Carpenter’s Guide To The SAS® REPORT Procedure, Art Carpenter
User Manual: Pdf
Open the PDF directly: View PDF 
.
Page Count: 497
| Download | |
| Open PDF In Browser | View PDF |
Praise from the Experts
“I am green with envy for the newest generation of SAS programmers because I wish that I had
had this book in front of me 20 years ago when I first started with SAS! Art’s book gives a
perspective on the REPORT procedure as no other user has done before by ingeniously
intertwining his extensive knowledge of PROC REPORT with the experiences and unique
approaches of over 100 PROC REPORT power users. His simple approach will give even the
most novice SAS user the necessary tools to get started with PROC REPORT, and his nicely
flowing buildup to PROC REPORT’s more complex usability makes this book a jewel for the
entire SAS community. Art’s PROC REPORT book has, without a doubt, given all SAS users
®
THE POWER TO KNOW !”
Rick Mitchell
Senior Systems Analyst
Westat
®
“Carpenter’s Complete Guide to the SAS REPORT Procedure is written in Art’s own friendly
®
and comfortable style, reminiscent of his major works on the topic of the SAS macro language.
Topics and features are introduced on a schedule that echoes how one might actually need to learn
them to get the job done. Manuals don’t do that. They have their purpose as reference tools and
they are certainly available when needed, but they are not typically productivity oriented. Art’s
book is.
“This book would make a welcome addition to the bookshelves of any serious SAS programmer.
I wish I had written it.”
Ray Pass, Ph.D.
Ray Pass Consulting
“This is the single best resource for PROC REPORT. I’m a huge fan of improving my SAS skills
through users group conferences, which is how I first learned PROC REPORT. The CD of
conference papers on this topic is an additional bonus to this book.”
Kim LeBouton
Independent Consultant
KJL Computing
SUGI 31 Conference Chair
“In his typical fashion, Art has taken on another facet of the SAS programming language and
provided a book that clears up a number of misconceptions about the REPORT procedure.
“Making the book even better is the logical approach not only to the training but also to the
development of a document. Art starts with the very basics of the REPORT procedure to define
the terms being used. From that point, he builds on the process by showing how to improve the
look of the output produced through the COLUMN and DEFINE statements. After the user
achieves some competence with these tools, Art introduces increasingly more complex topics,
such as compute blocks, along with clear explanations of how SAS processes the statements.
®
“Just like Carpenter’s Complete Guide to the SAS Macro Language, Second Edition, this is a
must-have book for any SAS developer’s reference library.”
Paul Slagle
Product Development Manager,
BI Systems
i3 Global
®
“Having read Carpenter’s Complete Guide to the SAS Macro Language, Second Edition, I knew
that this author was capable of presenting an exhaustive, in-depth, clever collection of tips,
examples, and references on a variety of levels. This new book does not disappoint.
“I am not a PROC REPORT programmer, yet I have been in the field for over 20 years. Art not
only brought me up to speed on probably the most flexible reporting tool SAS has to offer, he has
shown me how to integrate and leverage its use with other SAS resources (such as PROC
TRANSPOSE, PROC SUMMARY, SAS graphics, and ODS: RTF, PDF, HTML, and XML). He
has offered examples to build sophisticated yet simple reports that are eye catching and easily
understood.”
Thomas L. Lehmann
Sr. Programmer/Analyst II
RDA Group Inc.
“This book is very user-friendly with good examples and provides a useful and easily understood
way for readers to see, step by step, how to use PROC REPORT. As always, Art has done an
outstanding job of conveying the ‘how-to-do’ approach.”
Sue Douglas
Independent Consultant
“As an experienced SAS user, as well as an instructor of SAS, I find that this book is written well
for the new user, but also includes wonderful gems for the experienced user.”
Daphne Ewing
Sr. Director, Programming
Auxilium Pharmaceuticals, Inc.
®
“Carpenter’s Complete Guide to the SAS REPORT Procedure is well written, easy to
understand, and useful for the novice as well as the most advanced SAS user. Art’s technique of
using a spiral approach, starting off with the easiest examples and progressing to the most
complex, is ideal.
“This is a must-have book if you are a SAS user—period—no matter what level SAS user you
are! At some point in your IT activities, you must create a report of some nature. This book will
be an asset to you and your career by making you a better SAS IT professional.”
Charles Patridge
Sr. Data Engineer
Full Capture Solutions, Inc.
Carpenter’s
Complete Guide to
the SAS REPORT
Procedure
®
Art Carpenter
The correct bibliographic citation for this manual is as follows: Carpenter, Art. 2007. Carpenter’s Complete Guide to
the SAS® REPORT Procedure. Cary, NC: SAS Institute Inc.
Carpenter’s Complete Guide to the SAS® REPORT Procedure
Copyright © 2007, SAS Institute Inc., Cary, NC, USA
ISBN 978-1-59994-195-0
All rights reserved. Produced in the United States of America.
For a hard-copy book: No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in
any form or by any means, electronic, mechanical, photocopying, or otherwise, without the prior written permission
of the publisher, SAS Institute Inc.
For a Web download or e-book: Your use of this publication shall be governed by the terms established by the
vendor at the time you acquire this publication.
U.S. Government Restricted Rights Notice: Use, duplication, or disclosure of this software and related
documentation by the U.S. government is subject to the Agreement with SAS Institute and the restrictions set forth in
FAR 52.227-19, Commercial Computer Software-Restricted Rights (June 1987).
SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513.
1st printing, March 2007
SAS® Publishing provides a complete selection of books and electronic products to help customers use SAS software
to its fullest potential. For more information about our e-books, e-learning products, CDs, and hard-copy books, visit
the SAS Publishing Web site at support.sas.com/pubs or call 1-800-727-3228.
SAS® and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS
Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are registered trademarks or trademarks of their respective companies.
Contents
Preface
xi
Acknowledgments
About the Author
xiii
xv
How to Use This Book and the Accompanying CD
Part 1
Getting Started
Chapter 1
Creating a Simple Report
xvii
3
1.1
Basic Syntax 4
1.2
Routing Reports to ODS Destinations 6
1.3
Other Reporting Tools: A Brief Comparison of Capabilities 7
1.3.1 PROC REPORT vs. PROC PRINT 8
1.3.2 PROC REPORT vs. PROC TABULATE 8
1.3.3 PROC REPORT vs. DATA _NULL_ 8
1.4
The PROC REPORT Process: An Overview 9
1.4.1 PROC REPORT Terminology 9
1.4.2 Processing Phases 11
1.5
Chapter Exercises 12
Chapter 2
PROC REPORT: An Introduction
13
2.1
Introduction to the COLUMN Statement 16
2.2
Defining Types of Columns 17
2.2.1 Default Define Types DISPLAY and ANALYSIS 18
2.2.2 Using Define Usage ORDER 19
2.2.3 Using Define Type GROUP 22
2.2.4 Using Define Type ACROSS (and GROUP) 24
2.3
Doing More on the COLUMN Statement 25
2.3.1 Using the Comma to Form Nested Associations 26
2.3.2 Attaching Statistics with a Comma 26
2.3.3 Using Parentheses to Form Groups 28
2.3.4 Nesting Statistics under an ACROSS Variable 29
2.4
Other DEFINE Statement Options 31
2.4.1 Specification of an Analysis Statistic 31
2.4.2 Formatting the Values 33
2.4.3 Controlling the Order of the Displayed Values 34
2.4.4 Using the N Statistic without an ANALYSIS Variable 36
2.4.5 Associating Statistics with DEFINE Statements 37
iv Contents
2.5
Adding Text 39
2.5.1 Using the COLUMN Statement to Add Text 41
2.5.2 Using the DEFINE Statement to Add Text 43
2.5.3 Using the SPLIT= Option with Text 44
2.6
Compute Blocks 45
2.6.1 Inserting a Blank Line 46
2.6.2 Adding Lines of Text 47
2.6.3 Writing Formatted Values 49
2.6.4 Using SAS Language Elements 51
2.7
Sequencing of Step Events 52
2.8
Chapter Exercises 54
Chapter 3
Creating Breaks
57
3.1
Generating Breaks Using BREAK and RBREAK 57
3.2
BREAK Statement 59
3.2.1 Skipping a Line between Groups 59
3.2.2 Summarizing across a Group 61
3.2.3 Suppressing the Summarization Label 65
3.2.4 Generating a Page for Each Group Level 67
3.2.5 Combining Summaries with Detail Reports 68
3.3
RBREAK Statement 69
3.3.1 Using RBREAK in a Detail Report 69
3.3.2 Using RBREAK with BREAK in a Detail Report 70
3.3.3 Using RBREAK and BREAK in a Summary Report 71
3.4
Chapter 4
Chapter Exercises 73
Only in the LISTING Destination
75
4.1
Using the HEADLINE and HEADSKIP Options 76
4.2
Blank Lines, Overlines, and Underlines 78
4.3
Repeat Characters 79
4.3.1 Adding Repeated Characters to Spanning Headers 80
4.3.2 Repeat Characters with the SPLIT= Option 82
4.4
PROC REPORT Statement Options 83
4.4.1 Creating Boxes on the Report 83
4.4.2 Controlling the Centering of the Report 85
4.4.3 Adjusting the Width of Numeric and Computed Columns 85
4.4.4 Creating Multiple Panels on a Page 86
4.4.5 Using the PSPACE= Option 87
4.4.6 Controlling the Size of the Page 88
4.4.7 Using the FORMCHAR Option 89
4.4.8 Wrapping Data Lines 91
Contents
4.5
Other DEFINE Statement Options 92
4.5.1 Specifying the Column Width 93
4.5.2 Using the FLOW Option to Wrap Text 93
4.5.3 Adding Spaces between Columns 94
4.6
Chapter Exercises 96
Chapter 5
Creating and Modifying Columns Using the
Compute Block 97
5.1
Coordinating with the COLUMN and DEFINE Statements 98
5.2
Calculations Based on Statistics 99
5.3
Calculating Percentages within Groups 101
5.4
Using _PAGE_ with BEFORE and AFTER 103
5.5
Using the OUT= Option to View Report Break Information 104
5.6
Chapter Exercises 106
Part 2
Taking PROC REPORT Beyond the
Basics
Chapter 6
Refining Our Understanding of the
PROC REPORT Step 109
6.1
Additional DEFINE Statement Options 110
6.1.1 Changing Display Order with DESCENDING 110
6.1.2 Specification of Column Justification 111
6.1.3 Allowing the Use of Missing Classification Items 113
6.1.4 Controlling the Use of Analysis Items with All Missing or
Zero Values 115
6.1.5 Using NOPRINT 118
6.1.6 Identification Columns 119
6.1.7 Creating Vertical Page Breaks 120
6.2
Using Variable Aliases 121
6.3
Nesting Variables 122
6.4
Taking Full Advantage of Formats 123
6.4.1 User-Defined Formats 123
6.4.2 Preloading Formats 126
6.4.3 Order Based on Format Definition 130
6.5
Other PROC Statement Options 131
6.5.1 Removing Headers 131
6.5.2 Using NAMED Output 132
6.5.3 Debugging with the LIST Option 134
6.5.4 Including MISSING Classification Levels 134
v
vi Contents
6.6
BY-Group Processing 136
6.6.1 Using the BY Statement 137
6.6.2 Creating Breaks with BY Groups 138
6.6.3 Using the #BYVAL and #BYVAR Options 139
6.6.4 BY Groups and the Output Delivery System 141
6.7
Calculations Using the FREQ Statement 144
6.8
A Further Comment on Paging Issues 145
6.9
Chapter Exercises 146
Chapter 7
Extending Compute Blocks
147
7.1
Understanding the Events of the Compute Block Process 149
7.1.1 Setup Phase: Generating the Computed Summary
Information 150
7.1.2 Report Row Phase: Generating the Report 150
7.1.3 Process Example 151
7.2
Referencing Columns and Report Items in a Compute Block 154
7.2.1 Using Direct Variable Name References 156
7.2.2 Using Compound Variable Names 159
7.2.3 Using an Alias as a Column Reference 160
7.2.4 Using Absolute Column References: Referring to a Column
by Its Number 161
7.2.5 Using the Automatic Temporary Variable _BREAK_ 164
7.3
Using BEFORE and AFTER 166
7.4
Changing the Grouping Variable Values on Summary Lines 169
7.4.1 Specifying Text in a Compute Block 170
7.4.2 Using a Formatted Value 171
7.4.3 Creating a Dummy Column 173
7.5
Introducing the CALL DEFINE Routine 174
7.6
COMPUTE Statement Options and Switches 179
7.6.1 Justification of LINE Statement Text 179
7.6.2 Creating Character Variables with the CHARACTER and
LENGTH= Options 180
7.7
Using Logic and SAS Language Elements 182
7.7.1 Using the SUM Statement with Temporary Variables 183
7.7.2 Repeating GROUP and ORDER Variables on Each Row 185
7.7.3 Counting Items across Page Breaks in the LISTING
Destination 187
7.8
Doing More with the LINE Statement 191
7.8.1 Creating Group Summaries 192
7.8.2 Adding Repeated Characters 194
7.8.3 Understanding LINE Statement Execution 197
Contents
7.9
7.10
Chapter 8
vii
Examples of Common Tasks 199
7.9.1 Writing a Grand Total on Every Page 200
7.9.2 Combining Values into One Field or Column 202
7.9.3 Combining Values with Nested ACROSS Variables 204
7.9.4 Calculating a Weighted Mean 206
Chapter Exercises 209
Using PROC REPORT with ODS
211
8.1
Introduction to the STYLE= Option 213
8.2
Using STYLE= to Change Attributes 216
8.2.1 Changing Text and Cell Attributes 216
8.2.2 Adding a Logo to Your Report 219
8.2.3 Controlling Report Size 223
8.2.4 Adding Horizontal and Vertical Spaces to Separate
Data 223
8.3
Using CALL DEFINE to Change Style Attributes 227
8.3.1 Using CALL DEFINE in a Simple Report 228
8.3.2 Creating Shaded Rows 229
8.3.3 Conditional Assignment of Attributes 231
8.4
Creating Trafficlighting Effects 232
8.4.1 Building Trafficlighting Formats 233
8.4.2 Using Formats with the STYLE= Option 233
8.4.3 Controlling Trafficlighting with CALL DEFINE 236
8.4.4 Trafficlighting in the Presence of Computed Variables and
Summary Lines 236
8.4.5 Trafficlighting When Differentiating between Columns 240
8.4.6 Differentiating between Columns on Group Summary
Rows 242
8.4.7 Trafficlighting on the REPORT Summary Row 245
8.4.8 A Few Things to Remember When Using Formats for
Trafficlighting 249
8.5
Embedding Hyperlinks within Your Table 249
8.5.1 Linking Titles and Footnotes Using HTML Anchor Tags and
the LINK= Option 250
8.5.2 HTML Anchor Tags as Data Values 255
8.5.3 Establishing Links Using CALL DEFINE 257
8.5.4 Forming Links Using STYLE= 260
8.5.5 Creating Links in a PDF Document 262
8.5.6 Creating Links in an RTF Document 265
8.5.7 Automation Using the Macro Language 266
8.5.8 Using Formats to Build a Link 268
viii Contents
8.6
Using the Escape Character for In-Line Formatting 270
8.6.1 Controlling Superscripts and Subscripts 271
8.6.2 Displaying Page Numbers 273
8.6.3 Generating a Dagger 277
8.6.4 Using the Escape Character with S={ } and {STYLE} to
Change Style Attributes 279
8.6.5 Line Breaks and Wrapping 282
8.6.6 Passing Raw Destination-Specific Codes 289
8.7
Using TITLE and FOOTNOTE Statement Options
8.8
Creating Tip or “Flyover” Text for HTML and PDF
8.8.1 Using CALL DEFINE 293
8.8.2 Placing Tip Text Using STYLE= 295
8.8.3 Placing Tip Text Using ~S={ } 297
8.9
Specifying Multiple Columns for RTF and PDF
8.10 Adding Text through the TEXT= Option
292
293
298
300
8.11 RIGHTMARGIN: Aligning Numbers When Using CELLWIDTH
8.12 Chapter Exercises
304
Part 3
Extending PROC REPORT
Chapter 9
Reporting Specifics for ODS
Destinations 309
9.1
RTF
9.1.1
9.1.2
9.1.3
310
Using the BODYTITLE Option 311
Adding RTF Control Words 312
Post-processing of RTF Files 313
9.2
PDF 314
9.2.1 Adding PDF File Descriptors 314
9.2.2 Setting the Default Margins 315
9.3 HTML and Other Markup Destinations 316
9.3.1 Exporting a Report to Microsoft Excel 316
9.3.2 Setting Tagset Attributes 322
9.3.3 HTML Tags and Repeat Characters 323
Chapter 10 Solving Other Common Report
Problems 325
10.1 Creating Vertically Concatenated Tables 326
10.1.1 A Simple Table 326
10.1.2 Ordering the Generated Classifications 332
10.1.3 Text and Number Alignment in Derived Columns 335
10.1.4 Doing More in the PROC REPORT Step 338
301
Contents
10.2 Automating the PROC REPORT Process 341
10.2.1 Things to Think about When Automating 342
10.2.2 Macro Variable Resolution Issues 343
10.3 Coordinating Graphics with PROC REPORT 345
10.3.1 Using CALL DEFINE to Import Graphics 345
10.3.2 Using GPRINT and GREPLAY 352
10.3.3 Using the Annotate Facility to Generate Lines 355
10.4 Workarounds for Monospace-Only Options 357
10.5 Generating Separate Reports on the Same Page 360
10.5.1 ODS LAYOUT 360
10.5.2 HTML Reports 362
10.5.3 RTF and PDF Reports: Using STARTPAGE=NEVER 363
10.5.4 Aligning Columns across Reports 365
Chapter 11 Details of the PROC REPORT Process
367
11.1 Step Sequence Review 368
11.2 Building a Simple Table with Summary Lines 371
11.3 Compute Block Processing 372
11.3.1 Creating a Computed Variable 372
11.3.2 Multiple Compute Blocks 373
11.3.3 Summary Lines and Compute Blocks in the Same
Report 374
11.3.4 Using Compute BEFORE and COMPUTE AFTER with
Summary Lines 375
11.4 Using the ACROSS Define Usage 378
Appendix 1 Exercise Solutions
383
A1.1 Solutions to Chapter 1 Exercises 383
A1.2 Solutions to Chapter 2 Exercises 384
A1.3 Solutions to Chapter 3 Exercises 389
A1.4 Solutions to Chapter 4 Exercises 391
A1.5 Solutions to Chapter 5 Exercises 394
A1.6 Solutions to Chapter 6 Exercises 397
A1.7 Solutions to Chapter 7 Exercises 402
A1.8 Solutions to Chapter 8 Exercises 408
Appendix 2 Syntax and Example Index
417
A2.1 PROC REPORT Step 418
A2.1.1 Primary Statements 418
A2.1.2 PROC REPORT Statement Options 418
A2.1.3 BY Statement Options 419
ix
x Contents
A2.1.4 COLUMN Statement Options 420
A2.1.5 DEFINE Statement Options 420
A2.1.6 BREAK Statement Options 422
A2.1.7 RBREAK Statement Options 422
A2.1.8 COMPUTE Statement Options 422
A2.1.9 In the Compute Block 423
A2.1.10 Other PROC REPORT Step Statements 423
A2.2 Output Delivery System 423
A2.2.1 ODS Destinations 424
A2.2.2 ODS Statements and Options 424
A2.2.3 HTML Destination Options 425
A2.2.4 PDF Destination Options 425
A2.2.5 RTF Destination Options 426
A2.3 Attribute Control and Modification 426
A2.3.1 STYLE= Option 426
A2.3.2 CALL DEFINE Routine 427
A2.3.3 Attribute Modifiers 428
A2.4 System Options 429
Appendix 3 Example Locator
431
A3.1 Combination Detail and Summary Reports 431
A3.1.1 Transposing Rows and Columns 432
A3.1.2 Specifying and Calculating Statistics 432
A3.1.3 Enhancing Tables 432
A3.1.4 Controlling Pages 433
A3.1.5 Controlling the Order of the Report’s Rows 433
A3.2 Calculating Percentages 433
A3.3 Processing Weighted Means and Totals 433
A3.4 Understanding Processing Phases and Event Sequencing 434
References
Index 449
435
Preface
The presentation of data is an essential part of virtually every study, and SAS provides numerous
tools that enable the user to create a large variety of charts, reports, and data summaries. The
REPORT procedure is a particularly powerful and valuable procedure that can be used in this
process. It can be used to both summarize and display data, and it is highly customizable and
flexible.
Unfortunately, for many of those just starting to learn PROC REPORT, the terms “customizable”
and “flexible” often seem to be euphemisms for “hard to learn.” Fortunately, PROC REPORT
does not have to be “hard to learn”—not, that is, with the right approach. And that is why I have
written this book to offer that approach and the necessary tool sets to you,
This book introduces you to PROC REPORT by showing you how it works and how it “thinks.”
A progression of increasingly more complex examples are used to illustrate many features,
including options and capabilities new to SAS 8 and SAS®9. Along the way, we create a variety
of reports and tables that highlight some of the more common and even uncommon capabilities of
the procedure.
xii
Acknowledgments
I received quite a bit of essential assistance from a number of PROC REPORT experts. Cynthia
Zender, lead trainer for PROC REPORT at SAS, provided a number of examples and comments
on the sections dealing with trafficlighting and the Output Delivery System. Russ Lavery and I
had a number of long and interesting discussions on “the way things seem to work.” Pete Lund of
Looking Glass Analytics is a PROC REPORT programmer who is constantly pushing the
procedure’s limits. Not only did he review the manuscript, but he also provided a number of ODS
examples. Be sure to read Pete’s papers on various ODS, macro language, and reporting topics.
Others who shared examples and topics included Ben Cochran, Mike Metz, and Justina Flavin.
The clinical trial data (RPTDATA.CLINICS) that appears in a number of examples throughout
this book was used with permission of Mr. Kirk Lafler of Software Intelligence Corporation.
A great deal of input was received from the numerous people who offered their time to help with
the content of this book. Reviewers included the following people from SAS:
Vince DelGobbo
Nancy Goodling
Wayne Hester
Tim Hunter
Scott Huntley
Bari Lawhorn
Elizabeth Maldonado
Kathryn McLawhorn
Allison McMahill
Chevell Parker
Jane Stroupe
Cynthia Zender
The following people also reviewed the book:
Russ Lavery
Pete Lund
Finally, I would like to thank Marilyn, my wife and the owner of California Occidental
Consultants, for her unwavering support of this time-consuming project.
xiv
About the Author
Art Carpenter’s publications include three other books and numerous
papers and posters presented at SUGI, SAS Global Forum, and other
®
users group conferences. Art has been using SAS since 1976 and
has served in various leadership positions in local, regional, national,
and international users groups. He is a SAS Certified Advanced
TM
Programmer and through California Occidental Consultants he
teaches SAS courses and provides contract SAS programming
support nationwide.
Author Contact
Arthur L. Carpenter
California Occidental Consultants
P.O. Box 430
Oceanside, CA 92085-0430
(760) 945-0613
art@caloxy.com
www.caloxy.com
xvi
How to Use This Book and the Accompanying CD
This book was not written with the intention of providing a syntax reference manual, nor is there
any intent to completely describe all aspects of the REPORT procedure. Rather, this book was
written to assist new PROC REPORT programmers in getting started, while at the same time
providing techniques that I have found to be useful to more advanced users of the procedure.
Because we do not all approach problems in the same way, it is likely that others will find
alternate solutions to many of the reporting problems discussed in this book.
The book is divided into three primary parts.
Part 1 provides a step-by-step introduction to PROC REPORT and is designed to be read
linearly by someone who is unfamiliar with the procedure. Those familiar with the procedure
might wish to scan these chapters quickly.
Part 2 includes many of the more advanced options and concepts associated with the
procedure. This section of the book is designed to be used as much as a reference manual as
an instructional guide. A number of the techniques discussed in this section have been known
to cause a fair amount of consternation to those attempting to “figure out” the procedure on
their own.
Part 3 incorporates the options and statements described in the first two parts into a series of
examples that highlight many of the extended capabilities of PROC REPORT. This part
includes a discussion of a few ODS statements and options that might be useful to a PROC
REPORT programmer. This section also includes a more in-depth look at the PROC
REPORT process itself, especially as it relates to the execution of compute blocks.
PROC REPORT Tables
Many of the tables produced in the examples in this book are not displayed in their entirety. If you
want to see an entire table, the code is available on the CD that comes with this book. See the
section “Using the Code Examples in This Book.”
Logs and tables that are written to the
LISTING destination are displayed in a
text box enclosed in the borders
shown here.
Tables resulting from screen captures, especially from
the PDF, RTF, and HTML destinations, have a single
narrow border like this.
At times, summarizing remarks or
asides associated with a table or code
set are written in a text box like this
to the right of the listing or table.
xviii
SAS Versions
This book was written while SAS 9.1.3 was in production and before the beta version of SAS 9.2
was released. Occasionally, I have made comments with regard to what will probably happen or
options that may be available in SAS 9.2. The functionality of SAS 9.2 is, as of this writing, still
subject to change, and these comments are based on discussions with SAS developers.
Fonts
You will see these typographical conventions used throughout the book:
UPPERCASE
SAS language elements such as the names of procedures and options
italics
emphasis in text; user-supplied values
Monospace
SAS code
bold monospace
emphasis in SAS code
Using the Appendixes
Appendix 1: Exercise Solutions provides Solutions to the chapter exercises.
Appendix 2: Syntax and Example Index serves as a secondary index to this book. It is
organized from a syntax and example perspective, and contains information on where PROC
REPORT step elements and options are discussed as well as where they are used within the
book. System and ODS options are also included.
Appendix 3: Example Locator lists examples of the various types of reports that are found
throughout the book. Scanning this appendix should help you locate an example of a report
with the specified elements.
MORE INFORMATION
Additional information, about a particular topic and related topics within this book is detailed
under this heading. Generally the reader is referred to a particular section of the book (e.g. see
Section 2.3.4).
SEE ALSO
These sections are used to point the reader to references outside the scope of this book. The
references are in the form of the author’s last name and the year of publication—for example,
(Gupta, 2003) or Gupta (2003). Details appear in the “References” section at the back of the book.
References to sample programs that have been written by SAS are noted using the number of the
sample program, e.g., Sample 603. Occasionally, when references are made to a book or longer
article, the page number is also included in the citation.
Using the Code Examples in This Book
The SAS programs used as examples throughout this book are available as sample programs on
the book’s CD. They are all named according to the section in which they are used. For example,
a PROC REPORT step shown in Section 11.3.1 would be named S11_3_1.sas. Occasionally,
more than one example will appear in a given section. These will be named using a letter
extension, e.g., S11_3_1a.sas. There are also a few bonus programs that are not specifically
discussed in the book but still demonstrate or reinforce items of discussion in the section. These
follow a similar naming convention.
xix
For exercise questions that result in a SAS program, the program name will contain the chapter
and exercise number, e.g. E11_3.sas for Chapter 11 question 3. These programs also are available
on the CD.
About the Data Used in This Book
Although it makes the examples a bit less interesting, only a limited number of data tables have
been used in the book. The intent is that the reader will be able to concentrate on the code without
having to learn a new data table for each example.
RPTDATA.CLINICS was supplied by Kirk Lafler of Software Intelligence Corporation and
is used with his permission. It contains fictitious patient information from a small clinical
study. The data represents patient visits with patient information as well as visit-specific
information. Data was collected at clinics from across the country, which for management
purposes, has been divided into 10 regions. A DATA step that can be used to generate this
data set is included in the SAS code samples.
SASHELP.various are data sets that are shipped with SAS, and these are all available within
the SASHELP library. The table SASHELP.CLASS contains observations and some
demographic information on a class of teenagers. The sales data of an imaginary shoe
distribution company is contained in the table SASHELP.PRDSALE.
How to Use the Accompanying CD
The CD that accompanies this book has been designed to maximize the amount of knowledge that
I can offer to you. It contains a great deal of information that I would otherwise be unable to
include within the pages of the book itself. I hope you will find this bonus content of interest as
you read the book. The CD includes the following directories:
Sample programs
The code used to generate the examples is available and is ready for you to submit. There are
even bonus programs that are not discussed within the book.
Data Sets
The data sets used in the examples are provided.
Results
The tables that are generated by the example programs are provided so that you can view
them directly.
Cited Papers
The PDF files of nearly a hundred of the papers cited in the “References” section of this book
are provided.
Russ Lavery’s “An Animated Guide to the SAS REPORT Procedure”
The CD enables us to overcome one of the limitations of a book that is published using the
standard media of paper. This CD also includes Russ Lavery’s specially revised “An
Animated Guide to the SAS REPORT Procedure.” This PowerPoint presentation contains an
animated approach to the process of learning and assimilating not only how to use the
REPORT procedure, but how to understand its phases of operation. This presentation is
offered to you as an executable file that can be run only from the CD. You do not need to
have access to PowerPoint to view this presentation.
xx
P a r t
1
Getting Started
3
Chapter 1
Creating a Simple Report
Chapter 2
PROC REPORT: An Introduction
Chapter 3
Creating Breaks
Chapter 4
Only in the LISTING Destination
Chapter 5
Creating and Modifying Columns Using the Compute Block
13
57
75
97
You can use the REPORT procedure to generate a wide range of sophisticated tables and reports. The
code can be complex or fairly straightforward. This part of the book shows you how to create fairly
simple reports using the basic statements and their options.
2 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
1
Creating a Simple Report
1.1 Basic Syntax 4
1.2 Routing Reports to ODS Destinations 6
1.3 Other Reporting Tools: A Brief Comparison of Capabilities 7
1.3.1 PROC REPORT vs. PROC PRINT 8
1.3.2 PROC REPORT vs. PROC TABULATE 8
1.3.3 PROC REPORT vs. DATA _NULL_ 8
1.4 The PROC REPORT Process: An Overview 9
1.4.1 PROC REPORT Terminology 9
1.4.2 Processing Phases 11
1.5 Chapter Exercises 12
The syntax for PROC REPORT is quite different from that of most other Base SAS procedures. In
most procedures, the supporting statements define the scope and options of the procedure. In a
PROC REPORT step, on the other hand, the statements refer to and build on each other.
PROC REPORT can be used in two different modes, batch and interactive. This book discusses
the syntax of PROC REPORT in the batch environment, and does not discuss the interactive or
windowing environment.
4 Carpenter’s Complete Guide to the SAS REPORT Procedure
1.1 Basic Syntax
Like most procedures, PROC REPORT can be executed with a minimal understanding of even the
most basic syntax.
In its simplest form, PROC REPORT is similar to PROC PRINT in that it creates a data listing.
Here is the minimum coding required:
PROC REPORT;
run;
By default the REPORT procedure opens an interactive windowing environment. This
environment is not normally used and is not discussed in this book. The following is the simplest
PROC REPORT step that does not open the interactive windowing environment:
PROC REPORT nowd;n
run;
When executed, this simple step creates a listing of all rows and all columns in the most recently
modified data table. This plain vanilla result is of course rarely what we need or want, so we must
know more in order to create the report that we actually do need.
Some of the basic statements used in PROC REPORT include the following:
PROC REPORT Y DATA= datasetname ;
COLUMN variable list and column specifications;
DEFINE column / column usage and attributes;
COMPUTE column; compute block statements; ENDCOMP;
RUN;
A number of options and modifiers can be used along with these statements. Most of these are
discussed throughout this book. To locate the discussion of a specific statement, option, or modifier,
see Appendix 2, which provides a syntax and example reference locator for this book.
You can use the REPORT procedure to build reports interactively (LeBouton 2004). While
appealing in concept, in practice this feature is rarely used and is not discussed in this book.
Unfortunately, the procedure default is to initiate the interactive mode. You can disable this made
by using either the NOFS, NOWD, or NOWINDOWS option. NOWD n is most often used in the
documentation and in SAS literature.
As for most procedures that operate against data tables, you will want to be able to specify which
table PROC REPORT is to display. oThe DATA= option is used for this specification in the
REPORT procedure as it is in so many other SAS procedures.
A number of supporting statements are used in the PROC REPORT step. The following
statements are three of the most common:
COLUMN
DEFINE
identifies all variables (report items) used in the generation of the table.
specifies how the column is to be used and what its attributes are to be.
One DEFINE statement is used for each variable in the COLUMN
statement.
COMPUTE creates new columns and performs column-specific operations.
Chapter 1: Creating a Simple Report 5
The following PROC REPORT step creates a simple listing of a select few of the twenty or so
variables in the RPTDATA.CLINICS table:
* Simple report;
options nocenter;
title1 'Using Proc REPORT';
title2 'Simple Report';
proc report data=rptdata.clinics nowd;
columns region lname fname wt;
define region / display;
define lname / display;
define fname / display;
define wt / display;
run;
Here are the first few lines of the generated report.
Using Proc REPORT
Simple Report
re
gi
on
5
6
9
2
4
5
8
4
last name
Rose
Nolan
Tanner
Saunders
Jackson
Pope
Olsen
Maxim
first
name
Mary
Terrie
Heidi
Liz
Ted
Robert
June
Kurt
weight
in pounds
215
187
177
109
201
158
158
179
. . . Portions of the report are not shown . . .
A quick inspection of the output listing shows both similarities and differences between PROC
REPORT and PROC PRINT. As in PROC PRINT, variables/columns are listed across the page,
while rows/observations are listed down the page. Unlike PROC PRINT, there is no OBS column,
and the default is to print the variable label instead of the variable name. “Pretty” is not a default
characteristic, and the remainder of this book is devoted to controlling how the report looks.
Notice in this example that the default header of the column is the variable label. In Section 2.5,
several examples show how you can control this text. You can also use the system option
NOLABEL to make the variable name the default column header.
MORE INFORMATION
Appendix 2, “Syntax and Example Index,” is designed to help the reader navigate this book.
SEE ALSO
A nice introduction to the PROC REPORT windowing environment is presented by LeBouton
(2004). This interactive environment was first introduced in the SAS Guide to the REPORT
Procedure: Usage and Reference, Version 6 (1990).
6 Carpenter’s Complete Guide to the SAS REPORT Procedure
1.2 Routing Reports to ODS Destinations
Usually we need to route the output generated by PROC REPORT to one or more Output
Delivery System (ODS) destinations. The syntax and use of ODS is outside the direct scope of
this book. However, because we are going to depend on ODS for a great deal of the appearance of
the output generated by PROC REPORT, it is necessary to at least discuss the basics of ODS.
Since we use reports in different ways, we need to generate the reports as different types of files.
We declare the type of file to be generated by specifying the ODS destination. This means that
there is generally a correspondence between the name of the destination and the type of output
that is to be created (e.g., HTML, PDF, RTF).
Usually we surround the PROC REPORT step with what has been referred to as an ODS
sandwich. The sandwich consists of two ODS statements that turn the Output Delivery System on
and off. The physical name of the output file and the file’s location are included in the first ODS
statement. The second ODS statement closes (turns off) the ODS destination. In both statements,
the ODS destination name immediately follows the ODS keyword.
The general form of the ODS sandwich is something like this:
ods destination ;
proc report . . . . ;
. . . .
run;
ods destination close;
If you wanted to re-create the results of the previous step as an HTML document, you might write
ODS statements like the following. Note that all physical paths in the examples are created using
the macro variable &PATH. This should make it easier for you to replicate the results of these
same example programs on your own computer.
ods html file="&path\results\simple.html";
title1 'Using Proc REPORT';
title2 'Simple Report';
proc report data=rptdata.clinics nowd;
columns region lname fname wt;
define region / display;
define lname / display;
define fname / display;
define wt / display;
run;
ods html close;
Chapter 1: Creating a Simple Report 7
Here is a portion of the HTML report:
. . .
Portions of the report are not shown . . .
Throughout this book you will see examples of a number of other ODS statements, options, and
destinations.
MORE INFORMATION
The &PATH macro variable is used throughout the book to designate the upper portion of all
location references and is described in more detail in “About This Book.” Appendix 2 contains a
list of ODS-related references within this book. A number of other sections in this book contain
examples that utilize features of ODS. Chapter 8, “Using PROC REPORT with ODS,” and
Chapter 9, “Reporting Specifics for ODS Destinations,” are devoted to the topic.
SEE ALSO
Haworth (2001, 2003) and Gupta (2003) provide very good information on the Output Delivery
System and show how to get started using it. Kumar (2006) introduces ODS along with a PROC
REPORT example.
1.3 Other Reporting Tools: A Brief Comparison
of Capabilities
Since SAS provides a variety of reporting tools, there is sometimes some confusion about which
tool should be used in a given situation. Three of the primary reporting tools are the PRINT,
REPORT, and TABULATE procedures. All three have enough flexibility to produce a fairly
diverse set of reports. However, they are not the same and do not have the same overall
capabilities.
All three of these procedures work well with the ODS environment, and each supports the use of
the STYLE= option (see Section 8.1 for an introduction to this option).
8 Carpenter’s Complete Guide to the SAS REPORT Procedure
1.3.1 PROC REPORT vs. PROC PRINT
Both the PRINT and REPORT procedures can perform detail-level reporting (reporting of
individual data values). Although a number of supporting statements are available, PROC PRINT
has the advantage of being a fairly simple procedure and is generally one of the first procedures
that is learned by a new user.
Although both procedures are good at creating simple detail reports, the only real summary
capability of PROC PRINT is to calculate column totals. When the SUM statement is combined
with the BY statement, SUM (and SUMBY) can calculate group and sub-group totals. Unlike
PROC PRINT, PROC REPORT is not limited to group totals. PROC REPORT can calculate all
of the usual statistics that can be calculated by other procedures such as MEANS, SUMMARY,
and UNIVARIATE. In fact, the reason that PROC REPORT can calculate some of these same
statistics is that the MEANS/SUMMARY process is used behind the scenes for summarizing the
data set used with PROC REPORT.
Most users find that PROC PRINT is fine for simple straightforward detailed reports. However, if
you find that the limitations of PROC PRINT are causing extra work, then it is probably an
indication that it is time to switch to PROC REPORT.
SEE ALSO
Burlew (2005, pp. 18–19) provides a comparison of the default behaviors of these two procedures.
1.3.2 PROC REPORT vs. PROC TABULATE
Both the REPORT and TABULATE procedures can create summary reports, and each has access
to the same standard suite of summary statistics.
Unlike PROC TABULATE, the REPORT procedure can provide detail reporting as well as
summary reporting capabilities. PROC REPORT has the added flexibility to calculate and display
columns of information based on other columns in the report.
Because of the unique way that PROC TABULATE structures the report table, it has a great deal
more flexibility to present the groups, subgroups, and statistics as either rows or columns. This is
especially true for vertically concatenated reports, which are very straightforward in PROC
TABULATE and difficult in PROC REPORT (see Section 10.1).
SEE ALSO
Buck (1999, 2004), Bruns, Pass, and Eaton (2002), and Bruns and Pass (2004) compare the
strengths and weaknesses of these two procedures.
The TABULATE procedure is fully described by Haworth (1999).
1.3.3 PROC REPORT vs. DATA _NULL_
The DATA _NULL_ step is a reporting tool that offers extreme flexibility. Because it uses the
power of the DATA step, this methodology enables the user to generate reports in almost any
form.
Of course, this power comes with the price of complexity. Although the user has the power to
place every character “just so,” the process itself can become quite difficult. In PROC REPORT,
Chapter 1: Creating a Simple Report 9
the compute block, with its access to the majority of the SAS language elements, such as logic
processing, functions, and assignment statements, takes on some of the role of the DATA
_NULL_ step.
1.4 The PROC REPORT Process: An Overview
For most procedures, the internal processing is of little interest to the average user. This should
not be the case for PROC REPORT. Because PROC REPORT has the capability of creating
columns as well as group and report summaries, the process can be quite complex. When the
report is simple, such as those in this chapter and in Chapter 2, “PROC REPORT: An
Introduction,” the processing details are of less interest. However, as new columns are calculated
and perhaps then coordinated with report and group summaries, a more complete understanding
of the process becomes critically important.
MORE INFORMATION
The timing of the compute block is discussed in Section 7.1, and a detailed presentation of the
processing of the PROC REPORT step is provided in Chapter 11, “Details of the PROC REPORT
Process.”
SEE ALSO
SAS Technical Report P-258 (1993, Chapter 10), Lavery (2003), and Russ Lavery’s “An
Animated Guide to the SAS REPORT Procedure,” which is included on the CD that accompanies
this book, discuss the sequencing of events in detail.
1.4.1 PROC REPORT Terminology
Some of the terms and concepts associated with PROC REPORT are similar to those in other
types of PROC steps. However, PROC REPORT is unique in that it allows some DATA step-type
processing to be performed, and thus we need some specialized words and phrases to discuss this
processing. Of course, the special nature of PROC REPORT results in terminology that is unique
to this step, and an overview of basic PROC REPORT processing will highlight these terms.
Term usage has evolved since the introduction of the REPORT procedure in SAS 6.06. This not
only reflects the complexity of the procedure, but also the changes in how PROC REPORT
operates behind the scenes.
Some of the terminology used in this book is included here.
Current Terminology
Two general types of reports can be generated by PROC REPORT. Detail reports are most
similar to those generated by PROC PRINT and have one line in the report, called a report row,
for each observation in the incoming data set. When the incoming data is summarized or
collapsed into groups, PROC REPORT can create a summary report. PROC REPORT is flexible
enough to create a report that has characteristics of both of these types of reports.
10 Carpenter’s Complete Guide to the SAS REPORT Procedure
The report generated by PROC REPORT is called the final report output. Columns on the final
report output can include more than variables, so the report columns are often referred to as
report items. There are two general classes of report items used within the PROC REPORT step:
report variables
temporary variables
appear in the COLUMN statement and usually in one or more
of the report columns. They may or may not be created or used in
COMPUTE blocks.
are created and used in COMPUTE block calculations, but do not
appear in the COLUMN statement or on the report itself.
Through the DEFINE statement, report items are assigned a define type or define usage that
determines how the variables are to be processed by PROC REPORT. Report items can be used
during compute block execution to build or calculate other report items. Depending on the PROC
REPORT step, not all report items will necessarily be included in the final report output.
PROC REPORT builds each report row, one row at a time. However, in order for summary and
break information to be available when it is constructing summary rows, PROC REPORT goes
through a three-phase process to create the report. The first phase is the evaluation phase, and it
is during this phase that the submitted code is assessed. The final two phases are of special
interest to the PROC REPORT programmer, and these (the setup phase and the report row
phase) are described in more detail in the section on processing phases (see Section 1.4.2).
For reports that summarize the incoming data, the summary results are determined during the
setup phase and stored in memory in the computed summary information.
Each line of a report is a report row; for some reports, however, report rows are generated that are
not ultimately written to the final report output. The final report output is generated one row at a
time, and depending on the selection of statement options, not all summary report rows are
included in the final report output.
Outdated Terminology
Since its introduction in SAS 6, PROC REPORT has been the subject of a great many papers.
This unofficial documentation, as well as some of the initial official documentation, has generated
fairly extensive terminology for the internal processes of PROC REPORT. Although some of this
terminology reflects, to some degree, current internal processes, the majority has at best become
outdated. In order to assist readers of this older literature, the following table attempts to link the
older terminology with that used throughout this book.
Older, Outdated, or Inaccurate Terms
Terminology Used in This Book
Temporary Internal File, TIF
Computed summarized information
computed summary information area
Report Data Vector, RDV
Report variables, report items
DATA Step Variable
Temporary variable
DATA Variable Table, DVT
Temporary variables are stored in memory
and no special name is needed for this
location.
DATA Step Statement
DATA Step Functions
SAS language elements
Chapter 1: Creating a Simple Report 11
SEE ALSO
Extensive discussion contrasting report items and temporary variables can be found in Chapter
10 and more specifically on pages 250-251 of SAS Technical Report P-258 (1993). When you
read SAS Technical Report P-258, remember that it reflects some of the earliest documentation
available for PROC REPORT and does not use the current terminologies or in some cases reflect
the current processes of the PROC REPORT step.
1.4.2 Processing Phases
When a PROC REPORT step is submitted, SAS breaks down the processing into a series of steps
or phases. All of this processing, as well as the results of the processing, including the computed
summary information, takes place in memory.
Evaluation Phase
First, all the PROC REPORT statements are evaluated before anything else happens. The SAS
language elements and LINE statements (if there are any) in the compute blocks are simply set
aside to be executed as each report row is built (report row phase). This evaluation determines the
resources and levels of summarization that will be needed during the setup phase.
Setup Phase
After the code is evaluated, the setup phase reads and prepares the incoming data. If necessary,
the columns that will be used for summarizing are sent to the MEANS/SUMMARY engine,
where the summarization takes place.
Report Row Phase
Once PROC REPORT is done with these preliminary setup phase tasks and the computed
summarized information has been created, the report can be built row by row during the report
row phase. Finally, after each report row is built, it is sent to all open ODS destinations
(LISTING, HTML, RTF, PDF, etc.).
Summary of the Processing Phases
The following flowchart shows the general processing phases at the conceptual level described in
this section.
It is tempting to try to impose DATA step concepts
such as the Program Data Vector onto PROC
REPORT. However, between PROC REPORT's
original inception with SAS 6 and subsequent
upgrades and rewrites to the underlying PROC
REPORT code, these primary conceptual phases
(especially the setup and report row phases) should
suffice to explain how the final report is
constructed.
Of primary importance is to remember that during
the report row phase, processing is from left to
right. The order of the variables in the COLUMN
statement (see Section 2.1), therefore, becomes very
important.
12 Carpenter’s Complete Guide to the SAS REPORT Procedure
1.5 Chapter Exercises
1. What are the three processing phases of a PROC REPORT step?
2. What is the difference between temporary variables and report variables?
3. What two PROC REPORT statement options will you use within virtually every PROC
REPORT statement?
C h a p t e r
2
PROC REPORT: An Introduction
2.1 Introduction to the COLUMN Statement 16
2.2 Defining Types of Columns 17
2.2.1 Default Define Types DISPLAY and ANALYSIS 18
2.2.2 Using Define Usage ORDER 19
2.2.3 Using Define Type GROUP 22
2.2.4 Using Define Type ACROSS (and GROUP) 24
2.3 Doing More on the COLUMN Statement 25
2.3.1 Using the Comma to Form Nested Associations 26
2.3.2 Attaching Statistics with a Comma 26
2.3.3 Using Parentheses to Form Groups 28
2.3.4 Nesting Statistics under an ACROSS Variable 29
2.4 Other DEFINE Statement Options 31
2.4.1 Specification of an Analysis Statistic 31
2.4.2 Formatting the Values 33
2.4.3 Controlling the Order of the Displayed Values 34
2.4.4 Using the N Statistic without an ANALYSIS Variable 36
2.4.5 Associating Statistics with DEFINE Statements 37
2.5 Adding Text 39
2.5.1 Using the COLUMN Statement to Add Text 41
2.5.2 Using the DEFINE Statement to Add Text 43
2.5.3 Using the SPLIT= Option with Text 44
2.6 Compute Blocks 45
2.6.1 Inserting a Blank Line 46
14 Carpenter’s Complete Guide to the SAS REPORT Procedure
2.6.2 Adding Lines of Text 47
2.6.3 Writing Formatted Values 49
2.6.4 Using SAS Language Elements 51
2.7 Sequencing of Step Events 52
2.8 Chapter Exercises 54
In this chapter the basic statements and options associated with the PROC REPORT step are
introduced. Although you will not necessarily be able to make the perfect report using only what
is presented here, you will probably be able to get fairly close.
The appearance of a given report will depend on its output destination. LISTING is the default
destination for most of the examples in this chapter (LISTING is the default when no other ODS
destination has been specified). One of the biggest differences between destinations is in the way
that the text is placed in the tables. The first example in Section 2.5 highlights some of the more
immediately obvious of these differences. Most of the reports in this chapter have formatting and
appearance issues in the LISTING destination that automatically go away for other destinations.
The LISTING destination is useful in the context of this chapter because it emphasizes the
differences between a number of the demonstrated options and statements.
The examples presented in this chapter build on one another. We start by creating some fairly
ugly, and in some cases fairly silly, reports. As we add additional options and statements, the
reports become more useful.
The starting point for the PROC REPORT step, as it is for all procedure steps, is the PROC
statement. In virtually all of the steps contained in this book, the PROC REPORT statement
contains at least the following options:
proc report data=dataset_name
nowd;
The DATA= option identifies the name of the table on which the report is to be based, and the
NOFS, NOWINDOWS or NOWD option turns off the interactive PROC REPORT window
system (which this book ignores). A number of other options can be used on the PROC REPORT
statement (these are presented throughout the book, most specifically in Section 6.5).
Unlike PROC PRINT, which by default uses the column name as the column header, PROC
REPORT uses the column label as the default column header. You can change the header in
several ways, as discussed throughout this book. However, if you want to change the default
behavior, you can use the NOLABEL system option.
option nolabel;
The examples in this chapter utilize the data sets SASHELP.CLASS, which contains student
information for a small number of teenagers, and RPTDATA.CLINICS, which contains fictional
data on a small clinical study. The table RPTDATA.CLINICS is available both as a data set and
as the code that creates it on the CD that accompanies this book.
Chapter 2: PROC REPORT: An Introduction 15
A typical PROC REPORT step containing statements discussed in this chapter could include the
following:
title1 'Using Proc REPORT';
proc report data=sashelp.class nowd;
column name sex age height;
define name
/ display;
define sex
/ display;
define age
/ analysis;
define height / analysis;
run;
Here is the resulting report in the LISTING destination:
Using Proc REPORT
Name
Alfred
Alice
Barbara
Carol
Henry
James
Jane
Janet
Jeffrey
John
Joyce
Judy
Louise
Mary
Philip
Robert
Ronald
Thomas
William
S
e
x
M
F
F
F
M
M
F
F
M
M
F
F
F
F
M
M
M
M
M
Age
14
13
13
14
14
12
12
15
13
12
11
14
12
15
16
12
15
11
15
Height
69
56.5
65.3
62.8
63.5
57.3
59.8
62.5
62.5
59
51.3
64.3
56.3
66.5
72
64.8
67
57.5
66.5
This is a detail report. It contains one
row for each row of incoming data.
The individual columns and their
order is specified on the COLUMN
statement (see Section 2.1). The
DEFINE statements tell the PROC
REPORT step how to handle each of
these report items.
A portion of the same report
generated as a PDF file using the
PRINTER style shows some of the
formatting differences that can be
expected for different ODS
destinations.
Destination: PDF Style: PRINTER
16 Carpenter’s Complete Guide to the SAS REPORT Procedure
SEE ALSO
A very nice introduction to PROC REPORT is given by Cochran (2005). Simple introductions to
PROC REPORT can also be found in Carpenter (2005), Haworth (2003), Foley (2001), Ma,
Meimei, and Schlotzhauer (2000), Ma et al. (2002), LeBouton (2004), Pass (2000), and Smith
(2000). A number of introductory examples are provided in SAS Technical Report P-258 (1993),
which also discusses the NOLABEL option (p. 61). Also in P-258 is a general syntax review of
the entire step (p. 180) and of the various PROC REPORT step statements (Chapter 9).
2.1 Introduction to the COLUMN Statement
The COLUMN statement is primarily used to identify the variables of interest. In its simplest
form, it is merely a list; however, it can also be used to create headers and groups of variables, in
addition to associating statistics with variables. COLUMN may be either singular or plural, and
can be abbreviated as COL.
The primary function of the COLUMN statement is to provide a list of variables (report items) for
PROC REPORT to operate against. The space-separated variable names are listed, left to right, in
the order that they are to appear in the report. Variables in the incoming data set that are not listed
in the COLUMN statement are not available to the REPORT procedure. This means that unless an
incoming variable appears on the COLUMN statement, it will not be printed and it will be
unavailable for use in compute blocks.
The following COLUMN statement specifies four report variables. The DEFINE statement (see
Section 2.2) associated with each of these variables determines how it will appear in the report.
column region lname fname wt;
The COLUMN statement appears in the PROC REPORT step before any DEFINE statements. A
simple PROC REPORT step could be specified as follows:
title1 'Using Proc REPORT';
title2 'A Simple COLUMN Statement';
proc report data=sashelp.class nowd;
column name sex age height;
run;
The incoming data table contains the variable WEIGHT, which is not included in the report
because it does not appear on the COLUMN statement.
Chapter 2: PROC REPORT: An Introduction 17
Using Proc REPORT
A Simple COLUMN Statement
Name
Alfred
Alice
Barbara
Carol
Henry
James
Jane
Janet
Jeffrey
John
Joyce
Judy
Louise
Mary
Philip
Robert
Ronald
Thomas
William
S
e
x
M
F
F
F
M
M
F
F
M
M
F
F
F
F
M
M
M
M
M
Age
14
13
13
14
14
12
12
15
13
12
11
14
12
15
16
12
15
11
15
Height
69
56.5
65.3
62.8
63.5
57.3
59.8
62.5
62.5
59
51.3
64.3
56.3
66.5
72
64.8
67
57.5
66.5
There are a number of
immediately obvious
differences between the plain
vanilla report generated by
PROC REPORT and that of
PROC PRINT. Because PROC
REPORT expects that the
programmer will want to take
control, the default report is
even less pretty than that of
PROC PRINT.
This detail report has one row
for each incoming observation.
Although this report could be useful, it is not pretty. There are a number of things on this report
that need to be fixed. This book addresses those types of issues. Also, the ODS LISTING
destination, which is used here, is the least pretty of all the destinations. The default presentation
for most of the other destinations would make this report substantially more usable.
2.2 Defining Types of Columns
Although it can do much more than is shown in Section 2.1, the COLUMN statement is
insufficient to provide total control of the appearance of individual columns and how they are to
be used. This task falls to the DEFINE statement.
The DEFINE statement lists the name of the column to which it applies and, following a slash (/),
the attributes and options that are to be applied to that column. The syntax follows a pattern that
might include the following:
define columnname / ;
The primary attribute of the column is its define usage or define type. This attribute tells the
REPORT procedure how to use this variable. There are several different define types, including
the following:
DISPLAY
shows or displays the value of the variable (default usage for character
variables).
ANALYSIS
uses the variable in calculations with a statistic (default usage for numeric
variables).
GROUP
uses the variable to consolidate observations.
18 Carpenter’s Complete Guide to the SAS REPORT Procedure
ORDER
sorts the data and forms groups when summary statistics are requested.
ACROSS
creates groups across the page rather than down the page.
COMPUTED
specifies a report item, not on the incoming data set, that is to be created in
a compute block.
Although the DEFINE statement is not required unless you need to change one or more of the
default attributes for the column, it is generally considered good programming practice to have
one DEFINE statement for each variable in the COLUMN statement. Not only does this make the
PROC REPORT step easier to read, but it also helps during debugging and when making
modifications to the code. Another good programming practice (although it is not necessary to do
so) is to list the DEFINE statements in the same order as the items are listed in the COLUMN
statement.
DEFINE statements associated with variables that do not appear on the COLUMN statement
result in a warning being written to the log, but do not stop the processing of the step.
Depending on the define usages (or define types) specified in your PROC REPORT step, one of
two general types of reports are generated. A given report can contain elements of both of these
types.
Detail reports are roughly analogous to PROC PRINT reports. Data observations appear
individually in the generated report.
Summary reports are roughly analogous to the type of summary generated by PROC
MEANS. Individual data observations are collapsed (summarized) into groups.
2.2.1 Default Define Types DISPLAY and ANALYSIS
When not otherwise specified, the default define usages are DISPLAY for character variables and
ANALYSIS for numeric variables. These default define usages cause the generation of a detail
report. The PROC REPORT step in Section 2.1 could have also included the DEFINE statements
shown here:
title1 'Using Proc REPORT';
title2 'Including DEFINE Statements with Defaults';
proc report data=sashelp.class nowd;
column name sex age height;
define name
/ display;
define sex
/ display;
define age
/ analysis;
define height / analysis;
run;
Although you cannot use ANALYSIS for a character variable, in simple PROC REPORT steps
such as this one, it does not particularly matter whether you use ANALYSIS or DISPLAY for
numeric variables. As your reports become more complex, you will find that some operations,
such as the calculation of statistics, can only be accomplished on numeric variables with a define
type of ANALYSIS.
When the report includes summary lines using the BREAK and RBREAK statements (see
Section 3.1), the difference between ANALYSIS and DISPLAY can become more apparent for
numeric variables. Numeric variables with a define type of DISPLAY are generally not
summarized, whereas those with a define type of ANALYSIS are summarized on the summary
lines.
Chapter 2: PROC REPORT: An Introduction 19
2.2.2 Using Define Usage ORDER
The define usage ORDER allows you to change the order of the rows of the table without first
performing a sort.
title1 'Using Proc REPORT';
title2 'Define Type ORDER';
proc report data=sashelp.class nowd;
column name sex age height;
define sex
/ order;
define name
/ display;
define age
/ analysis;
define height / analysis;
run;
The order of the DEFINE statements has nothing to do with the order of the variables in the report
itself. In this example, SEX is the first DEFINE statement, but not the first variable on the
COLUMN statement. Usually the DEFINE statements are written in the same order as their
associated items on the COLUMN statement.
Inspection of the resulting report shows that the observations have been ordered by SEX. By
default the value of the ORDER variable is only printed for the first detail row for that group of
rows.
Using Proc REPORT
Define Type ORDER
Name
Alice
Barbara
Carol
Jane
Janet
Joyce
Judy
Louise
Mary
Alfred
Henry
James
Jeffrey
John
Philip
Robert
Ronald
Thomas
William
S
e
x
F
M
Age
13
13
14
12
15
11
14
12
15
14
14
12
13
12
16
12
15
11
15
Height
56.5
65.3
62.8
59.8
62.5
51.3
64.3
56.3
66.5
69
63.5
57.3
62.5
59
72
64.8
67
57.5
66.5
Destination: PDF Style: PRINTER
20 Carpenter’s Complete Guide to the SAS REPORT Procedure
The columns are ordered as they are specified in the COLUMN statement (name sex age
height). Usually, when we order the columns on a report, the columns used to form the
ordering are to the left. Exchanging the positions of the variables NAME and SEX on the
COLUMN statement results in a cleaner looking table.
title1 'Using Proc REPORT';
title2 'Define Type ORDER';
proc report data=sashelp.class nowd;
column sex name age height;
define sex
/ order;
define name
/ display;
define age
/ analysis;
define height / analysis;
run;
Using Proc REPORT
Define Type ORDER
S
e
x
F
M
Name
Alice
Barbara
Carol
Jane
Janet
Joyce
Judy
Louise
Mary
Alfred
Henry
James
Jeffrey
John
Philip
Robert
Ronald
Thomas
William
Age
13
13
14
12
15
11
14
12
15
14
14
12
13
12
16
12
15
11
15
Height
56.5
65.3
62.8
59.8
62.5
51.3
64.3
56.3
66.5
69
63.5
57.3
62.5
59
72
64.8
67
57.5
66.5
Because SEX is the first variable
listed in the COLUMN statement, it
is now the leftmost variable in the
table.
You are not limited to a single ORDER variable. As is true when you sort with multiple variables
in the BY statement, when you specify multiple ORDER variables, the variables to the right are
nested within those to the left (on the COLUMN statement). Here the table has been ordered by
AGE within SEX.
title1 'Using Proc REPORT';
title2 'Define Type ORDER';
proc report data=sashelp.class nowd;
column sex age name height;
define sex
/ order;
define age
/ order;
define name
/ display;
define height / analysis;
run;
Chapter 2: PROC REPORT: An Introduction 21
Using Proc REPORT
Define Type ORDER
S
e
x
F
Age
11
12
13
14
15
M
11
12
13
14
15
16
Name
Joyce
Jane
Louise
Alice
Barbara
Carol
Judy
Janet
Mary
Thomas
James
John
Robert
Jeffrey
Alfred
Henry
Ronald
William
Philip
Height
51.3
59.8
56.3
56.5
65.3
62.8
64.3
62.5
66.5
57.5
57.3
59
64.8
62.5
69
63.5
67
66.5
72
AGE has been sorted within SEX.
Notice how only the first row for
each group of the ORDER variables
is displayed.
By default the sort order is ascending; however, a DESCENDING option is available (see
Section 6.1.1). You can also change the default method of reordering rows by using the ORDER=
option, which is introduced in Section 2.4.3.
BY-group processing can be used with PROC REPORT. However, as in other SAS procedures,
the use of the BY statement does assume a sorted incoming data table. When BY-group
processing is used, a separate report is produced for each level of the BY variable.
proc sort data=sashelp.class
out=sclass;
by sex;
run;
title1 'Using Proc REPORT';
title2 'Define Type ORDER';
title3 'Using a BY';
proc report data=sclass nowd;
by sex;
column age name height;
define age
/ order;
define name
/ display;
define height / analysis;
run;
22 Carpenter’s Complete Guide to the SAS REPORT Procedure
The report displays the values for males as follows:
Using Proc REPORT
Define Type ORDER
Using a BY
Sex=M
Age
11
12
Name
Height
Thomas
57.5
James
57.3
John
59
Robert
64.8
13 Jeffrey
62.5
14 Alfred
69
Henry
63.5
15 Ronald
67
William
66.5
16 Philip
72
. . . portions of the report not shown . . .
Remember that for this example and
for most of the early examples in this
book, the system option
NOCENTER has been set. This leftjustifies tables and text for easier
inclusion in this book.
Although in this report SEX was not included on the COLUMN statement, it could have appeared
on both the COLUMN statement and the BY statement.
2.2.3 Using Define Type GROUP
The GROUP define type is used for the consolidation of rows. Usually this is done for summary
purposes. Using this define type also implies an ordering, and like the ORDER define type,
GROUP has a default order of ascending. As with the ORDER define type, this default order can
be modified through the use of the ORDER= option (see Section 2.4.3) or the DESCENDING
option (see Section 6.1.1).
In the previous examples, we have been creating detail reports that have implied groupings for
SEX and AGE. The following example actually consolidates the observations into the two groups
(one for each gender).
title1 'Using Proc REPORT';
title2 'Define Type GROUP';
proc report data=sashelp.class nowd;
column sex height weight;
define sex
/ group;
define height / analysis;
define weight / analysis;
run;
We have not specified how to summarize the numeric ANALYSIS variables HEIGHT and
WEIGHT. By default, ANALYSIS variables are totaled (this is the SUM statistic).
Chapter 2: PROC REPORT: An Introduction 23
Using Proc REPORT
Define Type GROUP
S
e
x
F
M
Height
545.3
639.1
Weight
811
1089.5
Unlike PROC PRINT, PROC
REPORT does not specify the same
formatting by default for all the values
within a column. The resulting numeric
columns can be difficult to read.
Section 2.4.2 discusses a solution to this
problem.
Because we are grouping on (or consolidating across) SEX, the variables AGE and NAME no
longer make sense and have been removed from the COLUMN statement.
GROUP is not always used to consolidate observations. When GROUP is used along with a
nongrouped DISPLAY variable on the COLUMN statement, grouping occurs without
consolidation. This technique generates a note in the LOG and produces a detail report, while at
the same time changing the way that the grouping variable is displayed. In this example, HEIGHT
has been changed to DISPLAY, and groups cannot be consolidated.
title1 'Using Proc REPORT';
title2 'Define Type GROUP';
title3 'With a DISPLAY variable';
proc report data=sashelp.class nowd;
column sex height weight;
define sex
/ group;
define height / display;
define weight / analysis;
run;
Inclusion of a nongrouped DISPLAY variable negates the ability to form groups. Although
groups cannot be formed, the group variable SEX is displayed only once for each level;
effectively GROUP has become ORDER.
Using Proc REPORT
Define Type GROUP
With a DISPLAY variable
S
e
x
F
M
Height
56.5
65.3
62.8
59.8
62.5
51.3
64.3
56.3
66.5
69
63.5
57.3
62.5
59
72
64.8
67
57.5
66.5
Weight
84
98
102.5
84.5
112.5
50.5
90
77
112
112.5
102.5
83
84
99.5
150
128
133
85
112
This is a detail report rather than a
summary report (there is one line in
the report for each incoming
observation).
This type of report is a direct result
of the SEX variable no longer being
able to form groups as it had been
able to do in the previous example,
because the variable HEIGHT has
been assigned a usage of DISPLAY.
24 Carpenter’s Complete Guide to the SAS REPORT Procedure
As an aside, notice that although EDU appears in a DEFINE statement, it does not also appear on
the COLUMN statement. Consequently EDU does not appear in the report.
Nested groups can be formed by using GROUP on more than one DEFINE statement. When you
do this, the second grouping variable is nested within the first. The order of the nesting is
determined by the order of the variables in the COLUMN statement and not by the order of the
DEFINE statements.
title1 'Using Proc REPORT';
title2 'Two Define Statements with Type GROUP';
proc report data=sashelp.class nowd;
column sex age height weight;
define sex
/ group;
define age
/ group;
define height / analysis;
define weight / analysis;
run;
There is now a single line for each value of AGE within each value of SEX. This collapses several
rows into a single group and allows group summarization.
Using Proc REPORT
Two Define Statements with Type GROUP
S
e
x
F
M
Age
11
12
13
14
15
11
12
13
14
15
16
Height
51.3
116.1
121.8
127.1
129
57.5
181.1
62.5
132.5
133.5
72
Weight
50.5
161.5
182
192.5
224.5
85
310.5
84
215
245
150
Because AGE is to the right of
SEX in the COLUMN statement,
groups for AGE are nested
within SEX.
2.2.4 Using Define Type ACROSS (and GROUP)
In the previous example, the variable AGE is grouped within SEX. This grouping results in one
row in the table for each SEX-by-AGE combination. A more efficient table could be achieved if
there was a separate column for each value of SEX. Effectively we want to perform a transpose
operation on SEX so that each value of SEX would form a column. Although PROC
TRANSPOSE can be used, it is more direct to transpose rows into columns by using the ACROSS
define type.
Chapter 2: PROC REPORT: An Introduction 25
Although not required, generally ACROSS is used when another variable has also been defined as
a GROUP variable. In the following report, SEX is designated as an ACROSS variable. Because
it is to the right of the GROUP variable (AGE) in the COLUMN statement, values of SEX are
nested within AGE.
title1 'Using Proc REPORT';
title2 'Define Types GROUP and ACROSS';
title3 'No ANALYSIS or DISPLAY Variables';
proc report data=sashelp.class nowd;
column age sex;
define age
/ group;
define sex
/ across;
run;
Because we have not requested otherwise, the number of observations for each SEX-by-AGE
combination is presented under the values of the student’s sex. This default behavior is changed in
the example in Section 6.3.
Using Proc REPORT
Define Types GROUP and ACROSS
No ANALYSIS or DISPLAY Variables
Age
11
12
13
14
15
16
Sex
F M
1 1
2 3
2 1
2 2
2 2
. 1
In the LISTING destination:
By default, since SEX is a $1 variable,
only one space is allocated for the
student counts. If the count had
exceeded 9, an asterisk would have been
printed.
2.3 Doing More on the COLUMN Statement
The COLUMN statement can be used to do much more than simply list the variables that are to be
available in the PROC REPORT step. Although some of the functionality available through the
COLUMN statement can be duplicated through other statements and options, some are unique to
the COLUMN statement. The following functionality of the COLUMN statement is discussed in
this book:
Section 2.3.1
discusses the formation of nested associations
Section 2.3.2
describes the association of statistics with analysis variables
Section 2.3.3
explains the use of parentheses to form groups of variables and statistics
Section 2.5.1
explains the addition of text in the header areas of the report
26 Carpenter’s Complete Guide to the SAS REPORT Procedure
2.3.1 Using the Comma to Form Nested Associations
In each of the preceding examples, the variables in the COLUMN statement were separated by
blanks. It is also possible to use a comma ( , ) to separate two or more items on the COLUMN
statement. When items are separated by a comma, the item on the right is nested within the item
on the left.
In the example in Section 2.2.4, the variable SEX has a define usage of ACROSS, and because
there are no analysis variables, the number of observations (N) is displayed for each combination
of AGE and SEX. In the following example, the numeric variable WEIGHT has been added to the
COLUMN statement. Notice that a comma separates SEX and WEIGHT in the COLUMN
statement. The comma forms an association with WEIGHT nested within SEX.
title1 'Using Proc REPORT';
title2 'Using the Comma to Form Associations';
title3 'WEIGHT is Nested Within SEX';
proc report data=sashelp.class nowd;
column age sex,weight;
define age
/ group;
define sex
/ across;
define weight / analysis;
run;
Using Proc REPORT
Using the Comma to Form Associations
WEIGHT is Nested Within SEX
Age
11
12
13
14
15
16
Sex
F
Weight
50.5
161.5
182
192.5
224.5
.
M
Weight
85
310.5
84
215
245
150
There is only limited control of the
appearance of the columns when the
statistics are assigned on the
COLUMN statement. Additional
control is available when statistics are
assigned on the DEFINE statement
(see Section 2.4).
As was the case in Section 2.2.3, when groups are formed in the presence of an analysis variable,
the value displayed is, by default, the total (SUM) of all the observations in the group. In this
case, the total weight of all the 12-year-old females in the data table is 161.5 pounds. Other
statistics are available and are discussed in Section 2.3.2.
2.3.2
Attaching Statistics with a Comma
The standard set of summary statistics (N, SUM, MEAN, VAR, STD, MEDIAN, etc.) are
available for use within PROC REPORT. These are the same summary statistics (with the same
names) that you can request from other procedures, such as MEANS, SUMMARY, and
UNIVARIATE.
These statistics can be requested in several ways. One way is to include the name of the desired
statistic on the COLUMN statement, using a comma to form the association between a statistic
and the variable for which that statistic is to be calculated.
Chapter 2: PROC REPORT: An Introduction 27
title1 'Using Proc REPORT';
title2 'Using the Comma to Attach Statistics';
title3 'Mean WEIGHT and HEIGHT';
proc report data=sashelp.class nowd;
column age weight,mean height,mean;
define age
/ group;
define weight / analysis;
define height / analysis;
run;
Because AGE is defined as a grouping variable, the mean for HEIGHT and WEIGHT will be
within age groups.
Using Proc REPORT
Using the Comma to Attach Statistics
Mean WEIGHT and HEIGHT
Age
11
12
13
14
15
16
Weight
mean
67.75
94.4
88.666667
101.875
117.375
150
Height
mean
54.4
59.44
61.433333
64.9
65.625
72
The name of the requested statistic, in
this case the MEAN, has automatically
been added to the column header.
Statistics can be associated only with
numeric variables.
The statistic can either follow the variable name (as was the case in the previous example), or it
can precede the variable name. When the statistic precedes the name, the label is placed higher in
the column header.
title1 'Using Proc REPORT';
title2 'Using the Comma to Attach Statistics';
title3 'Mean WEIGHT and HEIGHT';
proc report data=sashelp.class nowd;
column age weight,mean
Mean,height n;
define age
/ group;
define weight / display; o
define height / analysis;
run;
n The statistic has been placed before the analysis variable. This position nests the variable
within the statistic and causes the order of the labels in the header to reverse. Notice that the
case used in the name of the statistic (Mean) is preserved in the label.
o The statistic can be associated with a numeric variable with the define usage of DISPLAY.
28 Carpenter’s Complete Guide to the SAS REPORT Procedure
Using Proc REPORT
Using the Comma to Attach Statistics
Mean WEIGHT and HEIGHT
Age
11
12
13
14
15
16
Weight
Mean
mean
67.75
94.4
88.666667
101.875
117.375
150
Height
54.4
59.44
61.433333
64.9
65.625
72
The label for the statistic can be placed
either above or below the label for the
variable.
n
The case of the statistic’s name that is
used in the COLUMN statement is
preserved in the column header in the
report.
Statistics must be associated with a numeric variable. This association can be done on the
COLUMN statement, as it is here, or on the DEFINE statement (see Section 2.4). The only
exception is the N statistic, which can appear on the COLUMN statement without being
associated with a numeric variable. When this happens, N counts observations (see Section 2.4.4
for an example).
SEE ALSO
SAS Technical Report P-258 (1993, pp. 25, 195) discusses the stacking of variables and statistics.
2.3.3 Using Parentheses to Form Groups
In the examples in Section 2.3.2, the mean was calculated for both height and weight. This
resulted in the specification of the same statistic for each analysis variable. This syntax can be
simplified by forming groups through the use of parentheses on the COLUMN statement. You can
use the parentheses to form groups of items that include variables, statistics, and even text for
labels. The following COLUMN statement was used in the previous section:
column age weight,mean
height,mean;
This statement could be replaced with the following more succinct statement:
column age (weight height),mean;
The same technique allows you to assign multiple statistics to the same analysis variable. The
following example displays three statistics for WEIGHT within AGE:
title1 'Using Proc REPORT';
title2 'Using Parentheses to form Groups';
title3 'Grouping Statistics';
proc report data=sashelp.class nowd;
column age weight,(N Mean Stderr);
define age
/ group;
define weight / analysis;
run;
Chapter 2: PROC REPORT: An Introduction 29
Using Proc REPORT
Using Parentheses to form Groups
Grouping Statistics
Age
11
12
13
14
15
16
N
2
5
3
4
4
1
Weight
Mean
67.75
94.4
88.666667
101.875
117.375
150
Stderr
17.25
9.1806862
4.6666667
4.6069467
5.2096665
.
Destination: PDF Style: JOURNAL
2.3.4 Nesting Statistics under an ACROSS Variable
You are not limited to a single level of nesting. In the following example, statistics are displayed
for WEIGHT within each value of SEX, which has been defined as an ACROSS variable.
title1 'Using Proc REPORT';
title2 'Using Parentheses to form Groups';
title3 'Grouping Under an ACROSS Variable';
proc report data=sashelp.class nowd;
column age sex,weight,(N Mean);
define age
/ group;
define sex
/ across;
define weight / analysis;
run;
The resulting table shows the statistics for WEIGHT for each level of SEX.
Using Proc REPORT
Using Parentheses to form Groups
Grouping Under an ACROSS Variable
Age
11
12
13
14
15
16
Sex
F
Weight
N
Mean
1
50.5
2
80.75
2
91
2
96.25
2
112.25
.
.
M
Weight
N
1
3
1
2
2
1
Mean
85
103.5
84
107.5
122.5
150
The statistic does not
have to be the last
item in the list. We
could nest SEX
within the analysis
variable so that we
could make easy
comparisons between
genders.
30 Carpenter’s Complete Guide to the SAS REPORT Procedure
If we had failed to nest WEIGHT within SEX, the summary for WEIGHT would have been
independent of SEX.
title1 'Using Proc REPORT';
title2 'Using Parentheses to form Groups';
title3 'An ACROSS Variable';
title4 'With Non-nested Statistics';
proc report data=sashelp.class nowd;
column age sex weight,(N Mean);
define age
/ group;
define sex
/ across;
define weight / analysis;
run;
Because no statistics are associated with the ACROSS variable, the number of students (N) is
displayed under SEX.
Using Proc REPORT
Using Parentheses to form Groups
An ACROSS Variable
With Non-nested Statistics
Age
11
12
13
14
15
16
Sex n
F M
1 1
2 3
2 1
2 2
2 2
. 1
Weight Y
N
Mean
2
67.75
5
94.4
3 88.666667
4
101.875
4
117.375
1
150
By including a non-nested ANALYSIS
variable, we have effectively created a
horizontally concatenated report. The first
report is for SEX n and the second for
WEIGHT Y. The ability to generate multiple
types of concatenated reports within a single
report is a very powerful feature of the
REPORT procedure.
This report is interesting for several reasons. Because AGE is a GROUP variable, there is one row
per AGE. Therefore the N and MEAN for the variable WEIGHT are across all observations
within an AGE, and the statistics ignore SEX. Also, we did not tell PROC REPORT what we
wanted to show under the columns ‘F’ and ‘M’ n. Consequently, the counts are displayed and we
see the number of students within each AGE-by-SEX combination.
If the same PROC REPORT step were re-executed without a GROUP variable (but still with the
ACROSS variable), there would be one row in the table for each observation in the data set. When
we calculate statistics, we generally need to group observations to make the statistics meaningful.
Chapter 2: PROC REPORT: An Introduction 31
2.4 Other DEFINE Statement Options
In all of the previous examples in this chapter, the display of information has primarily relied on
defaults. Although the results have usually been at least acceptable in appearance, we can do
better. Fortunately, a number of supplemental options can be used with the DEFINE statement to
augment the display of the information. These include the following:
statistic
specifies a statistic to be calculated (used only with analysis variables).
FORMAT= specifies how the information within this column is to be formatted.
ORDER=
determines a scheme for ordering the values.
MORE INFORMATION
This is definitely not an exhaustive list of the options that can be used with the DEFINE
statement. A number of options specific to the LISTING destination are demonstrated in
Chapter 4, “Only in the LISTING Destination.” Additional DEFINE statement options are also
discussed in Section 6.1.
SEE ALSO
Burlew (2005, p. 16) includes an example with a variety of DEFINE statement options.
2.4.1 Specification of an Analysis Statistic
Although you can specify the calculation of one or more statistics on the COLUMN statement
(see Section 2.3), it is usually easier and more straightforward to specify the statistics on the
DEFINE statement. All of the same statistics that are available in PROC TABULATE (except for
SKEWNESS and KURTOSIS) can be specified on either the COLUMN statement or on the
DEFINE statement. These statistics include SUM (the default), N, MEAN, VARIANCE, and
MEDIAN. You can specify only one statistic per DEFINE statement; however, it is possible to
specify multiple DEFINE statements per variable through the use of aliases (see Section 6.2) or
by creating columns with compute blocks (see Chapter 5, “Creating and Modifying Columns
Using the Compute Block”).
The examples in this section use the data from a small health care study. In the following
example, the average (mean) weight is calculated. Notice that since no statistic is requested for the
EDU (years of education) variable, the SUM is calculated.
title1 'Using Proc REPORT';
title2 'Specify the MEAN on the DEFINE Statement';
proc report data=rptdata.clinics nowd;
column region sex edu wt;
define region / group;
define sex
/ across;
define wt
/ analysis mean;
define edu
/ analysis;
run;
32 Carpenter’s Complete Guide to the SAS REPORT Procedure
In this particular example, the order of the variables in the COLUMN statement is not the same as
the order of the DEFINE statements. As far as PROC REPORT is concerned, the order of the
DEFINE statements is not important. However, the order of the columns depends on the order of
the variables in the COLUMN statement. Here is the report in the LISTING destination:
Using Proc REPORT
Specify the MEAN on the DEFINE Statement
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
years of
education
40
82
136
142
210
116
140
60
56
120
weight
in pounds
195
172.33333
107.8
145.8
159.14286
157.75
198
151
160
187.8
Although the formatting of the EDU
column is acceptable (EDU is an
integer), the values for mean weight
are ugly and hard to read. This
issue is addressed in the next section.
Notice that the number of male patients in region 4 exceeds 9, and because there is only one space
allocated (SEX is a $1 variable), an asterisk is displayed instead of the number. This loss of
information is not an issue for other destinations. Here is the same table generated for the PDF
destination using the RTF style:
The RTF style is optimized for
use with the RTF destination;
however, it can also be used
with other destinations as well.
Destination: PDF Style: RTF
Chapter 2: PROC REPORT: An Introduction 33
SEE ALSO
Rohowsky (2005) creates a list of statistics.
2.4.2 Formatting the Values
As in a majority of procedures that display values, you can apply a format directly to a variable
through a FORMAT statement. In a PROC REPORT step you can also specify the format on the
DEFINE statement itself by using the FORMAT= option.
In the following PROC REPORT step, the mean for both height (HT) and weight (WT) are
calculated.
title1 'Using Proc REPORT';
title2 'MEAN for Height and Weight';
proc report data=rptdata.clinics nowd;
column region sex ht wt;
define region / group;
define sex
/ across;
define ht
/ analysis mean format=4.1;
define wt
/ analysis mean;
format wt 6.2;
run;
This example shows that both the FORMAT statement and the FORMAT= option can be used. In
the LISTING destination, shown here, the format can also be used to control the width of the
column so that the headings are more readable.
Using Proc REPORT
MEAN for Height and Weight
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
heig
ht
in
inch
es
74.0
67.0
63.8
68.2
69.0
66.0
66.6
66.0
70.0
67.4
weight
in
pounds
195.00
172.33
107.80
145.80
159.14
157.75
198.00
151.00
160.00
187.80
The width of the column for height
(HT) is still a bit too narrow, and the
label is split awkwardly. This is not a
problem for weight (WT), as the
format has a width that is two digits
wider. For ODS destinations other
than LISTING, column width is
generally not a problem.
MORE INFORMATION
More robust methods for controlling column width are available. For the LISTING destination,
COLWIDTH (see Section 4.4.3) and WIDTH= (see Section 4.5.1) can also be used to control
column width independently from the formatted width. For other destinations, the CELLWIDTH
attribute modifier can be used (see Section 8.2.4).
SEE ALSO
SAS Technical Report P-258 (1993, pp. 46–47) discusses the use of the FORMAT= option on the
DEFINE statement.
34 Carpenter’s Complete Guide to the SAS REPORT Procedure
2.4.3 Controlling the Order of the Displayed Values
In Section 2.2.2, the define type of ORDER was used to change the order of the rows of the table.
Within SAS there are several ways for you to determine the ordering criteria, and within PROC
REPORT these criteria are controlled through the use of the ORDER= option. This same option is
used in a number of other procedures, primarily when there is a classification or some other
similar type of variable that is used to build the table. In PROC REPORT this option can be used
on DEFINE statements with a GROUP, ORDER, or ACROSS define type.
The ORDER= option can take on values such as the following:
DATA
The order is based on the incoming data.
FORMATTED
Values are first formatted and then ordered on their formatted (external)
values.
FREQ
The ascending frequency count is used to determine the order.
INTERNAL
As in PROC SORT, the unformatted values are used. This sort sequence is
particularly useful for displaying dates chronologically.
The following example shows the use of the ORDER=FREQ option to display the regions in
ascending frequency.
title1 'Using Proc REPORT';
title2 'Ordering Values with ORDER=FREQ';
proc report data=rptdata.clinics nowd;
column region sex ht wt;
define region / group order=freq;
define sex
/ across;
define ht
/ analysis mean format=4.1;
define wt
/ analysis mean;
format wt 6.2;
run;
Using Proc REPORT
Ordering Values with ORDER=FREQ
re
gi
on
7
1
8
10
5
3
2
9
6
4
pati
F M
. 4
. 4
4 .
2 4
5 3
5 5
6 4
2 8
4 6
4 *
heig
ht
in
inch
es
66.0
74.0
70.0
67.0
66.0
68.2
63.8
67.4
66.6
69.0
weight
in
pounds
151.00
195.00
160.00
172.33
157.75
145.80
107.80
187.80
198.00
159.14
The order of the regions is determined
by the number of patients in each
region (with one observation per
patient visit, the order is actually
based on the number of observations
within REGION). Regions with the
fewest patients are listed first.
The DESCENDING option can also be used on the DEFINE statement, as it can be with the BY
statement. Here the previous example is repeated using the DESCENDING option.
Chapter 2: PROC REPORT: An Introduction 35
title1 'Using Proc REPORT';
title2 'Ordering Values with ORDER=FREQ';
proc report data=rptdata.clinics nowd;
column region sex ht wt;
define region / group order=freq descending;
define sex
/ across;
define ht
/ analysis mean format=4.1;
define wt
/ analysis mean;
format wt 6.2;
run;
The order of the regions is determined
by the number of patients in each
region. Because of the inclusion of the
DESCENDING option, regions with
the most observations (patients) are
listed first.
Destination: PDF Style: SANSPRINTER
By happenstance, the first observation on the data set RPTDATA.CLINICS is for a male patient.
If we use ORDER=DATA for the ACROSS variable SEX, the order of the two columns for SEX
are reversed.
title1 'Using Proc REPORT';
title2 'Ordering Values with ORDER=DATA';
proc report data=rptdata.clinics nowd;
column region sex ht wt;
define region / group;
define sex
/ across
order=data;
define ht
/ analysis mean format=4.1;
define wt
/ analysis mean;
format wt 6.2;
run;
36 Carpenter’s Complete Guide to the SAS REPORT Procedure
Using Proc REPORT
Ordering Values with ORDER=DATA
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
M F
4 .
4 2
4 6
5 5
* 4
3 5
6 4
4 .
. 4
8 2
heig
ht
in
inch
es
74.0
67.0
63.8
68.2
69.0
66.0
66.6
66.0
70.0
67.4
weight
in
pounds
195.00
172.33
107.80
145.80
159.14
157.75
198.00
151.00
160.00
187.80
The column for males is before females,
because the first observation on the
incoming data is for a male
(ORDER=DATA). The default (previous
example) is for DATA=INTERNAL,
which places the columns in
alphabetically ascending order.
MORE INFORMATION
Section 6.3 uses the ORDER=FORMAT option to control the order based on a user-defined
format.
2.4.4 Using the N Statistic without an ANALYSIS Variable
When you want to count observations within a group, you can use the N statistic on the
COLUMN statement. Unlike other statistics, which must be associated with an analysis variable,
N is a special case and can be used alone. In the following example, N is used in two different
ways. It appears on the COLUMN statement, both as a statistic that is not associated with any
variable n, and as a statistic nested within the procedure variable (PROCED) o.
title1 'Using Proc REPORT';
title2 'Using N';
proc report data=rptdata.clinics nowd;
column region n n proced,n o;
define region / group;
define proced / across;
run;
n N is not associated with any analysis variable and therefore counts observations.
o This occurrence of N is a counter of procedure codes within REGION.
Because REGION is a GROUP variable and PROCED is an ACROSS variable, we have
effectively nested procedure code within REGION. This nesting is also often referred to as
crossing REGION and procedure code.
Chapter 2: PROC REPORT: An Introduction 37
Using Proc REPORT
Using N
re
procedure code
gi
n
1
2
3
on
1
10
2
3
5
6
7
8
n
4
6
8
6
6
4
2
4
n
.
.
.
.
.
.
2
.
n
.
6
.
.
4
4
.
2
n
4
.
8
6
2
.
.
2
9
2
.
.
2
o
In the previous example (Section 2.4.3)
there were 10 patients in region 2,
however in this table there are only 8. It
turns out that the variable PROCED
has missing values, and by default
missing values are excluded from the
table. This exclusion effects the overall
value of N as well. The MISSING
option is used to include missing values
of a classification variable (see Section
6.1.3).
MORE INFORMATION
Observations with missing values for classification variables are not counted unless the MISSING
option (see Section 6.1.3) is included. Compute blocks and aliases can also be associated with the
N statistic (see Section 7.8.1).
SEE ALSO
The N statistic is used in an example in SAS Technical Report P-258 (1993, pp. 245–248).
2.4.5 Associating Statistics with DEFINE Statements
In the example in Section 2.4.4, there are four columns that contain counts. However, if we want
to control the characteristics of the individual columns, we need a DEFINE statement. If we try to
format the counts for each of the procedure codes using something like the following statement,
we get an error:
define proced / across format=2.;
This error occurs because defining the character variable PROCED as an ACROSS variable
applies the format to the values of PROCED in the header and not to the counts.
We can get around this dilemma by adding a DEFINE statement for the statistic itself. This
DEFINE statement does not have a define type and exists solely to allow the placement of
DEFINE statement options. The example from Section 2.4.4 becomes the following:
title1 'Using Proc REPORT';
title2 'Using N with a Format';
proc report data=rptdata.clinics nowd;
column region n proced,n;
define region / group;
define proced / across;
define n
/ format=2.;
run;
Notice that the format was applied to all of the N statistics, including those associated with the
ACROSS variable.
38 Carpenter’s Complete Guide to the SAS REPORT Procedure
Using Proc REPORT
Using N with a Format
re
gi
on
1
10
2
3
5
6
7
8
9
n
4
6
8
6
6
4
2
4
2
procedure code
1
2
3
n
n
n
.
.
4
.
6
.
.
.
8
.
.
6
.
4
2
.
4
.
2
.
.
.
2
2
.
.
2
In the previous example, the format on the DEFINE N statement is applied to each of the columns
that have an N statistic. If we did not specify the statistic to be used with the ACROSS variable,
the result would still have the counts (because that is the default), but the format on the DEFINE
N statement would not apply to the counts for PROCED.
* Using N without an ANALYSIS variable;
title1 'Using Proc REPORT';
title2 'No Statistic for PROCED';
proc report data=rptdata.clinics nowd;
column region n proced;
define region / group;
define proced / across;
define n
/ format=8.;
run;
The FORMAT=8. has been applied only to the one N column. Note that in the LISTING
destination shown here, the columns that contain the procedure code counts have a width of one.
This width is applied because the N statistic is not explicitly requested for PROCED, and
PROCED has a width of one.
Using Proc REPORT
No Statistic for PROCED
re
gi
on
1
10
2
3
5
6
7
8
9
n
4
6
8
6
6
4
2
4
2
procedure code
1 2 3
. . 4
. 6 .
. . 8
. . 6
. 4 2
. 4 .
2 . .
. 2 2
. . 2
In Section 2.4.4, the N statistics are
not explicitly formatted and therefore
receive a default format width. When
only one column for N is formatted, as
is done in this example, the
unformatted columns receive a width
based on the width of the ACROSS
variable (PROCED is $1).
In addition to N, the DEFINE statement can be used for any of the other statistics that appear on
the COLUMN statement. This allows you to control text and formatting for individual report item
statistics that are introduced in the COLUMN statement.
Chapter 2: PROC REPORT: An Introduction 39
* Using Define Statements with Statistics;
title1 'Using Proc REPORT';
title2 'Using DEFINE Statements for Statistics';
proc report data=rptdata.clinics nowd;
column region wt,(n mean var);
define region / group;
define n
/ format=4.;
define mean / format=6.2;
define var / format=7.2;
run;
Using Proc REPORT
Using DEFINE Statements for Statistics
re
gi
on
1
10
2
3
4
5
6
7
8
9
weight in pounds
n
mean
var
4 195.00
0.00
6 172.33
52.27
10 107.80
17.07
10 145.80 1238.84
14 159.14
562.59
8 157.75 1957.93
10 198.00
565.33
4 151.00
21.33
4 160.00
5.33
10 187.80
726.40
DEFINE statements are used to supply
formats to the report items, which in
this case are statistics and not
variables.
Because there is only one column for each statistic, each of these DEFINE statements necessarily
applies to only one column. However, if we had asked for the mean of both height and weight,
the DEFINE MEAN statement would have applied to both means. This happened in the first
example in this section. When you need to control the options differently for two columns with
the same statistic, it is usually easier to declare an alias (see Section 6.2), which can have its
own DEFINE statement.
2.5 Adding Text
The default text associated with our report is limited to titles, footnotes, and column labels. As is
quite apparent from all of the preceding LISTING examples in this chapter, this text is not
adequate for any but the simplest of reports. We need to be able to take control of the text that
appears on our report. We need to be able to augment the text, modify it, and add new text such as
headers that span columns. Fortunately, there are a number of ways to do this, and these form a
great deal of the discussion throughout the remainder of this book.
As was stated earlier, when no other action is taken, column headers are built from either the
column label (when present) or the column name. For the LISTING destination, the width
requirements of the values determine the column widths, and the labels are often unreadable when
the column is narrow. For many of the other destinations, the column labels are by default much
more legible; however, we still often need more control.
40 Carpenter’s Complete Guide to the SAS REPORT Procedure
The following PROC REPORT step includes the specifications of two ODS destinations. This
enables us to create two reports with a single PROC REPORT step. One will be written to the
LISTING destination (e.g., the OUTPUT window) and the other to an HTML file.
ods html file="&path\results\ch2_5.html";
ods listing;
* Using GROUP and ACROSS;
title1 'Using Proc REPORT';
title2 'Define Types GROUP and ACROSS';
proc report data=rptdata.clinics nowd;
column region sex wt,(n mean);
define region / group;
define sex
/ across;
define wt
/ analysis;
run;
ods html close;
ods listing close;
Notice the differences in the labels and headers for the two types of reports. Here is the HTML
report:
Unlike the LISTING destination, most nonmonospace destinations, such as the HTML
report shown here, are much better at
properly spacing columns to show all
values and labels.
Destination: HTML Style: DEFAULT
Chapter 2: PROC REPORT: An Introduction 41
Here is the LISTING report:
Using Proc REPORT
Define Types GROUP and ACROSS
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
weight in pounds
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
Although there is sufficient space
to display the counts for all the
regions in the HTML report, the
LISTING report is unable to
display the patient count for males
in region 4.
In the LISTING report, the variable REGION, which is a $2 variable without a label, has its text
stacked. The label for SEX (an ACROSS variable) is truncated.
In the resulting report for both destinations, the label for WT is used to span the two columns of
statistics (N and MEAN) calculated for WT. In a similar fashion, the label for SEX is used to span
the two columns generated by the two values of SEX. These are known as spanning headers and
can be very helpful in annotating a report. Fortunately, there are also other ways to create these
types of text headers.
There are two major types of headers:
those that span more than one report column
those that apply to each column individually
Two of the easiest ways to control and add headers to your report are through the DEFINE and
COLUMN statements.
MORE INFORMATION
Section 2.6.2 shows how text can also be added with the LINE statement in compute blocks, and
Chapter 4 deals exclusively with issues associated with reports written to the LISTING
destination (OUTPUT window).
SEE ALSO
Mitchell (2006) has a PROC REPORT example that prints only titles and no report.
2.5.1 Using the COLUMN Statement to Add Text
As was shown earlier in this chapter, parentheses can be used to group variables and statistics
(Sections 2.3.3 and 2.3.4). You can also use parentheses to create an association between a
variable and a text string. In the following example, the text “Gender” is added to the columns
that display the patient sex.
42 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Text headers in the COLUMN statement;
title1 'Using Proc REPORT';
title2 'Column Text';
proc report data=rptdata.clinics nowd;
column region ('Gender' sex) ;
define region / group;
define sex
/ across format=$3.;
run;
Using Proc REPORT
Column Text
re
gi
on
1
10
2
3
4
5
6
7
8
9
Gender
patient sex
F
M
.
4
2
4
6
4
5
5
4
10
5
3
4
6
.
4
4
.
2
8
Destination: RTF Style: RTF
Parentheses can also be used to group text headers that span columns. Again the spanning text
precedes the variables, groups of variables, statistics, and groups of statistics. Because parentheses
that form groups can be nested, you can easily create text associations in layers.
* Text headers in the COLUMN statement;
title1 'Using Proc REPORT';
title2 'Grouped Header';
proc report data=rptdata.clinics nowd;
column region sex ('Patient Weight (lb)' wt,(n mean));
define region / group;
define sex
/ group format=$6.;
define wt
/ analysis;
run;
Chapter 2: PROC REPORT: An Introduction 43
Using Proc REPORT
Grouped Header
re
gi
on
1
10
2
3
4
5
6
7
8
9
patien
t sex
M
F
M
F
M
F
M
F
M
F
M
F
M
M
F
F
M
Patient Weight (lb)
weight in pounds
n
mean
4
195
2
163
4
177
6 109.66667
4
105
5
127.8
5
163.8
4
143
10
165.6
5
146.2
3
177
4
187
6 205.33333
4
151
4
160
2
177
8
190.5
The text in the COLUMN statement
adds a new header, but does not alter
the existing headers.
Although the previous examples have added text, the original text supplied by PROC REPORT
has not been replaced. Thus your control of the text is limited. When you want to control the text
associated with an individual column, you can also use the text options associated with the
DEFINE statement (see Section 2.5.2).
2.5.2 Using the DEFINE Statement to Add Text
Text can also be added directly through the DEFINE statements by adding a text string. Unlike
text specified on the COLUMN statement, this text will replace the text that would otherwise
appear as the column header.
* Text headers in the DEFINE statement;
title1 'Using Proc REPORT';
title2 'DEFINE Statement Text';
proc report data=rptdata.clinics nowd;
column region sex ('Weight' wt);
define region / group
format=$6.;
define sex
/ across
format=$3.'Gender';
define wt
/ analysis mean '(Mean)';
run;
44 Carpenter’s Complete Guide to the SAS REPORT Procedure
Using Proc REPORT
DEFINE Statement Text
region
1
10
2
3
4
5
6
7
8
9
Gender
M
.
4
2
4
6
4
5
5
4
10
5
3
4
6
.
4
4
.
2
8
F
Weight
(Mean)
195
172.33333
107.8
145.8
159.14286
157.75
198
151
160
187.8
The column for mean weight now has
two sources of text for the header:
'Weight' from the COLUMN
statement and '(Mean)' from the
DEFINE statement.
SEE ALSO
Tsykalov and Yeh (2006) demonstrate several ways to suppress the text that would otherwise be
displayed. Their techniques include the use of a nonprintable character.
2.5.3 Using the SPLIT= Option with Text
As has been shown in a number of the previous examples, header text can be wrapped or
truncated depending on the use of the text and the available space. When header text is too long to
fit into the available space, or when you want to control how the header text is to be split, you can
use the SPLIT= option on the PROC REPORT statement to add some control. This option
specifies an unprinted character that is used to designate where the line break is to take place. This
is similar to the SPLIT= option in PROC PRINT.
In the following example, the text associated with the variable WT is split using the asterisk (*).
* Text headers in the DEFINE statement;
title1 'Using Proc REPORT';
title2 'Using SPLIT=';
proc report data=rptdata.clinics nowd split='*';
column region sex ('Weight*Pounds' wt);
define region / group
format=$6. ;
define sex
/ across
format=$3. 'Gender';
define wt
/ analysis mean
'(Mean)';
run;
Notice that the split character (*) does not appear in the header in the report.
Chapter 2: PROC REPORT: An Introduction 45
Using Proc REPORT
Using SPLIT=
region
1
10
2
3
4
5
6
7
8
9
Gender
F
M
.
4
2
4
6
4
5
5
4
10
5
3
4
6
.
4
4
.
2
8
Weight
Pounds
(Mean)
195
172.33333
107.8
145.8
159.14286
157.75
198
151
160
187.8
It took three different options on
three different statements to create
this header for the mean weight: the
SPLIT= option, text on the
COLUMN statement, and text on the
DEFINE statement.
Since not all ODS destinations and styles handle the splitting of text in the same way, the effect of
the SPLIT= option can be destination-specific.
SEE ALSO
Ping and Schiefelbein (2006) preprocess the data to add specialized split characters in order to
control where long data lines are to break.
2.6 Compute Blocks
Unlike most other SAS procedures, PROC REPORT has the ability to modify values within a
column, to insert lines of text into the report, to create columns, and to control the content of a
column. Through compute blocks, it is possible to use a number of SAS language elements, many
of which can otherwise only be used in the DATA step. As the report itself is built, the statements
associated with each compute block are executed.
There are two basic types of compute blocks: those that are associated with a location (the option
BEFORE or AFTER follows the COMPUTE keyword), and those associated with a report item.
Although the structure and execution of these two types of compute blocks are similar, how they
are used and the timing of their execution can be quite different. These differences are noted
throughout the sections dealing with compute blocks.
The compute block starts with the COMPUTE statement and terminates with the ENDCOMP
statement. Usually the compute block is placed in the PROC REPORT step after the DEFINE
statements. The syntax of the compute block looks something like this:
compute ;
. . . one or more SAS language elements . . .
endcomp;
46 Carpenter’s Complete Guide to the SAS REPORT Procedure
The COMPUTE statement includes the following components:
location
This component specifies when the compute block is to execute and ultimately what is to
be done with the result of the compute block. Accepted values include BEFORE and
AFTER. When a location is specified without also specifying a report_item, the location
is at the start (BEFORE) or at the end (AFTER) of the report. When location is used in
conjunction with the location modifier _PAGE_, the action of the compute block takes
place BEFORE or AFTER page breaks (see Section 5.4).
report_item
When the result of the compute block is associated with a variable or report item, its
name is supplied here. This report_item variable can be any variable on the COLUMN
statement. When report_item is a variable that either groups or orders rows (usage of
GROUP or ORDER) you can also use BEFORE and AFTER to apply the result at the
start or end of each group.
options
Several options are available that can be used to determine the appearance and location of
the result of the compute block.
SAS language elements
Any number of SAS language elements can be used within the compute block. These
include executable statements, logical processing (IF-THEN/ELSE), and most of the
functions available in the DATA step (see Section 2.6.4).
2.6.1 Inserting a Blank Line
One of the more interesting programming statements within the compute block is the LINE
statement. This statement is roughly analogous to the PUT statement in the DATA step and can be
used to introduce lines of text into the report.
Chapter 2: PROC REPORT: An Introduction 47
In the following example, this line of text is a blank line after each level of the grouping variable
(REGION).
* Blank Line using COMPUTE;
title1 'Using Proc REPORT';
title2 'Blank Line After Region';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
compute after region;
line ' ';
endcomp;
run;
Using Proc REPORT
Blank Line After Region
weight in pounds
n
mean
4
195
region
1
Gender
M
2
F
M
6
4
109.66667
105
3
F
M
5
5
127.8
163.8
4
F
M
4
10
143
165.6
The LINE statement has been
used to insert a blank line after
each regional group.
Because the COMPUTE statement contains the location specification AFTER and also designates
the report item REGION, we have effectively requested that the results of the compute block (a
blank space) be written after each group of the grouping variable REGION.
2.6.2 Adding Lines of Text
In the previous example, the LINE statement added a blank line. The LINE statement is more
commonly used to add lines of text at specific locations in the report.
In the following example, the text is written as a footnote at the end of the report. We can specify
the options BEFORE and AFTER to indicate the location, and since in this case no grouping
variable appears on the COMPUTE statement, the AFTER option applies to the whole PROC
REPORT step.
48 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Text Line using COMPUTE;
title1 'Using Proc REPORT';
title2 'Footnote Using LINE';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
compute after region;
line ' ';
endcomp;
compute after;
line @20 'Weight taken during';
line @20 'the entrance exam.';
endcomp;
run;
Using Proc REPORT
Footnote Using LINE
weight in pounds
n
mean
4
195
region
1
Gender
M
2
F
M
6
4
109.66667
105
3
F
M
5
5
127.8
163.8
4
F
M
4
10
143
165.6
This report was generated with
the system option NOCENTER
in effect (notice the titles). Even
so, the results of the LINE
statement are centered unless a
column specification is provided.
Weight taken during
the entrance exam.
In these LINE statements the @ is used, as it is in the DATA step PUT statement, to designate the
column number. If a specific column is not specified with the @, and no justification options are
specified, text generated by the LINE statement is centered. This is a different default behavior
than that of the PUT statement in the DATA step.
When writing to ODS destinations other than LISTING, proportional fonts might make exact
placement of values difficult, and might require you to use a trial-and-error approach. To make
things more interesting, some destinations ignore the @ altogether. Here is the same report
generated in RTF (using STYLE=RTF):
Chapter 2: PROC REPORT: An Introduction 49
Destination: RTF Style: RTF
MORE INFORMATION
The LINE statement is used in a compute block in an example that shows how to write a leftjustified line at the bottom of each page in Section 5.4.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 114–119) uses the $VARYING. format to control both
text and numeric variables in a LINE statement. The LINE statement and the PUT statement are
contrasted in P-258 on page 216.
2.6.3 Writing Formatted Values
You can place formatted variable values on the report by using the LINE statement, and you can
use the BEFORE or AFTER option to place these values before or after each group. In the
following example, the user-defined format $REGNAME provides a text name for the first four
regions. These names are than added to the report through the use of a LINE statement and a
compute block.
proc format;
value $regname
'1' = 'New England'
'2' = 'New York'
'3' = 'Maryland'
'4' = 'South East';
run;
50 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Text Line using COMPUTE;
title1 'Using Proc REPORT';
title2 'Formatted Values';
footnote1 'at the bottom';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
compute before region;
line @3 region $regname8.; n
endcomp;
compute after region;
line ' '; o
endcomp;
compute after;
line @20 'Weight taken during'; Z
line @20 'the entrance exam.';
endcomp;
run;
Using Proc REPORT
Formatted Values
region
Gender
New England
1
M
weight in pounds
n
mean
n
4
195
New York
2
F
M
6
4
109.66667
105
Maryland
3
F
M
5
5
127.8
163.8
South East
4
F
M
4
10
143
165.6
Three compute blocks are
used to generate the three
types of text: a formatted
region name before each
region n, a blank line after
each region summary o,
and footnote text at the end
of the report Z.
o
Weight taken during
the entrance exam.
Z
The format $REGNAME could have also been used in the DEFINE statement; however, in this
case we wanted to show the unformatted value as well as the formatted group header.
SEE ALSO
Burlew (2005, p. 121) makes extensive use of the LINE statement with formats.
Chapter 2: PROC REPORT: An Introduction 51
2.6.4 Using SAS Language Elements
Much of the power of the DATA step is available within the compute block. This radically
increases the flexibility of PROC REPORT, as most of the SAS language elements, such as
routines, functions, arithmetic operations, and executable statements, can also be used within
the compute block. These include DO loops, assignment and SUM statements, arrays, and
IF-THEN/ELSE processing.
Throughout this book there are numerous examples of the use of these SAS language elements
within the compute block. As part of Base SAS, %INCLUDE statements, macro variables, and
macro invocations work the same in compute blocks as they do in all other parts of SAS
procedures.
The following example demonstrates, in a simple way, some of this power. The compute block
performs a transformation of weight from pounds to kilograms. The conversion is done in a
compute block with the same name as the variable that is being modified. The assignment
statement in the compute block is of the same form as you would expect to find in the DATA
step.
title1 'Using Proc REPORT';
title2 'Converting Weight to Kg';
proc report data=rptdata.clinics
(where=(region in('4')))
nowd;
column lname fname sex wt;
define lname / display;
define fname / display;
define sex
/ display format=$6. 'Gender';
define wt
/ display 'Weight in Kg'
format=6.2;
compute wt;
* Convert pounds to KG;
wt = wt/2.2;
endcomp;
run;
52 Carpenter’s Complete Guide to the SAS REPORT Procedure
Using Proc REPORT
Converting Weight to Kg
last name
Halfner
Johnson
Rodgers
Cordoba
Baron
Adams
Rymes
Most
Jackson
Maxim
Perez
East
Batell
Rumor
first
name
John
Randal
Carl
Juan
Roger
Mary
Carol
Mat
Ted
Kurt
Mathew
Clint
Mary
Stacy
Gender
M
M
M
M
M
F
F
M
M
M
M
M
F
F
Weight
in Kg
70.45
91.36
81.36
60.45
72.73
70.45
59.55
70.45
91.36
81.36
60.45
72.73
70.45
59.55
The label for the variable WT
contains the units (pounds),
so a new column header is
also required.
Although you can have a compute block for any report item (variable on the COLUMN
statement), the naming conventions for report items that are addressed within the compute block
are not nearly as straightforward as they would seem to be in this example. This example is
almost the simplest case in which the name of the column is used directly. Naming conventions
used within a compute block are discussed in more detail in Section 7.2.
MORE INFORMATION
The majority of the examples in Chapter 7, “Extending Compute Blocks,” have compute blocks
that use SAS language elements. Section 7.2 covers how columns and report items are addressed
in a compute block under various conditions. Section 7.7 provides additional detail about the SAS
language elements that are allowed in the compute block.
2.7 Sequencing of Step Events
When a PROC REPORT step includes one or more compute blocks, it is important to understand the
sequencing of events when the step is processed. The general process is described in
Section 1.4.2; however, no detail is provided for the interaction of the process with the compute block.
When you have multiple compute blocks, they can appear in any order in your PROC REPORT
step. The order in which you include the blocks does not affect when they are executed. I usually
try to group my compute blocks after my BREAK and RBREAK statements, and I also try to
order them nominally in the same order as they will execute. This arrangement helps me with my
coding, but does not change how the compute blocks work.
As the report is generated (one row at a time—top to bottom and left to right—during the report
row phase), the various compute blocks are executed at the appropriate time. Remember that
compute blocks are tied both vertically and horizontally to locations in the report, and the location
controls the timing of when the compute block will be executed. These ties are specified on the
COMPUTE statement. Vertical ties are established by using the timing options BEFORE and
AFTER, and horizontal ties are made by the specification of report items.
Chapter 2: PROC REPORT: An Introduction 53
When a compute block is tied to a specific report item (such as a variable name) on the COLUMN
statement, it is executed for each report row. The execution takes place when PROC REPORT
processes that specific column, and any given row is processed from left to right based on the
order of the items on the COLUMN statement. In the following step, because SEX is to the left of
WEIGHT on the COLUMN statement, the compute block for SEX is executed before the
compute block for WEIGHT.
proc report .......;
column region sex weight;
.....
compute weight;
.....
endcomp;
compute sex;
.....
endcomp;
.....
run;
Although it would make no difference to the processing of the step, I would reorder the code to
place the compute blocks in the order that they will be processed. I would change the step as
follows:
proc report .......;
column region sex weight;
.....
compute sex;
.....
endcomp;
compute weight;
.....
endcomp;
.....
run;
Compute blocks that are defined with BEFORE or AFTER on the COMPUTE statement are
executed at the time that the specified event takes place. When BEFORE or AFTER appears on
the COMPUTE statement and there is no report item, the compute block is executed only once—
before or after the report.
Including a BEFORE or AFTER on a compute statement is sufficient to generate a row in the
computed summary information during the setup phase. That report row may or may not be
written out to the final report, depending on the BREAK and RBREAK specifications. In either
case, as the report rows are processed during the report row phase, the compute block is executed
as its associated report row is processed.
The COMPUTE BEFORE statement
compute before;
generates what will be the first row in the computed summary information. The COMPUTE
AFTER statement
compute after;
generates the last row in the computed summary information and would be the last compute block
to execute.
54 Carpenter’s Complete Guide to the SAS REPORT Procedure
When you have a report item that is a grouping or ordering variable, you can specify the
COMPUTE statement with either the BEFORE or AFTER options and the report item. This
allows you to execute the compute block when values of the grouping variable change. The
processing opportunities are roughly similar to FIRST. and LAST. processing in the DATA step.
The following COMPUTE statement sets up a compute block that is executed just before each
new value of REGION.
compute before region;
It is not at all unusual to have compute blocks that specify both the BEFORE and AFTER options
in the same PROC REPORT step. Compute blocks that execute before the report or group are
very useful for initializing variables, whereas compute blocks that execute after the report or
group are more generally used for summaries.
Using a compute block with a BEFORE or AFTER option does not require a corresponding
BREAK or RBREAK statement. However, when both are present in the same step, the compute
block and its corresponding BREAK or RBREAK statement share a common row in the
computed summary information. Remember that the compute block itself does not transfer
information to the report (unless it contains a LINE statement). The compute block can be used to
modify information on the computed summary information, and if there is a corresponding
BREAK or RBREAK statement with a SUMMARIZE option, then that information can also be
written to the final report.
MORE INFORMATION
Compute block timing is of special interest and is discussed further in Section 7.1. The BREAK
and RBREAK statements are introduced in Chapter 3, “Creating Breaks.”
SEE ALSO
Chapman (2002) describes the process of events.
2.8 Chapter Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Generate a report that lists total PROFIT for each YEAR. Apply the
DOLLAR. format to the total PROFIT.
2. Building on the solution for Exercise 1:
a.
List the profit for each PRODUCT_LINE as a subgroup of YEAR.
b. Repeat with PRODUCT_LINE as an ACROSS variable.
3. Building on the solution for Exercise 2a, add the following to the report:
Specify text for at least one column.
Use the SPLIT= option.
Use the LINE statement to create a break after each year.
Chapter 2: PROC REPORT: An Introduction 55
Create a user-defined format that redisplays the product line “Sports” to “Sports
Equipment”.
Add a line of text after the report using the LINE statement.
4. The following step contains no typos and all the variables exist; why will it fail?
proc report data=sashelp.class nowd;
column age sex,height;
define age / group;
define sex /across;
define height/display;
run;
5. We would like to create a numeric counter (CNT) for each age group, and then we want to
display the counter along with the age group. We are expecting the following code to result in
a table with one row per age group. What goes wrong and how can it be fixed?
proc sort data=sashelp.class out=cl1;
by age;
run;
data class;
set cl1;
by age;
if first.age then cnt+1;
run;
title 'Count the Age Groups';
proc report data=class nowd;
column cnt age sex,height;
define cnt / order;
define age / group;
define sex /across;
define height/analysis mean;
run;
56 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
3
Creating Breaks
3.1
Generating Breaks Using BREAK and RBREAK 57
3.2
BREAK Statement 59
3.2.1 Skipping a Line between Groups 59
3.2.2 Summarizing across a Group 61
3.2.3 Suppressing the Summarization Label 65
3.2.4 Generating a Page for Each Group Level 67
3.2.5 Combining Summaries with Detail Reports 68
3.3
RBREAK Statement 69
3.3.1 Using RBREAK in a Detail Report 69
3.3.2 Using RBREAK with BREAK in a Detail Report 70
3.3.3 Using RBREAK and BREAK in a Summary Report 71
3.4
Chapter Exercises 73
3.1 Generating Breaks Using BREAK and RBREAK
It is often advantageous to provide additional information at specific locations within the report.
Compute blocks (Section 2.6) can be used in some circumstances. However, when you want to
summarize across the entire report or across groups within a report, the BREAK and RBREAK
statements can be especially useful. The syntax of these statements is as follows:
break location break-variable;
rbreak location ;
58 Carpenter’s Complete Guide to the SAS REPORT Procedure
location
The BREAK and RBREAK statements are used to specify events, and the location of
each event is indicated with either BEFORE or AFTER.
break-variable
This report item indicates for which group the summary will take place. It generally has a
usage of GROUP.
options
Several appearance options can appear on these two statements as well. Some of these
include the following:
PAGE
starts a new page after the last break line summary.
SUMMARIZE
calculates an across group or across report. Summary lines are
written to the report only when this option is present.
The following options are available only on the BREAK statement:
SKIP
SUPPRESS
skips a line after the break (LISTING destination).
suppresses the printing of the value of the grouping variable and, in the
LISTING destination, suppresses any overlining and underlining around the
grouping variable as well.
When a summary line is requested (through the SUMMARIZE option), numeric variables with
the define type of ANALYSIS are summarized. This is important because your report might
include numeric variables such as DATE or AGE with a usage of DISPLAY that are not to be
summarized.
You can specify multiple BREAK and RBREAK statements for a given PROC REPORT step,
and the order of the statements does not change the appearance of the report or the timing of the
execution of any compute blocks.
When you create a break with either the BREAK or RBREAK statement, you must specify when
the break is to take place. The break is event-specific, and you specify the event location as being
before or after the event by using either BEFORE or AFTER as the target location. The event can
be the start or end of the entire report (RBREAK) or simply when a grouping variable changes
values (BREAK).
The RBREAK statement summarizes across the entire report, and therefore its result appears at
either the top of the report (BEFORE) or at the end of the report (AFTER). On the BREAK
statement, a variable, usually a grouping variable, is specified in addition to the location
specification, and the summaries appear before or after the groups of this variable.
Chapter 3: Creating Breaks 59
A BREAK or RBREAK statement can generate more than one row in the table, depending on the
options used with the statement. Here is the order in which these options are applied when they
are present:
Order
Actions
Options
Book Sections
1
Overlining
OL and DOL
4.2 (LISTING only)
2
Summarization
SUMMARIZE
3.3.2
3
Underlining
UL and DUL
4.2 (LISTING only)
4
Skipping lines
SKIP
3.2.1 (LISTING only)
5
New page
PAGE
3.2.4
MORE INFORMATION
The BEFORE and AFTER locations are also used with compute blocks and are first discussed in
Section 2.6.
A number of additional BREAK and RBREAK statement options that are only used with the ODS
LISTING destination are discussed in Chapter 4, “Only in the LISTING Destination.” For reports
explicitly controlled by ODS, the STYLE= option can be used on these statements to override the
ODS style characteristics (See Sections 8.1 and 8.2).
SEE ALSO
SAS Technical Report P-258 (1993, p. 219) discusses the order of break line events. Burlew
(2005, pp. 26–41) includes several examples that use the BREAK and RBREAK statements.
3.2 BREAK Statement
The BREAK statement is used to create breaks and summaries within the report, either before or
after specific changes in some grouping variable. That grouping variable appears on the BREAK
statement and must be specified as such on the DEFINE statement.
The PROC REPORT step rolls up or summarizes on the specified groups. However, if the
COLUMN statement also contains a variable with a define type of DISPLAY, the summarizations
do not take place and a detail report will result. This is not always bad, because although the rollups are not displayed, the BREAK statement can still be used (see Section 3.3.2).
3.2.1 Skipping a Line between Groups
In the LISTING destination, one of the simplest actions of the BREAK statement is for it to insert
a blank line between groups. In the following example, the SKIP option is used with a location of
AFTER so that the break (and therefore the blank line) follows each group (REGION).
60 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Using BREAK to SKIP a Space';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region/skip;
run;
Creating Breaks in the Report
Using BREAK to SKIP a Space
weight in pounds
n
mean
4
195
region
1
Gender
M
2
F
M
6
4
109.66667
105
3
F
M
5
5
127.8
163.8
4
F
4
143
The BREAK statement has
inserted a blank line after each
summary group, but it has not
itself created any summary output.
Group summaries are created
using the SUMMARIZE option,
which is discussed in Section 3.2.2.
For ODS destinations other than LISTING, the SKIP option is ignored. Instead, these destinations
emphasize the summary line by presenting it in a different style (Data Emphasis is the default
emphasis style). The same report generated through the PDF destination does not show the blank
lines, because PDF ignores the SKIP option.
The PDF destination does not
support the SKIP option;
consequently, other techniques are
needed if you want to separate
groups with a blank line.
Destination: PDF Style: PRINTER
Chapter 3: Creating Breaks 61
MORE INFORMATION
For destinations that do not support the SKIP option, you can also skip or insert blank lines
through the use of the compute block (see Section 2.6). More options and techniques that apply
only to the LISTING destination are discussed in Chapter 4, and the SKIP option specifically in
Section 4.2.
3.2.2
Summarizing across a Group
Usually we would like to summarize across groups. This is one of the primary purposes of the
BREAK statement, and it is accomplished using the SUMMARIZE statement option.
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Summarizing With BREAK';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / summarize;
run;
Creating Breaks in the Report
Summarizing With BREAK
region
1
1
2
2
3
3
4
4
Gender
M
F
M
F
M
F
M
weight in pounds
n
mean
4
195
4
195
6 109.66667
4
105
10
107.8
5
127.8
5
163.8
10
145.8
4
143
10
165.6
14 159.14286
For destinations other than LISTING, the Data
Emphasis style attribute is used on the summary
lines. In the RTF table to the right, the summary
lines appear in italics.
Destination: RTF Style: RTF
62 Carpenter’s Complete Guide to the SAS REPORT Procedure
Notice that by itself the SUMMARIZE option does not create any visual separation between the
summary and the next group. When the SUMMARIZE and SKIP options are used in conjunction
with each other in the LISTING destination, the report becomes much easier to read.
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Summarizing and Skipping With BREAK';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / summarize skip;
run;
Creating Breaks in the Report
Summarizing and Skipping With BREAK
region
1
1
Gender
M
2
F
M
6
4
109.66667
105
10
107.8
F
M
5
5
10
127.8
163.8
145.8
F
M
4
10
143
165.6
14
159.14286
2
3
3
4
4
weight in pounds
n
mean
4
195
4
195
Z
n
Y
The summary line n appears after
the group. The SKIP option creates
a blank line Y. By default, the
SUMMARIZE option restates the
current value of the grouping
variable Z. The value of the group
variable can be suppressed (see
Section 3.2.3).
The order of the BREAK statement options does not make a difference, and for that matter one
could issue two separate BREAK statements, each with one of the options. The BREAK
statement used in this example
break after region / summarize skip;
could be rewritten as
break after region / summarize;
break after region / skip;
Chapter 3: Creating Breaks 63
Essentially the same report could be generated by using the SKIP option before the summary.
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Summarizing and Skipping With BREAK';
title3 'Using Two BREAK Statements';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break before region / skip; q
break after region /summarize; r
run;
Creating Breaks in the Report
Summarizing and Skipping With BREAK
Using Two BREAK Statements
region
Gender
weight in pounds
n
mean
q
1
qThe blank line for each group
M
1
2
4
4
195
6 109.66667
4
105
10
107.8
F
M
5
5
10
F
M
4
143
10
165.6
14 159.14286
3
4
195
F
M
2
3
4
r
now appears before the group. The
summary row r still appears after
the group.
127.8
163.8
145.8
The group summary line can also be placed before the group.
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Using BEFORE With BREAK';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / skip; s
break before region / summarize; t
run;
64 Carpenter’s Complete Guide to the SAS REPORT Procedure
Creating Breaks in the Report
Using BEFORE With BREAK
region
1
1
s
2
2
3
3
4
4
Gender
M
weight in pounds
n
mean
4
195
4
195
F
10
6
107.8
109.66667
M
4
105
F
M
10
5
5
145.8
127.8
163.8
F
M
14
4
10
159.14286
143
165.6
t
As was mentioned earlier, the SKIP option is for use in the LISTING destination and is ignored
when applied to other ODS destinations. This is demonstrated in the following example, where
the first example from this section is repeated and routed to the HTML destination _.
ods html file="&path/results/ch3_2_2d.html";
_
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Summarizing With BREAK';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / summarize skip;`
run;
ods html close;
An examination of the HTML report (CH3_2_2D.HTML) shows that not only has the SKIP
option ` not been used, but the summary line has a different appearance.
Chapter 3: Creating Breaks 65
The SKIP option has been ignored and the
summary lines have been written in italics.
Italics is one of the appearance
characteristics of the Data Emphasis style
attribute.
Destination: HTML Style: DEFAULT
MORE INFORMATION
The value of the group variable is repeated for the summary line. This repeated label can be
suppressed using the SUPPRESS option (see Section 3.2.3).
SEE ALSO
SAS Technical Report p-258 (1993, p. 136) introduces the SUMMARIZE option.
3.2.3
Suppressing the Summarization Label
When you use the SUMMARIZE option to request the printing of a summary, the value of the
group variable is repeated for the summary line. Although this can be especially useful for groups
with a large number of values, often we want to remove this repeated value. In the reports
generated in Section 3.2.2, these repeated values are distracting. They can be eliminated through
the use of the SUPPRESS option n.
66 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Creating Breaks;
title1 'Creating Breaks in the Report';
title2 'Suppressing the Group Value';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / skip summarize suppress; n
run;
Creating Breaks in the Report
Suppressing the Group Value
region
1
Gender
M
n
weight in pounds
n
mean
4
195
4
195
F
M
6
4
10
109.66667
105
107.8
3
F
M
5
5
10
127.8
163.8
145.8
4
F
M
4
10
14
143
165.6
159.14286
2
n The group value for the
summary line has been
suppressed and is no longer
shown.
For the LISTING destination shown here, the summary line is still not as distinct from the other
values as we might wish it to be. There are several underlining options (see Section 4.4) that can
be used to help with this issue in the LISTING destination. Most other ODS destinations, even
when used with the simpler styles, attempt to distinguish the summary group automatically. The
following example shows the same report generated using the HTML destination and the
MINIMAL style.
ods html file="&path\Results\ch3_2_3b.html"
style=minimal;
Chapter 3: Creating Breaks 67
Destination: HTML Style: MINIMAL
3.2.4
Generating a Page for Each Group Level
When the PAGE option is used on the BREAK statement, a new page is generated for each level
of the grouping variable. Here is the example from Section 3.2.3 with the PAGE option on the
BREAK statement.
title1 'Creating Breaks in the Report';
title2 'One PAGE per Group';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
break after region / skip summarize suppress page;
run;
The same report is generated; however, this time each region appears on its own page. (Only the
page for REGION=3 is shown here.)
68 Carpenter’s Complete Guide to the SAS REPORT Procedure
Creating Breaks in the Report
One PAGE per Group
region
3
Gender
F
M
weight in pounds
n
mean
5
127.8
5
163.8
10
145.8
SEE ALSO
The PAGE option is discussed in SAS Technical Report P-258 (1993, pp. 93–95, 189).
Humphreys (2006) uses the PAGE option on the BREAK statement.
3.2.5
Combining Summaries with Detail Reports
Sometimes you will need to create a report that contains both detail-level and summary-level
information. You do this by taking advantage of the fact that PROC REPORT cannot form groups
if there is also a display variable.
In the following example, REGION has a define type of GROUP; however, because LNAME is a
display variable, the regions are not collapsed. Instead a BREAK statement has been added to
form summaries for REGION.
title1 'Creating Breaks in the Report';
title2 'Detail Report With BREAK';
proc report data=rptdata.clinics
(where=(region in('2' '3')))
nowd;
column region lname wt;
define region / group width=6;
define lname / display;
define wt
/ mean;
break after region / summarize skip;
run;
Creating Breaks in the Report
Detail Report With BREAK
region
2
last name
Maxwell
Little
Long
Harbor
Harbor
Haddock
Saunders
Ingram
Leader
Atwood
weight
in pounds
105
109
115
105
105
105
109
115
105
105
The report contains
individual weights as well as
the average weight for the
region.
This mixture of report types
is generated by using both the
GROUP and DISPLAY
define types.
(continued)
Chapter 3: Creating Breaks 69
(continued)
2
3
Smith
Jones
Marksman
Candle
Masters
Henry
Stubs
Upston
Panda
Herbal
3
107.8
162
105
112
195
155
162
105
112
195
155
145.8
We get the best of both worlds!
3.3 RBREAK Statement
When you want to summarize across the entire report, you use the RBREAK statement. The
syntax and options for this statement are similar to those of the BREAK statement, except, of
course, that there is no grouping variable.
3.3.1
Using RBREAK in a Detail Report
The RBREAK statement enables us to summarize across the entire report. The following PROC
REPORT step places this summary before the report.
ods html file="&path\results\ch3_3_1.html"
style=statdoc;
title1 'Creating Breaks in the Report';
title2 'Summarizing After the Detail REPORT';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column lname fname sex dob wt,(n mean) n;
define lname / order;
define fname / order;
define sex
/ group format=$6. 'Gender';
define dob
/ display 'Birthday'; o
define wt
/ analysis format=5.1;
rbreak before / summarize; p
run;
ods html close;
70 Carpenter’s Complete Guide to the SAS REPORT Procedure
n Statistics have been requested for a detail report. Because the mean of a single observation is
that individual value, the mean only comes into play on the summary line.
o Although DOB is numeric, it has a define type of DISPLAY and is not summarized. This is
important, because the average birthday would have little meaning.
The summary line shows the total number of patients and the average weight at the top
(BEFORE) of the report.
Destination: HTML Style: STATDOC
Notice that the summary line has chosen the correct type of summary for each statistic (SUM for
N and MEAN for the mean WT).
3.3.2
Using RBREAK with BREAK in a Detail Report
When the report contains one or more grouping variables, you can use a combination of the
BREAK and RBREAK statements to create summaries for both the individual groups as well as
across the entire report.
Because the following COLUMN statement contains both grouping variables (REGION and
SEX) as well as nongrouping variables (LNAME and FNAME), the report is not rolled up to the
group level. Instead a detailed report is generated. However, because there is a grouping variable,
the BREAK statement can now be used as well.
ods html file="&path\results\ch3_3_2.html"
style=brick;
HTML using STYLE=statdoctitle1 'Creating Breaks in the Report';
title2 'Summarizing Groups in a Detail REPORT';
proc report data=rptdata.clinics
Chapter 3: Creating Breaks 71
(where=(region in('1' '2' '3' '4')))
nowd;
column region lname fname sex wt,(n mean);
define region / group 'Region' format=$6.; n
define lname / order; o
define fname / order; o
define sex
/ group format=$6. 'Gender'; n
define wt
/ analysis format=5.1;
break after region / summarize suppress; p
rbreak before / summarize;
run;
ods html close;
n Grouping variables are defined. However, because the presence of nongrouping variables o
prevents the summarization of observations into individual groups, the report is still at the
detail level.
p Although individual detail rows are not collapsed, the BREAK statement can be used, and it
generates group summaries for each region.
Destination: HTML Style: BRICK
This combination of BREAK and RBREAK statements along with a detail report allows us to
generate a report with group summaries, even though the individual rows of the report itself are at
the detail level.
3.3.3
Using RBREAK and BREAK in a Summary Report
The BREAK and RBREAK statements are commonly used together in reports that summarize
across observations. When grouping variables are used without the presence of nongrouping
variables, the individual detail rows are collapsed into summaries.
72 Carpenter’s Complete Guide to the SAS REPORT Procedure
In the following step there are two grouping variables (REGION and SEX), and a summary line has
been requested through the use of the BREAK statement after each group formed by REGION.
title1 'Creating Breaks in the Report';
title2 'Summarizing After the REPORT';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd;
column region sex wt,(n mean);
define region / group format=$6.;
define sex
/ group format=$6. 'Gender';
define wt
/ analysis;
define mean
/ format=5.1; n
break after region / skip summarize suppress; o
rbreak after / summarize; p
run;
n A format is attached to the column containing the mean values.
o The BREAK statement generates a summary line after each region. The SKIP option only
has meaning in the LISTING destination, which has been used here.
p The summary across all observations is generated by the RBREAK statement.
Creating Breaks in the Report
Summarizing After the REPORT
weight in pounds
region
1
Gender
M
n
4
4
mean
195.0
195.0
2
F
M
6
4
109.7
105.0
10
107.8
3
F
M
5
5
10
127.8
163.8
145.8
4
F
M
4
10
14
143.0
165.6
159.1
38
145.9
n
o
p
The order of the BREAK o
and RBREAK p statements
is not important. The SKIP
option on the BREAK
statement will work only for
the LISTING destination.
Chapter 3: Creating Breaks 73
3.4 Chapter Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Create a report that shows total PROFIT for each PRODUCT_LINE
within each year. You can build on the results of Exercise 3 in Chapter 2.
Using the BREAK and RBREAK statements:
Additionally summarize across product lines and across years.
Experiment with the SUPPRESS, SUMMARIZE, and SKIP options.
2. The data table SASHELP.RETAIL contains quarterly sales information. For each YEAR use
the quarterly sales to calculate the following sales statistics:
number of quarters (why might we need this statistic, when there are always four quarters
in a year?)
mean quarterly sales
standard deviation of the quarterly sales
74 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
4
Only in the LISTING Destination
4.1 Using the HEADLINE and HEADSKIP Options 76
4.2 Blank Lines, Overlines, and Underlines 78
4.3 Repeat Characters 79
4.3.1 Adding Repeated Characters to Spanning Headers 80
4.3.2 Repeat Characters with the SPLIT= Option 82
4.4 PROC REPORT Statement Options 83
4.4.1 Creating Boxes on the Report 83
4.4.2 Controlling the Centering of the Report 85
4.4.3 Adjusting the Width of Numeric and Computed Columns 85
4.4.4 Creating Multiple Panels on a Page 86
4.4.5 Using the PSPACE= Option 87
4.4.6 Controlling the Size of the Page 88
4.4.7 Using the FORMCHAR Option 89
4.4.8 Wrapping Data Lines 91
4.5 Other DEFINE Statement Options 92
4.5.1 Specifying the Column Width 93
4.5.2 Using the FLOW Option to Wrap Text 93
4.5.3 Adding Spaces between Columns 94
4.6 Chapter Exercises 96
76 Carpenter’s Complete Guide to the SAS REPORT Procedure
Of all the ODS destinations, LISTING is unique. Not only was this the first (and effectively only)
destination before the development of the Output Delivery System; it was also, and still is, a textonly destination. This means that ODS features such as styles, fonts, and colors do not apply.
Because of these limitations of the LISTING destination, many users now generate their reports
using other destinations. However, the LISTING destination is still useful.
To help provide some control over the appearance of reports that use the LISTING destination, a
number of options and techniques are available for various statements within the PROC REPORT
step. These options, which can be used only with the LISTING destination, are discussed in this
chapter.
MORE INFORMATION
It is often useful to simulate, in other destinations, some of the exclusive LISTING options
described in this chapter. Fortunately, alternatives enable you to do this. These are discussed
throughout the book and summarized in Section 10.4.
SEE ALSO
A number of the options and techniques demonstrated in this chapter are also discussed in
examples by Young (2003) and Flavin (1996). For examples of how exclusive LISTING options
can be simulated in other destinations, see
http://support.sas.com/rnd/base/topics/templateFAQ/repoption.html
Burlew (2005, p. 190) lists options that are used strictly with the LISTING destination.
4.1 Using the HEADLINE and HEADSKIP Options
The HEADLINE and HEADSKIP options provide a separation between the header portion of the
report and the report itself. These options can be used separately or in conjunction with each
other. Used in the PROC REPORT statement, they result in the addition of an underline and a
space below the column header text.
* Using Headline and Headskip;
title1 'Only in LISTING';
title2 'With HEADLINE and HEADSKIP';
proc report data=rptdata.clinics
nowd headline headskip;
column region sex wt;
define region / group;
define sex
/ across;
define wt
/ analysis mean format=6.2;
run;
In the report, the underline n and the space o that separates the underline from the body of the
table are formed by the HEADLINE and HEADSKIP options.
Chapter 4: Only in the LISTING Destination 77
Only in LISTING
With HEADLINE and HEADSKIP
re
weight
gi pati
in
on F M pounds
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
1
10
2
3
4
5
6
7
8
9
.
2
6
5
4
5
4
.
4
2
4
4
4
5
*
3
6
4
.
8
n
o
195.00
172.33
107.80
145.80
159.14
157.75
198.00
151.00
160.00
187.80
The HEADLINE option generates
an underline n below the header,
and HEADSKIP adds a blank row
o.
The asterisk that appears instead
of the count of males in region 4
indicates that the count was
greater than a single digit (see
Section 2.2.4).
In the previous table, the default font, SAS Monospace, has been used, and the line has been
formed using a special character that belongs to that specific font. Converting the font from SAS
Monospace to some other font, such as Courier New, as has been done in the following table,
causes the characters in the line to change into characters that no longer form a line.
Only in LISTING
With HEADLINE and HEADSKIP
re
weight
gi pati
in
on F M pounds
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
1
10
2
3
4
5
6
7
8
9
.
2
6
5
4
5
4
.
4
2
4
4
4
5
*
3
6
4
.
8
195.00
172.33
107.80
145.80
159.14
157.75
198.00
151.00
160.00
187.80
The underline is formed by a character unique to
the font SAS Monospace, and does not convert
well to other fonts.
One of the limitations of the use of the HEADLINE option is this lack of portability.
MORE INFORMATION
If you need to change the font, but also need to show lines correctly, you can specify the character
used to create the line by using the FORMCHAR option (see Section 4.4.7). The example in
Section 7.8.2 simulates the HEADLINE and HEADSKIP options by using LINE statements.
78 Carpenter’s Complete Guide to the SAS REPORT Procedure
4.2 Blank Lines, Overlines, and Underlines
In the BREAK and RBREAK statements, a number of options are available to produce blank
lines, overlines, and underlines. These include the following:
OL
DOL
UL
DUL
SKIP
inserts an overline.
inserts a double overline.
inserts an underline.
inserts a double underline.
skips a line after the break.
In the following example, a double overline (DOL) is inserted before the report summary
generated by the RBREAK statement.
* Options with BREAK and RBREAK;
title1 'Only in LISTING';
title2 'RBREAK with Double Overline (DOL)';
proc report data=rptdata.clinics nowd;
column region sex wt;
define region / group;
define sex
/ across;
define wt
/ analysis mean format=6.2;
rbreak after / summarize dol;
run;
Only in LISTING
RBREAK with Double Overline (DOL)
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
= =
* *
weight
in
pounds
195.00
172.33
107.80
145.80
159.14
157.75
198.00
151.00
160.00
187.80
======
161.78
The LISTING destination does not automatically
scale the width of a column to accommodate a
value. Consequently, where the patient count
exceeds 9 (the sums across regions, as well as for
males within region 4), an asterisk (*) is
displayed instead of the value. This can be
corrected by using a format (see Section 2.4.2) or
by specifying a column width (see Section 4.4).
Unlike the example in Section 4.1, this example forms the double overline and double underline
with the equal sign (=). Consequently, if you want to convert the font to something other than
SAS Monospace, it should convert correctly. The previous table has been displayed in Courier.
Chapter 4: Only in the LISTING Destination 79
MORE INFORMATION
The SKIP option in the BREAK statement is introduced and discussed in Section 3.1.
SEE ALSO
The OL, DOL, UL, and DUL options are discussed in SAS Technical Report P-258 (1993, pp.
138-141).
4.3 Repeat Characters
When you place text as a header, it can be helpful to clarify the specific columns to which that
header is to apply. You can do this easily if the text fully spans the header space. However, you
usually don’t know exactly how much space is needed, and filling the space manually is awkward.
Fortunately, you can use repeat characters to fill any extra space automatically. To take advantage
of repeat characters, use one of the following characters as the first and last characters in the text
string.
=
_
.
*
+
(Hyphen, or minus sign)
(Equal)
(Underscore)
(Period)
(Asterisk)
(Plus)
®
Some of the documentation also incorrectly (at least for SAS 9) indicates that the following can
also be used as repeat characters.
:
\
(Colon)
(Backslash) If used, the \ is stripped off.
The greater than (>) and less than (<) characters can also be used in pairs as spanning headers, and
they can be used in either order:
< >
> <
For destinations other than LISTING, repeat characters are generally ignored or, in SAS 9.2 and
later, removed. The repeat character pairs < > and > < are not removed in other destinations, as
they can also be used with HTML tags (see Section 8.5.1).
SEE ALSO
SAS Technical Report P-258 (1993, pp. 66–69) has several examples that use repeated characters.
80 Carpenter’s Complete Guide to the SAS REPORT Procedure
4.3.1 Adding Repeated Characters to Spanning Headers
Spanning headers are text strings that span multiple columns. These headers are designated as
quoted strings in the COLUMN statement. You can use repeated characters to fill the available
space on spanning headers by making one of the special characters both the first and last character
in that string. In the following example, hyphens are used to provide extended fill lines before and
after the text.
* Repeated Text headers in the COLUMN statement;
title1 'Only in LISTING';
title2 'Repeated Text';
proc report data=rptdata.clinics nowd;
column region sex ('-wt(lb)-' wt,(n mean));
define region / group;
define sex
/ across;
define wt
/ analysis;
run;
In the following table, notice that the characters immediately after and before the quotes (in this
example the characters are hyphens or minus signs) are repeated as many times as necessary to fill
the full space allocated to the columns.
Only in LISTING
Repeated Text
re
gi
on
1
10
2
3
4
5
6
7
8
9
pati
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
ƒƒƒƒƒƒƒwt(lb)ƒƒƒƒƒƒƒ
weight in pounds
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
This table actually has two spanning
headers. We defined the first one in the
COLUMN statement. The second,
which is the label of the analysis
variable (weight in pounds), is produced
automatically to span the columns
containing the nested statistics.
As was discussed in Section 4.1, the repeated hyphen is automatically converted to a repeated
character that forms a line, but this character does not translate well into fonts other than SAS
Monospace. Fortunately, the other repeat characters translate much better when you change fonts.
The following example uses the equal sign (=) as the repeat character.
Repeat characters are not restricted to the COLUMN statement. You can also use the repeat
character syntax in a DEFINE statement, where they have the advantage of replacing the
variable’s label.
* Repeated Text headers in the DEFINE statement;
title1 'Only in LISTING';
title2 'Repeated Text in the DEFINE Statement';
proc report data=rptdata.clinics nowd;
column region sex ('Patient Weight' wt,(n mean));
define region / group;
define sex
/ across 'Sex';
Chapter 4: Only in the LISTING Destination 81
define wt
run;
/ analysis '=(lb)=';
Only in LISTING
Repeated Text in the DEFINE Statement
re
gi
on
1
10
2
3
4
5
6
7
8
9
Sex
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
Patient Weight
========(lb)========
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
Here the text on the DEFINE statement
becomes a spanning header as it
replaces the label for an analysis
variable with more than one statistic.
You can also use < and > together and have them repeat independently. The following example
replaces the equal signs in the DEFINE statement with < and >.
* Repeated Text headers in the DEFINE statement;
title1 'Only in LISTING';
title2 'Repeated < and >Text in the DEFINE Statement';
proc report data=rptdata.clinics nowd;
column region sex ('Patient Weight' wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '<(lb)>';
run;
Only in LISTING
Repeated < and >Text in the DEFINE Statement
re
gi
on
1
10
2
3
4
5
6
7
8
9
Sex
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
Patient Weight
<<<<<<<<(lb)>>>>>>>>
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
MORE INFORMATION
Repeat characters are established as part of a format specification in the second example in
Section 6.4.1.
82 Carpenter’s Complete Guide to the SAS REPORT Procedure
SEE ALSO
Burlew (2005, pp. 31–34) has an example that shows spanning headers with repeat characters.
4.3.2 Repeat Characters with the SPLIT= Option
As was shown in Section 2.4.3, a split character can be designated to indicate the location that text
should be broken across lines of output. This option can be especially useful if you want to create
repeated text within a spanning header.
* Splitting text headers in the DEFINE statement;
title1 'Only in LISTING';
title2 'Repeated Text and the SPLIT Option';
proc report data=rptdata.clinics
nowd split='*'; n
column region sex wt,(n mean);
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis 'Patient Weight*=(lb)=';
run;
o
n The SPLIT= option designates the character that is used to determine the line split
location.
o In this text string, the * is used to break the line.
Only in LISTING
Repeated Text and the SPLIT Option
re
gi
on
1
10
2
3
4
5
6
7
8
9
Sex
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
Patient Weight
========(lb)========
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
This technique allows you to create
multiple line-spanning headers from
the DEFINE statement alone. Header
text has been removed from the
COLUMN statement.
Chapter 4: Only in the LISTING Destination 83
You can also create lines of text using only repeat characters.
* Splitting text headers in the DEFINE statement;
title1 'Only in LISTING';
title2 'Text Line with ONLY Repeated Text';
proc report data=rptdata.clinics nowd split='*';
column region sex ('Patient Weight'wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '(lb)*__';
run;
In this example, the repeat character is an underscore ( _ ). When forming lines, this character
translates well to other fonts, although it does not appear as a solid line in all fonts. The following
table is displayed using Courier.
Only in LISTING
Text Line with ONLY Repeated Text
re
gi
on
1
10
2
3
4
5
6
7
8
9
Sex
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
Patient Weight
(lb)
____________________
n
mean
4
195
6 172.33333
10
107.8
10
145.8
14 159.14286
8
157.75
10
198
4
151
4
160
10
187.8
The header for the analysis variable in
this example uses three elements: the
SPLIT= option on the PROC statement,
the text on the COLUMN statement,
and the text on the DEFINE statement.
The two text strings could have been
combined.
MORE INFORMATION
See Section 8.6.5 for a discussion of in-line formatting sequences that can be used to force splits
in character strings in some of the other ODS destinations.
4.4 PROC REPORT Statement Options
The PROC REPORT statement supports a fairly wide variety of options. Some of these have to do
with the layout and formatting of the report, and some of these apply only to tables that are
directed to the LISTING destination. The options discussed in the following sections fall into the
latter category.
4.4.1 Creating Boxes on the Report
The BOX option can be used to add boxes that surround the various portions of the text of the
report. Although the vertical lines and the juncture points work well in the Output window of the
Display Manager, these same text characters do not usually translate well into word processing
84 Carpenter’s Complete Guide to the SAS REPORT Procedure
documents (see Section 4.1). This is often true even when the same font (SAS Monospace) is used
in the document.
The following example uses the BOX option to build boxes.
* Using the BOX option;
title1 'Only in LISTING';
title2 'Using the BOX Option';
proc report data=rptdata.clinics
nofs split='*'
box;
column region sex ('Patient Weight'wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '(lb)*__';
run;
The resulting table is displayed here using the SAS Monospace font. Even though this is the target
font, the resulting table is ugly at best when it is pasted into a word processing document. This problem
alone should be a sufficient reason for the programmer to consider alternate ODS destinations.
Only in LISTING
Using the BOX Option
„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚
Patient Weight
‚
‚re
(lb)
‚
‚gi Sex
____________________‚
‚on F M
n
mean‚
‡ƒƒ…ƒƒ…ƒƒ…ƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒ‰
‚1 ‚ .‚ 4‚
4‚
195‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚10‚ 2‚ 4‚
6‚ 172.33333‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚2 ‚ 6‚ 4‚
10‚
107.8‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚3 ‚ 5‚ 5‚
10‚
145.8‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚4 ‚ 4‚ *‚
14‚ 159.14286‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚5 ‚ 5‚ 3‚
8‚
157.75‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚6 ‚ 4‚ 6‚
10‚
198‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚7 ‚ .‚ 4‚
4‚
151‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚8 ‚ 4‚ .‚
4‚
160‚
‡ƒƒ ƒƒ ƒƒ ƒƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒƒ‰
‚9 ‚ 2‚ 8‚
10‚
187.8‚
Šƒƒ‹ƒƒ‹ƒƒ‹ƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒŒ
You can use the FORMCHAR option (see Section 4.4.7) to change the characters used to form
any segment of the boxes, including the interior corners that are displayed here as small squares.
Chapter 4: Only in the LISTING Destination 85
4.4.2 Controlling the Centering of the Report
The CENTER/NOCENTER option can be used on the PROC REPORT statement to override (but
not change) the CENTER system option. This option enables the PROC REPORT step to generate
a report that does not follow the value stored in the system option.
In all of the examples in this chapter, the system option has been set to NOCENTER, as shown here:
options nocenter;
This option results in all the titles and tables being left-justified. We could have left the system
option at its default value, centered, but instead changed the PROC REPORT option to
NOCENTER. The code would have looked something like this:
* This option statement sets the CENTER/NOCENTER
* option to its default value;
options center;
proc report data=rptdata.clinics nocenter
. . . Portions of the code are not shown . . .
Any subsequent steps, say a PROC PRINT or even another PROC REPORT, would be centered
by the system option.
4.4.3 Adjusting the Width of Numeric and Computed Columns
As can be seen in the tables for the examples in Sections such as 4.3.2, PROC REPORT does not
always pick a reasonable default width for the numeric values. The COLWIDTH= option sets a
default width for numeric and computed columns.
* Using the COLWIDTH option;
title1 'Only in LISTING';
title2 'Using the COLWIDTH Option';
proc report data=rptdata.clinics nowd
split='*' colwidth=4;
column region sex ('Patient Weight'wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '(lb)*__';
run;
Only in LISTING
Using the COLWIDTH Option
re
gi
on
1
10
2
3
4
5
6
7
8
9
Sex
F M
. 4
2 4
6 4
5 5
4 *
5 3
4 6
. 4
4 .
2 8
Patient Weight
(lb)
__________
n mean
4
195
6
172
10
108
10
146
14
159
8
158
10
198
4
151
4
160
10
188
Because COLWIDTH is applied only to
numeric and computed columns, the
patient counts under SEX are still at a
width of one space.
86 Carpenter’s Complete Guide to the SAS REPORT Procedure
The resulting table is much more compact, but has fewer significant digits for the mean.
MORE INFORMATION
This new default width can be overridden through the use of the WIDTH= option (see
Section 4.5.1) on the DEFINE statement. The FORMAT= option can also be used to specify
column width (see Section 2.4.2).
4.4.4 Creating Multiple Panels on a Page
When your report is narrow (few columns), you can sometimes save space and perhaps create a
more pleasing report by placing multiple panels on the page. This is the approach used in the
phone book. The PANELS= option is used to specify the maximum number of panels that you
would like to allow. When one set of columns has been written, instead of going to the next page,
PROC REPORT checks to see if the next set of columns will fit in the next panel.
In the following example, the report has only three columns, which creates a long narrow report.
With the PANELS= option set to 2, the report can contain twice the information on each page. In
this example, the final page contains 6 columns.
* Creating Panels;
title1 'Only in LISTING';
title2 'Using the PANELS Option';
proc report data=rptdata.clinics nowd
split='*' colwidth=3
panels=2
;
column lname region wt;
define lname / order;
define region / display 'region';
define wt
/ analysis ' ';
run;
Only in LISTING
Using the PANELS Option
last name
Adams
Adamson
Alexander
Antler
Atwood
Banner
Baron
Batell
re
gi
on
4
8
6
6
2
6
4
4
155
158
175
240
105
175
160
155
last name
Masters
Maxim
Maxwell
Mercy
Moon
Most
Nabers
Nolan
re
gi
on
3
4
2
1
5
4
1
6
155
179
105
195
160
155
195
187
. . .Portions of the table are not shown. . .
Chapter 4: Only in the LISTING Destination 87
MORE INFORMATION
The ODS COLUMNS= option, which is discussed in Section 8.9, can be used to produce a similar
effect in the RTF and PDF destinations.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 184, 229) discusses the PANELS= option.
4.4.5 Using the PSPACE= Option
When you specify multiple panels you can specify the space between the panels with the
PSPACE= option. In the example in Section 4.4.4, there are 6 spaces between the patient weight
and the last name in the second panel. You might want to emphasize the two panels by increasing
the distance between them by a couple of spaces. The following example repeats that of Section
4.4.4; however, the PSPACE= option is set to 8 (increased from the default by 2).
* Expanding the space between panels;
title1 'Only in LISTING';
title2 'Using the PSPACE Option';
proc report data=rptdata.clinics nowd
split='*' colwidth=3
panels=2 pspace=8
;
column lname region wt;
define lname / order;
define region / display 'region';
define wt
/ analysis ' ';
run;
Only in LISTING
Using the PSPACE Option
re
gi
last name on
Adams
4
Adamson
8
Alexander 6
Antler
6
Atwood
2
Banner
6
Baron
4
Batell
4
155
158
175
240
105
175
160
155
last name
Masters
Maxim
Maxwell
Mercy
Moon
Most
Nabers
Nolan
re
gi
on
3
4
2
1
5
4
1
6
155
179
1
195
160
155
195
187
. . .Portions of the table are not shown. . .
SEE ALSO
Burlew (2005, p. 176) has an example that uses the PSPACE= option.
The space between the
first three columns and
the second three has
been increased.
88 Carpenter’s Complete Guide to the SAS REPORT Procedure
4.4.6 Controlling the Size of the Page
The LINESIZE= (LS=) and PAGESIZE= (PS=) system options can be overridden through the use
of the LS= and PS= options on the PROC REPORT statement. Much like the
CENTER/NOCENTER option discussed in Section 4.4.2, these options can override the
corresponding system options, but they will not change the system option settings themselves.
The following example uses these options to specify a page size of 80 columns and 30 rows. With
the reduced number of rows, the data now fills all three panels specified in the PANELS= option.
* Changing the size of the page;
title1 'Only in LISTING';
title2 'Using the PS and LS Options';
proc report data=rptdata.clinics nowd
split='*' colwidth=3
panels=3
ls=80 ps=30
;
column lname region wt;
define lname / order;
define region / display 'region';
define wt
/ analysis ' ';
run;
Only in LISTING
Using the PS and LS Options
last name
Adams
Adamson
Alexander
Antler
Atwood
Banner
Baron
re
gi
on
4
8
6
6
2
6
4
155
158
175
240
105
175
160
last name
Harbor
Henderson
Henry
Herbal
Hermit
Holmes
re
gi
on
2
2
5
3
3
10
10
105
105
158
162
155
177
177
last name
Masters
Maxim
Maxwell
Mercy
Moon
Most
Nabers
re
gi
on
3
4
2
1
5
4
1
155
179
105
195
160
155
195
. . .Portions of the table are not shown. . .
There are two observations for the patient with the last name Harbor. Since LNAME is an
ORDER variable, each value of LNAME appears only once.
In most non-monospace destinations, the size of the page is considered to be infinite in both
length and width, and these options have no effect.
SEE ALSO
SAS Technical Report P-258 (1993) discusses the LS= (p. 183), PS= (p. 185), and the PANELS=
(p. 184–185) options. Burlew (2005, p. 176) has an example that uses the PANELS= option.
Chapter 4: Only in the LISTING Destination 89
4.4.7 Using the FORMCHAR Option
The characters used to form elements such as lines, corners (see Sections 4.1 and 4.4.1), boxes,
and underlines are determined by the FORMCHAR system option. This option can be overridden
through the use of the FORMCHAR option on the PROC REPORT statement. The syntax of the
use of this option is similar to that of the FORMCHAR system option.
The FORMCHAR option specifies 13 individual characters that are each used as needed under various
circumstances. You can specify the full set of characters, changing those that you would like to have
different from the current settings. However, this is awkward at best. Fortunately, you can also specify
the characters individually by directly indicating which character is to be replaced.
In the example in Section 4.4.1, the character used to create the juncture of a horizontal and
vertical line is supposed to be a plus sign (+). Instead, it is rendered as a small square ( ) when
the report is copied into a word processor. This rendering changes the spacing and causes
misaligned lines.
The character that forms the junction between a vertical and horizontal line is specified in the seventh
position of FORMCHAR. The following example replaces the seventh character with a vertical bar.
* Using the FORMCHAR option;
title1 'Only in LISTING';
title2 'Using the FORMCHAR Option';
proc report data=rptdata.clinics
formchar(7)='|'
nowd split='*'
box;
column region sex ('Patient Weight'wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '(lb)*__';
run;
Only in LISTING
Using the FORMCHAR Option
„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚
Patient Weight
‚
‚re
(lb)
‚
‚gi Sex
____________________‚
‚on F M
n
mean‚
‡ƒƒ…ƒƒ…ƒƒ…ƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒ‰
‚1 ‚ .‚ 4‚
4‚
195‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚10‚ 2‚ 4‚
6‚ 172.33333‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚2 ‚ 6‚ 4‚
10‚
107.8‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚3 ‚ 5‚ 5‚
10‚
145.8‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚4 ‚ 4‚ *‚
14‚ 159.14286‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚5 ‚ 5‚ 3‚
8‚
157.75‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚6 ‚ 4‚ 6‚
10‚
198‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
When rendered in Microsoft Word, the
characters line up quite well. In other
word processors the alignment might
still not be perfect. If you really need
things to look right, an alternate ODS
destination is almost certainly a better
and more robust solution.
(continued)
90 Carpenter’s Complete Guide to the SAS REPORT Procedure
(continued)
‚7 ‚ .‚ 4‚
4‚
151‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚8 ‚ 4‚ .‚
4‚
160‚
‡ƒƒ|ƒƒ|ƒƒ|ƒƒƒƒƒƒƒƒƒƒ|ƒƒƒƒƒƒƒƒƒƒ‰
‚9 ‚ 2‚ 8‚
10‚
187.8‚
Šƒƒ‹ƒƒ‹ƒƒ‹ƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒ
You can specify more than one character by listing the positions and their corresponding
characters. The following FORMCHAR option replaces vertical lines with blanks and
intersections with dashes.
* Using the FORMCHAR option;
title1 'Only in LISTING';
title2 'Replacing Vertical lines with the FORMCHAR Option';
proc report data=rptdata.clinics
formchar(1,4,6,7,8,10)=' - - -'
nowd split='*' box;
column region sex ('Patient Weight'wt,(n mean));
define region / group;
define sex
/ across
'Sex';
define wt
/ analysis '(lb)*__';
run;
Only in LISTING
Replacing Vertical lines with the FORMCHAR Option
„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
Patient Weight
re
(lb)
gi Sex
____________________
on F M
n
mean
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
1
. 4
4
195
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
10 2 4
6 172.33333
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
2
6 4
10
107.8
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
3
5 5
10
145.8
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
4
4 *
14 159.14286
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
5
5 3
8
157.75
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
6
4 6
10
198
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
7
. 4
4
151
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
8
4 .
4
160
ƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒ
9
2 8
10
187.8
Šƒƒ-ƒƒ-ƒƒ-ƒƒƒƒƒƒƒƒƒƒ-ƒƒƒƒƒƒƒƒƒƒŒ
Chapter 4: Only in the LISTING Destination 91
SEE ALSO
The FORMCHAR system option is discussed in more detail in the documentation for Base SAS.
The SAS Guide to the REPORT Procedure, Reference, Release 6.11 (1995, pp. 63–64) uses the
FORMCHAR system option in an example.
4.4.8 Wrapping Data Lines
When a report is too wide for the page, as when a given logical line does not fit within the
constraints of the physical line, both the REPORT and PRINT procedures form groups of lines
that are wrapped together. This problem is rarely an issue in destinations other than LISTING,
because page definitions and row and column wrapping are handled very differently. When the
report is too wide in the LISTING destination, however, the report can become very difficult to
follow.
Line and column wrapping is most common when you have a series of long text strings such as
free-form narrative comments.
The following example builds on the data and code provided online in Sample 637. In addition to
some scoring and demographic information, the data contains a long comment.
* Using the WRAP option;
title1 'Only in LISTING';
title2 'Using WRAP';
data demog;
input name $ sex $ idnum score1-score10 / comment $65.;
format name $5. sex $3. idnum 5. score1-score10 6.
comment $65.;
datalines;
Russ M 123 1 9 3 8 4 7 5 8 6 3
Occasionally has difficulty with verbal communication
Kevin M 456 4 7 5 6 8 5 4 3 2 3
Is very particular about the placement of personal objects
Paige F 789 6 7 4 3 5 8 9 2 3 4
Gets very excited by the success of those people close to her
run;
proc report data=demog
ls=70 nowd
wrap;
column name comment;
define name
/ group;
define comment / display;
break after name / skip;
run;
The LS= option has been set to 70, which leaves the page too narrow to support the two columns.
The comment is therefore wrapped to the following line and is placed below the name. Notice that
the column headers are also stacked.
92 Carpenter’s Complete Guide to the SAS REPORT Procedure
Only in LISTING
Using WRAP
name
comment
Kevin
Is very particular about the placement of personal objects
Paige
Gets very excited by the success of those people close to
her
Russ
Occasionally has difficulty with verbal communication
Without using the WRAP option, the values of NAME and COMMENT would have appeared on
separate pages.
MORE INFORMATION
When a given column’s value does not fit in the space allocated to it, you can also use the FLOW
option (Section 4.5.2) so that you can get the text to wrap within the column. When entire lines
wrap, as in this section, the ID option on the DEFINE statement can be used to help identify the
lines (Section 6.1.6).
SEE ALSO
In a very similar example, Sample 637 shows the use of the WRAP option with the NOHEADER
and NAMED options. SAS Technical Report P-258 (1993, p. 225) uses the WRAP option.
Whereas Ping and Schiefelbein (2006) insert split characters to control where text is to split, Jiang
and Boisvert (2006) determine split locations by counting characters and then calculating the total
width required. Before wrapping, Li (2006) searches long strings for special characters, including
carriage controls and line feeds.
4.5 Other DEFINE Statement Options
There are a few DEFINE statement options that have an effect only when used with traditional
SAS Monospace output. These include the following:
WIDTH= specifies the number of characters to allocate for the width of the column.
SPACING= specifies the number of spaces to leave between columns.
FLOW
specifies that text wider than the WIDTH= value wraps within the block (split
characters are supported).
SEE ALSO
The WIDTH= and FLOW options are used in an example in the SAS Guide to the REPORT
Procedure, Reference, Release 6.11 (1995, pp. 68–70).
Chapter 4: Only in the LISTING Destination 93
4.5.1 Specifying the Column Width
The following example adjusts the column widths for the variables REGION and SEX. The table
in Section 4.4.3 showed problems for each of these columns. Since REGION has a default width
of 2 spaces, the column label was wrapped. SEX had a default width of 1 space, leaving
insufficient space to allow numbers greater than 9. Both of these problems are solved in the
following example with the WIDTH= option.
* Using the WIDTH option;
title1 'Only in LISTING';
title2 'Using the WIDTH Option';
proc report data=rptdata.clinics nowd
split='*'
colwidth=4;
column region sex ('Patient Weight' wt,(n mean));
define region / group width=6;
define sex
/ across
'Sex' width=2;
define wt
/ analysis '(lb)*__';
run;
When applied to numeric and computed variables, the values of the WIDTH= option override the
overall column width specified in the COLWIDTH= option on the PROC statement.
Only in LISTING
Using the WIDTH Option
Sex
region
1
10
2
3
4
5
6
7
8
9
F
M
.
2
6
5
4
5
4
.
4
2
4
4
4
5
10
3
6
4
.
8
Patient Weight
(lb)
__________
n mean
4
195
6
172
10
108
10
146
14
159
8
158
10
198
4
151
4
160
10
188
The label for a character variable
such as SEX is left-justified.
Consequently, in this table, the values
seem not to be under the column
label.
SEE ALSO
Burlew (2005, p. 63) has an example that uses the WIDTH= option.
4.5.2 Using the FLOW Option to Wrap Text
When the field width is not sufficient to display a text value, truncation can occur. This
information loss can be avoided by using the FLOW option. In the following example, the
variable CLINNAME can have a length of up to 27 characters. Because it has been given space
for only 15 characters (WIDTH=15) truncation could occur. However, the FLOW option allows
the text to wrap within the column as needed.
94 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Define Statement wrap option;
title1 'Using Proc REPORT';
title2 'Define Statement FLOW Option';
proc report data=rptdata.clinics nowd;
column region clinnum clinname;
define region
/ group width=6;
define clinnum / group ;
define clinname / group width=15 flow;
run;
Using Proc REPORT
Define Statement FLOW Option
region
1
clinic
number
011234
014321
10
107211
108531
2
023910
clinic name
Boston
National
Medical
Vermont
Treatment
Center
Portland
General
Seattle
Medical Complex
New York Metro
Medical Ctr
When wrapping, SAS attempts to
break lines at word boundaries.
The FLOW option honors the split
character. (The default split
character is a “ / ”.)
. . . Portions of the table are not shown. . .
MORE INFORMATION
Truncation and wrapping issues become less problematic for most other ODS destinations.
Section 8.6.5 discusses in-line formatting sequences that can be used to control how text is split
and indented when it wraps.
SEE ALSO
The online sample program (Sample 634) demonstrates text wrapping, and Michel (2005) uses the
FLOW option in a CALL EXECUTE example.
The two very similar papers by Petersen and Garlach (1999) and Petersen and Garacani (2005)
both discuss a macro that controls the flow and wrapping of text without using the FLOW option.
Whitlock (2000) also discusses a macro that wraps text. SAS Technical Report P-258 (1993,
p. 236) and Burlew (2005, pp. 50–52) both have examples that demonstrate the FLOW option.
4.5.3 Adding Spaces between Columns
The number of spaces between individual columns can be adjusted by using the SPACING=
option. The columns in the example in Section 4.5.2 can be spread apart by specifying the
SPACING= option, which adds spaces to the left of the designated column.
Chapter 4: Only in the LISTING Destination 95
* Define Statement spacing option;
title1 'Using Proc REPORT';
title2 'Define Statement SPACING Option';
proc report data=rptdata.clinics nowd;
column region clinnum clinname;
define region
/ group width=6;
define clinnum / group
spacing=5;
define clinname / group width=15 spacing=5 flow;
run;
The additional space before the clinic number and before the clinic name give the report a less
crowded look.
Using Proc REPORT
Define Statement SPACING Option
region
1
clinic
number
011234
014321
10
107211
108531
2
023910
024477
clinic name
Boston
National
Medical
Vermont
Treatment
Center
Portland
General
Seattle
Medical Complex
New York Metro
Medical Ctr
New York
General
Spaces are inserted to the left of
the designated column.
The additional spaces between
the REGION and CLINNUM
columns are specified by the
SPACING= option on the
DEFINE statement for
CLINNUM.
. . .Portions of the table are not shown. . .
MORE INFORMATION
The SPACING= option can also be used to effectively concatenate columns. See the last example
in Section 6.4.1.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 44–45) discusses the SPACING = option for both the
PROC REPORT and DEFINE statements.
96 Carpenter’s Complete Guide to the SAS REPORT Procedure
4.6 Chapter Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Generate a report that lists total PROFIT for each PRODUCT_LINE
within each YEAR. List the products ACROSS the report. You might want to build on the
results of Exercise 2b in Chapter 2.
Include the following:
HEADLINE and HEADSKIP options
repeated characters in the spanning header for product line
2. The data table SASHELP.RETAIL contains quarterly sales information. List the columns
YEAR, DATE, and SALES. Do the following:
Use YEAR as a grouping variable.
Use the PANELS=, BOX, and PSPACE= options.
3. Building on Exercise 2 in this section, use the WIDTH= and SPACING= options.
C h a p t e r
5
Creating and Modifying Columns Using the
Compute Block
5.1 Coordinating with the COLUMN and DEFINE Statements 98
5.2 Calculations Based on Statistics 99
5.3 Calculating Percentages within Groups 101
5.4 Using _PAGE_ with BEFORE and AFTER 103
5.5 Using the OUT= Option to View Report Break Information 104
5.6 Chapter Exercises 106
One of the great strengths of PROC REPORT is its ability to create columns based on calculations
carried out during the execution of the procedure. Columns that are not on the incoming data set
can be created and displayed within the PROC REPORT step. Often this can eliminate one or
more DATA steps. The compute block provides the power and flexibility to mold the column in a
variety of ways.
As was seen in Section 2.6, the use of the compute block can be fairly straightforward. However,
as the compute block is used to complete more complex tasks, it becomes more important for the
programmer to achieve a more thorough understanding of tasks associated with the compute
block.
Compute blocks can be used to create both numeric and character variables, and an extensive
number of the SAS language elements that give the DATA step much of its power and
functionality are also available for use in the compute block. In this chapter we examine some of
the issues associated with the generation of additional columns through the use of the compute
block.
98 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
When you use existing or computed columns in a compute block, naming conventions are not
always either intuitive or straightforward. How to name a report item in a compute block is
discussed briefly in some of the following sections and in detail in Section 7.2.
SEE ALSO
Cochran (2005) includes several easy-to-understand examples of the use of compute blocks. A
number of compute statement options are listed in SAS Technical Report P-258 (1993, p. 53).
Burlew (2005, pp. 31–49) introduces the compute block through a series of examples.
5.1 Coordinating with the COLUMN and DEFINE
Statements
Variables on the data set that is processed by PROC REPORT are named on the COLUMN
statement, and the way that these variables are used is specified on the DEFINE statement. This is
also true for columns created through the use of the compute block. In addition, the name of the
computed variable also appears on the COMPUTE statement.
When a column is created through a compute block, the new column is named and that column
name is used on both the COLUMN statement and on a DEFINE statement. The compute blocks
that were discussed in Sections 2.6.1 through 2.6.3 used the BEFORE and AFTER location
specifications. This means that although these compute blocks could be tied to a report item, by
necessity they could not be associated with a new column. Whether or not the BEFORE or
AFTER location is specified, when a compute block is associated with a report item, such as a
computed variable, the name of the report item appears on the COLUMN, DEFINE, and
COMPUTE statements. It is this name that is used to create and coordinate the link between the
three statements.
In the following example, the patients’ weights are displayed in both pounds and kilograms (the
example in Section 2.6.4 only converts to kilograms without creating a new column). Since the
units for the variable WT are in pounds, a new column containing the weight in kilograms needs
to be created. This is accomplished in a compute block.
* Creating a new column with a compute block;
title1 'Using The COMPUTE Block';
title2 'Adding a Computed Column';
proc report data=rptdata.clinics nowd split='*';
column lname sex (' Weight *--' wt wtkg); n
define lname
/ order
width=18 'Last Name*--';
define sex
/ display width=6 'Gender*--';
define wt
/ display format=6. 'Pounds*--';
define wtkg o / computed p format=9.2 'Kilograms*--';
compute wtkg; q
wtkg = wt / 2.2; r
endcomp; s
run;
Chapter 5: Creating and Modifying Columns Using the Compute Block 99
n A name (WTKG) for the new column is added to the COLUMN statement.
o The DEFINE statement associated with the new column (WTKG) has a define type of
COMPUTED p.
q The COMPUTE statement contains the name of the computed variable.
r The new variable (WTKG) is calculated by dividing the weight (WT) in pounds by 2.2.
s Compute blocks are terminated with an ENDCOMP statement.
The resulting table shows the weight both in pounds and kilograms.
Using The COMPUTE Block
Adding a Computed Column
Last Name
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Adams
Adamson
Alexander
Antler
Atwood
Banner
Baron
Batell
Gender
ƒƒƒƒƒƒ
F
F
M
M
M
M
M
F
Weight
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Pounds Kilograms
ƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒ
155
70.45
158
71.82
175
79.55
240
109.09
105
47.73
175
79.55
160
72.73
155
70.45
Because this LISTING
destination table uses the
SAS Monospace font, the
horizontal lines are
preserved (see Section 4.1).
. . . Portions of the table are not shown . . .
MORE INFORMATION
A simple compute block that modifies the value of an existing column by transforming the value
of weight from pounds to kilograms is presented in Section 2.6.4.
5.2 Calculations Based on Statistics
In the previous example, because WT was given a define type of DISPLAY, the variable name
was used explicitly in the compute block. In this example, the mean for each region is calculated,
and it is from this mean that we create a column for kilograms. This time WT is an ANALYSIS
variable, and a compound column name (WT.MEAN o) is used in the calculation.
100 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Calculations based on statistics block;
title1 'Using The COMPUTE Block';
title2 'Calculations Based on a Statistics Column';
proc report data=rptdata.clinics nowd split='*';
column region (' Weight *--' wt wtkg);
define region / group
width=7
'Region*--';
define wt
/ analysis mean
format=8.1
'Pounds*--'; n
define wtkg
/ computed format=9.2
'Kilograms*--';
compute wtkg;
wtkg = wt.mean / 2.2; o
endcomp;
run;
n The mean weight (in pounds) is calculated for WT across the group variable (REGION).
o The weight in kilograms is calculated by converting it from the mean weight in pounds. Notice
that the compound name, WT.MEAN, reflects the calculated statistic (in pounds).
Using The COMPUTE Block
Calculations Based on a Statistics Column
Region
ƒƒƒƒƒƒƒ
1
10
2
3
4
5
6
7
8
9
Weight
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Pounds Kilograms
ƒƒƒƒƒƒƒƒ ƒƒƒƒƒƒƒƒƒ
195.0
88.64
172.3
78.33
107.8
49.00
145.8
66.27
159.1
72.34
157.8
71.70
198.0
90.00
151.0
68.64
160.0
72.73
187.8
85.36
The following somewhat silly example demonstrates more fully the need to be able to use
compound names. The compute block is used to calculate the standard error, which is based on
the standard deviation (STD) and the square root of N. This is silly because the standard error is
also available directly as STDERR and does not really need to be calculated using a formula.
Chapter 5: Creating and Modifying Columns Using the Compute Block 101
* Calculations based on two statistics columns;
title1 'Using The COMPUTE Block';
title2 'Calculations Based on Two Statistics Columns';
proc report data=rptdata.clinics nowd split='*';
column region (wt,(mean std stderr n) wtse); n
define region / group
width=7 'Region*--';
define wt
/ analysis 'Pounds*--';
define wtse
/ computed format=9.2 'STD Error*--';
compute wtse;
wtse = wt.std / sqrt(wt.n); o
endcomp;
run;
n A list of statistics has been requested for the single analysis variable WT. The computed
variable (WTSE), which is to be calculated, must also appear on the COLUMN statement.
o The calculation involves the use of both the standard deviation of WT (WT.STD) and the
number of observations (WT.N). The use of compound names allows this refined specification.
In the resulting table, the standard error has been generated twice. It is first calculated directly (as
a named statistic on the COLUMN statement). Then the value is calculated in the compute block.
For comparison purposes, both have been shown in the table.
Using The COMPUTE Block
Calculations Based on Two Statistics Columns
Region
Pounds
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
STD Error
ƒƒƒƒƒƒƒ
1
10
2
3
4
5
6
7
8
9
mean
195
172.33333
107.8
145.8
159.14286
157.75
198
151
160
187.8
ƒƒƒƒƒƒƒƒƒ
0.00
2.95
1.31
11.13
6.34
15.64
7.52
2.31
1.15
8.52
std
0
7.2295689
4.1311822
35.197222
23.719052
44.248487
23.776739
4.6188022
2.3094011
26.951809
stderr
0
2.9514591
1.3063945
11.130339
6.3391832
15.644202
7.5188652
2.3094011
1.1547005
8.5229103
n
4
6
10
10
14
8
10
4
4
10
MORE INFORMATION
The spanning headers in this LISTING destination example were created using a combination of a
split character and repeat characters (see Section 4.3.2).
5.3 Calculating Percentages within Groups
Because a percentage calculation is the ratio of the current value to a total for the group, the total
must be available for the calculation. As a result, the calculation of percentages requires the use of
two compute blocks. The first is used to determine the total across the group (COMPUTE
102 Carpenter’s Complete Guide to the SAS REPORT Procedure
BEFORE n), and then this total is used on each detail row of the table p to calculate the
percentage in the second compute block.
Notice in this example that the compound variable name WT.SUM has two different meanings,
depending on which compute block is being executed. When it is used in a summary across all
visits to a clinic (n and r), it contains the total weight for that clinic. When a detail row is being
processed q, WT.SUM contains the value of WT for that row.
* Calculating percentages within groups;
title1 'Using the COMPUTE Block';
title2 'Percentages Within Groups';
proc report data=rptdata.clinics nowd split='*';
where clinnum in('031234', '036321');
column clinnum lname wt prctwt ;
define clinnum / group width=10;
define lname
/ order
'Last Name';
define wt
/ analysis format=6.
'Weight';
define prctwt
/ computed format=percent8.1
'Percent*of Total';
compute before clinnum;
totwt = wt.sum; o
endcomp;
n
compute prctwt; p
prctwt = wt.sum / totwt;
endcomp;
break after clinnum
run;
q
/ dol skip summarize suppress;
r
n This compute block is executed before the individual rows of the table are processed. This
makes the temporary variable TOTWT available for use in the second compute block p. The
keyword BEFORE is available for this compute block because CLINNUM is a GROUP
variable.
o The total weight is calculated as the total weight within the specific clinic number. Since
CLINNUM is a GROUP variable, WT.SUM holds the total weight. During the execution of
this compute block, WT.SUM is the total for the upcoming CLINNUM, and this value is what
is saved in the temporary variable TOTWT.
p This compute block is used to calculate the percentages for each row of the table.
q In this equation, we use the total weight for all the patients that have visited the clinic,
TOTWT. This total is used with the individual patient’s weight (WT.SUM). Eventually, the
value is displayed using the PERCENT8.1 format.
r The BREAK statement requests a summary line after each clinic. This statement causes an
additional summary row to be added to the final report. The compute block for PRCTWT
is also executed for the corresponding row in the computed summary information.
Chapter 5: Creating and Modifying Columns Using the Compute Block 103
Using the COMPUTE Block
Percentages Within Groups
clinic
number
031234
Last Name
Candle
Henry
Panda
Smith
Weight
195
162
195
162
======
714
036321
Herbal
Jones
Masters
Stubs
155
105
155
105
======
520
Percent
of Total
27.3%
22.7%
27.3%
22.7%
========
100.0%
29.8%
20.2%
29.8%
20.2%
========
100.0%
r
The percentage across the
clinic is of course 100%. This
value is calculated
automatically when the row
resulting from the BREAK
AFTER CLINNUM statement
r is processed.
On this row, the temporary
variable TOTWT and
WT.SUM contain the same
value.
MORE INFORMATION
Additional information on the formation of the computed summary information can be found in
Sections 1.4.2 and 7.1, as well as in Chapter 11. This specific example is examined further in
Section 5.5.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 122–123) has an example that initializes a variable in a
compute block. Percentages are calculated in several examples in P-258 (Chapter 8).
5.4 Using _PAGE_ with BEFORE and AFTER
In Section 2.6.2, the compute block was used with the LINE statement to generate a footnote
using a COMPUTE AFTER statement.
compute after;
line @20 'Weight taken during';
line @20 'the entrance exam.';
endcomp;
The two lines of text generated by these two LINE statements appear at the end of the report.
Unlike a footnote generated by a FOOTNOTE statement, the text generated by these two LINE
statements does not appear at the bottom of each page if multiple pages are required. This is
because the compute block is targeted to execute only at the end of the report. Its target is implied;
essentially the COMPUTE AFTER statement is interpreted as COMPUTE AFTER REPORT.
For the LISTING destination, we can use the _PAGE_ option to give us additional control by
explicitly specifying _PAGE_ as the target. The compute block target determines when the
compute block executes, and when _PAGE_ is used as a target, it forces the compute block’s
execution at page boundaries. In a sense, _PAGE_ is used as if it were a grouping variable.
However, the text is instead written after the page breaks.
104 Carpenter’s Complete Guide to the SAS REPORT Procedure
compute after _page_;
line @10 'Weight taken during';
line @10 'the entrance exam.';
endcomp;
The same technique can be used to place text at the top of the page (between the titles and the
body of the report) by using BEFORE.
compute before _page_;
line @10 'Weight taken during';
line @10 'the entrance exam.';
endcomp;
Of course, the concept of a page is not the same for all ODS destinations, and this difference can
be an issue for some destinations, such as HTML and PDF. For other destinations, page break
determination is not always directly controlled by SAS. For these destinations, which include the
RTF destination, _PAGE_ either does not work at all, or, at best, does not work as you might
anticipate.
MORE INFORMATION
The _PAGE_ target is used with justification options in a similar example in Section 7.6.1, and
the example in Section 8.2.2 uses a similar compute block to place a logo at the top of each report
page. In the example in Section 7.7.3, it is used in a report that does page counting.
SEE ALSO
In similar papers, Gonzalez (2003) and Dunn (2004) both use the compute block to control title
and footnote lines. Two online sample programs (Sample 607 and Sample 610) discuss various
aspects of printing text headers before and after pages. Humphreys (2006) uses computed
variables to control page breaks.
5.5 Using the OUT= Option to View Report Break
Information
When you use compute blocks, it is sometimes helpful to be able to see the how the break information
is being used in the construction of the report rows. This break information can be visualized by using
the OUT= option n on the PROC REPORT statement to create an output data set.
The PROC REPORT step’s computed summary information is processed in a sequence of events
that are further described in Section 7.1 and in more detail in Chapter 11, “Details of the PROC
REPORT Process.” The data set that is saved when you use the OUT= option is a final result of
those processing steps. The output data set shows the report rows (including those that will not be
written to the final report). When this data set is written for us by PROC REPORT, all the
compute block processing has been performed, and although there is really no way to trap or trace
the compute block processing, an examination of the output data set created by PROC REPORT
can still be informative.
In addition to the other report items, the output data set includes the automatic temporary variable
_BREAK_. This variable notes observations in the output data set that are a result of data
summarizations, such as those generated by BREAK and RBREAK statements.
Chapter 5: Creating and Modifying Columns Using the Compute Block 105
The following example repeats the REPORT step used in Section 5.3, but it adds an OUT= option
to the PROC REPORT statement n.
* Calculating percentages within groups;
title1 'Using the COMPUTE Block';
title2 'Examining the Output Data Set';
proc report data=rptdata.clinics
out=r5_5out n
nowd split='*';
where clinnum in('031234', '036321');
column clinnum lname wt prctwt ;
define clinnum / group width=10;
define lname
/ order
'Last Name';
define wt
/ analysis format=6.
'Weight';
define prctwt
/ computed format=percent8.1
'Percent*of Total';
compute before clinnum;
totwt = wt.sum;
endcomp;
o
compute prctwt;
prctwt = wt.sum / totwt;
endcomp;
break after clinnum
run;
/ dol skip summarize suppress;
p
proc print data=r5_5out;
title3 Summarized Data Set;
run;
The PROC PRINT of the data set generated by the OUT= option in this example results in the
following:
Using the COMPUTE Block
Examining the Output Data Set
Summarized Data Set
Obs
1
2
3
4
5
6
7
8
9
10
11
12
clinnum
031234
031234
031234
031234
031234
031234
036321
036321
036321
036321
036321
036321
lname
Candle
Henry
Panda
Smith
Herbal
Jones
Masters
Stubs
wt
prctwt
_BREAK_
714
195
162
195
162
714
520
155
105
155
105
520
.
0.27311
0.22689
0.27311
0.22689
1.00000
0.72829
0.29808
0.20192
0.29808
0.20192
1.00000
clinnum
o
clinnum
clinnum
p
o
clinnum
p
106 Carpenter’s Complete Guide to the SAS REPORT Procedure
o This compute block generates a summary line before each distinct value of CLINNUM.
This row is only a result of the compute block and does not appear in the final report. Because
WT.SUM was referenced in a compute block, we can see how that summary information was
captured. Because there is no BREAK statement associated with the COMPUTE BEFORE
statement, this row does not appear in the final report. However, we can see from the output
data set that WT.SUM was available before each CLINNUM group in order to be assigned
to the temporary variable TOTWT.
p After each group of values for CLINNUM, the BREAK statement generates a summary or
break row in the final report, and this summary row is also reflected in the OUT= data set.
Notice the inclusion of the automatic column _BREAK_. This column can be used to track break
events as the table is calculated. The values of _BREAK_ can be checked by using
IF-THEN/ELSE processing in the compute block itself.
MORE INFORMATION
The OUT= option and the contents of the resulting table are used extensively in the examples that
discuss the compute block in Chapter 7, “Extending Compute Blocks.”
5.6 Chapter Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Create a report that shows total PROFIT for each PRODUCT_LINE
within each YEAR. You might wish to build on the results of Exercise 3 in Chapter 2.
Compute the percentage of annual sales that can be attributed to each product line.
2. Building on the solution to Exercise 1 in this section, add a summary line for each year by
using a BREAK statement.
Do you need to do anything extra for the percentage on this summary line?
What if you also used an RBREAK statement?
3. Building on the solution to Exercise 2 in this section, use the OUT= option to see the final
output data table.
P a r t
2
Taking PROC REPORT Beyond the Basics
Chapter 6
Refining Our Understanding of the PROC REPORT Step
Chapter 7
Extending Compute Blocks
Chapter 8
Using PROC REPORT with ODS
147
211
109
108 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
6
Refining Our Understanding of the PROC REPORT
Step
6.1 Additional DEFINE Statement Options 110
6.1.1 Changing Display Order with DESCENDING 110
6.1.2 Specification of Column Justification 111
6.1.3 Allowing the Use of Missing Classification Items 113
6.1.4 Controlling the Use of Analysis Items with All Missing or Zero
Values 115
6.1.5 Using NOPRINT 118
6.1.6 Identification Columns 119
6.1.7 Creating Vertical Page Breaks 120
6.2 Using Variable Aliases 121
6.3 Nesting Variables 122
6.4 Taking Full Advantage of Formats 123
6.4.1 User-Defined Formats 123
6.4.2 Preloading Formats 126
6.4.3 Order Based on Format Definition 130
6.5 Other PROC Statement Options 131
6.5.1 Removing Headers 131
6.5.2 Using NAMED Output 132
6.5.3 Debugging with the LIST Option 134
6.5.4 Including MISSING Classification Levels 134
110 Carpenter’s Complete Guide to the SAS REPORT Procedure
6.6 BY-Group Processing 136
6.6.1 Using the BY Statement 137
6.6.2 Creating Breaks with BY Groups 138
6.6.3 Using the #BYVAL and #BYVAR Options
139
6.6.4 BY Groups and the Output Delivery System 141
6.7 Calculations Using the FREQ Statement 144
6.8 A Further Comment on Paging Issues 145
6.9 Chapter Exercises 146
As our desire to create more complex reports increases, so too must our understanding of the
PROC REPORT step’s available options and statements. Depending on the types of reports that
you create, you may or may not use all or even most of the options and techniques discussed in
this chapter. Knowing of them, however, is very important to your overall understanding of the
PROC REPORT step.
6.1 Additional DEFINE Statement Options
A number of supplemental options can be used with the DEFINE statement to augment the
display of the information in the associated variable. DEFINE statement options not discussed
earlier in this book include the following:
DESCENDING
CENTER / LEFT / RIGHT
MISSING / NOZERO
NOPRINT
ID
PAGE
reverses the order of values when used with define types GROUP,
ORDER, and ACROSS.
controls the justification of the column header and formatted value
within a column.
controls how report items are to be handled when all values are
either missing or zero.
specifies that this report item is not to be displayed, although it
appears in the COLUMN statement.
identifies one or more columns to repeat when a logical report line
wraps to a new physical line. This option is usually used in
conjunction with the PAGE option.
inserts a page break between columns. This option is usually used
with the ID option.
MORE INFORMATION
The DEFINE statement and a number of its display options are introduced and discussed in
Sections 2.4 and 4.5.
6.1.1 Changing Display Order with DESCENDING
Typically when items are ordered, the default order is ascending. The DESCENDING option can
be used when you would like to reverse the order of items in columns with define types of
GROUP, ACROSS, or ORDER.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 111
In the following example, CLINNAME appears as a GROUP variable. Since GROUP implies an
ordering of the values, the default presentation would be for the values to appear in ascending
order. The DESCENDING option reverses that order.
* Using DESCENDING;
options nocenter;
title1 'Refining REPORT Appearance';
title2 'Using the DESCENDING Option';
proc report data=rptdata.clinics
(where=(region in('1','2','3'))) nowd;
columns clinname ht wt;
define clinname / group descending;
define ht
/ analysis mean format=6.1;
define wt
/ analysis mean format=6.1;
run;
The clinic names (CLINNAME) are now in descending order.
Refining REPORT Appearance
Using the DESCENDING Option
clinic name
Vermont Treatment Center
Philadelphia Hospital
New York Metro Medical Ctr
New York General Hospital
Naval Memorial Hospital
Geneva Memorial Hospital
Boston National Medical
Bethesda Pioneer Hospital
height
in
inches
74.0
65.0
64.0
63.5
65.5
64.0
74.0
72.5
weight
in
pounds
195.0
112.0
105.0
107.0
130.0
115.0
195.0
178.5
6.1.2 Specification of Column Justification
The documentation states that the justification options CENTER, RIGHT, and LEFT change the
justification of column headings as well as formatted values within a column. These options were
originally designed to work with the LISTING destination, and the impact of their use is less than
inspiring for other destinations. In fact, the results of the CENTER, RIGHT, and LEFT options
depend on the ODS destination and whether the variable is numeric or character. The following
table shows the default justifications. The values in bold can be overridden with these DEFINE
statement options.
Character
Variables
Numeric
Variables
Header
Value
Header
Value
LISTING
Destination
left
left
right
right
Non-text
Destination
center
left
center
right
112 Carpenter’s Complete Guide to the SAS REPORT Procedure
In the following example, each of the three columns receives a different justification option. By
default REGION, which is a character variable, is left-justified.
ods html file="&path\results\ch6_1_2.html";
* Using Justification options;
title1 'Refining REPORT Appearance';
title2 'Using Justification Options';
proc report data=rptdata.clinics
(where=(region in('1','2','3')))
nowd;
columns region ht wt;
define region / group width=7 center;
define ht
/ analysis mean left format=6.1;
define wt
/ analysis mean right format=6.1;
run;
ods html close;
In the LISTING destination, the HEADER, but not the value for mean HT, is left-justified.
Refining REPORT Appearance
Using Justification Options
region
1
2
3
height
in
inches
74.0
63.8
68.2
weight
in
pounds
195.0
107.8
145.8
In the non-LISTING (HTML) destination, the mean value for HT has been left-justified as
requested.
In future versions of SAS, it is possible that these options will be ignored by non-LISTING
destinations.
MORE INFORMATION
For non-LISTING destinations, it is more appropriate to change justification with the JUST=
attribute modifier in the STYLE= option (see Section 8.2) or in the CALL DEFINE routine (see
Section 8.3).
Chapter 6: Refining Our Understanding of the PROC REPORT Step 113
6.1.3 Allowing the Use of Missing Classification Items
Missing values can cause problems in our reports, and unless we are aware of the issues involved,
the report itself might not reflect the true underlying data. This can happen to us on both summary
and detail reports.
When classification variables (GROUP, ORDER, or ACROSS) have missing values, those
observations are by default excluded from the report. The MISSING option specifies that missing
values are valid levels of the classification variable and should therefore be included. MISSING
can appear on the PROC REPORT statement, where it applies to all classification variables, or on
the DEFINE statement, where it is applied to an individual variable.
The following report counts the number of observations within a region and the number of
procedure types (PROCED) that were used (nominally there should be a procedure type for each
visit).
* Counting Procedure types within Region;
title1 'Refining REPORT Appearance';
title2 'Counting Procedures Without MISSING';
proc report data=rptdata.clinics nowd;
column region n proced,n;
define region / group width=6;
define n
/
width=3; n
define proced / across width=3;
run;
Refining REPORT Appearance
Counting Procedures Without MISSING
procedure code
n
region
1
10
2
3
5
6
7
8
9
n
4
6
8
6
6
4
2
4
2
1
n
.
.
.
.
.
.
2
.
.
2
n
.
6
.
.
4
4
.
2
.
3
n
4
.
8
6
2
.
.
2
2
In this report it is easy to see that the
total number of procedures matches
the overall number of patient visits n.
Oops, the totals do not count patients
with missing procedure codes. Notice
that region 4 is not even included in the
report.
In fact, the number of visits n is not correct. It is actually the number of visits with nonmissing
procedure codes. If we include a MISSING option Y on the DEFINE statement for PROCED, we
get an entirely different count.
title1 'Refining REPORT Appearance';
title2 'Counting Procedures with MISSING';
proc report data=rptdata.clinics nowd;
column region n proced,n;
define region / group width=6;
define n
/
width=3;
define proced / across width=3 missing;
run;
o
114 Carpenter’s Complete Guide to the SAS REPORT Procedure
Refining REPORT Appearance
Counting Procedures with MISSING
procedure code
n
region
1
10
2
3
4
5
6
7
8
9
n
4
6
10
10
14
8
10
4
4
10
Y 1
n
.
.
2
4
14
2
6
2
.
8
n
.
.
.
.
.
.
.
2
.
.
2
n
.
6
.
.
.
4
4
.
2
.
3
n
4
.
8
6
.
2
.
.
2
2
The overall total n still reflects the
sum of the types of procedure
codes, PROCED. However, since
missing values are also now
included in the total Y, the overall
number now reflects the actual
number of visits.
The total N now reflects all visits and not just the visits with nonmissing procedure codes. Not
only was the number of visits not correct, but region 4 was completely eliminated from the first
report!
The same type of distortion can occur on a detail report as well. This is demonstrated in the
following example. REGION=11 has been artificially created with two levels of the variable
CLINNAME, one of the levels being missing.
Refining REPORT Appearance
Artificial Region 11 Data
Obs
1
2
3
4
clinname
Boston National Medical
Boston National Medical
region
lname
wt
ht
11
11
11
11
Lawless
Nabers
Mercy
Taber
195
.
195
.
74
0
74
0
When the MISSING option is not used, only nonmissing values of the GROUP variable
CLINNAME are included in the table.
* Using MISSING on the DEFINE statement;
title1 'Refining REPORT Appearance';
title2 'Without MISSING Option';
proc report data=reg11 nowd;
column region clinname ht wt;
define region
/ group
format=$6.;
define clinname / group;
define ht
/ analysis mean format=6.1;
define wt
/ analysis mean format=6.1;
run;
Chapter 6: Refining Our Understanding of the PROC REPORT Step 115
We see that the data associated with the missing clinic name has been excluded from the table.
Refining REPORT Appearance
Without MISSING Option
region
11
clinic name
Boston National Medical
height
in
inches
0.0
weight
in
pounds
.
Including the MISSING option on the DEFINE statement makes the missing clinic name a valid
classification (grouping) variable.
* Using MISSING on the DEFINE statement;
title1 'Refining REPORT Appearance';
title2 'MISSING Option on DEFINE Statement';
proc report data=reg11 nowd;
column region clinname ht wt;
define region
/ group
format=$6.;
define clinname / group missing;
define ht
/ analysis mean format=6.1;
define wt
/ analysis mean format=6.1;
run;
Refining REPORT Appearance
MISSING Option on DEFINE Statement
region
11
clinic name
Boston National Medical
height
in
inches
74.0
0.0
weight
in
pounds
195.0
.
MORE INFORMATION
Missing classification levels can also be included through the use of the MISSING option on the
PROC REPORT statement (see Section 6.5.4).
6.1.4 Controlling the Use of Analysis Items with All Missing or
Zero Values
Whereas the MISSING option (Section 6.1.3) controls the display of a classification variable with
missing values, the NOZERO option prevents the display of a nonclassification column with
items that have all zero, all missing, or some combination of only zero and missing values.
To demonstrate in the following examples, let’s create an artificial region (12) which has all
missing values for WT and all zero values for HT.
data reg12(keep=region clinname lname wt ht edu);
set rptdata.clinics(where=(region in('1','2')));
region='12';
wt=.;
ht=0;
run;
116 Carpenter’s Complete Guide to the SAS REPORT Procedure
When the NOZERO option is not used, the columns associated with HT and WT are displayed.
* DEFINE statement without NOZERO;
title1 'Refining REPORT Appearance';
title2 'Artificial Region 12 Data';
title3 'Without the NOZERO Option';
proc report data=reg12 nowd;
column region lname edu ht wt;
define region / group
format=$6.;
define lname / order;
define edu
/ analysis format=9.0;
define ht
/ analysis format=6.1;
define wt
/ analysis format=6.1;
run;
This report shows that the columns HT and WT contain all zero or missing values.
Refining REPORT Appearance
Artificial Region 12 Data
Without the NOZERO Option
region
12
last name
Atwood
Haddock
Harbor
Ingram
Lawless
Leader
Little
Long
Maxwell
Mercy
Nabers
Saunders
Taber
years of
education
14
14
14
14
14
10
14
12
14
14
10
10
12
10
height
in
inches
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
weight
in
pounds
.
.
.
.
.
.
.
.
.
.
.
.
.
.
The columns for HT
and WT are
displayed even
though only zero or
missing values are
in the report.
Adding the NOZERO option on the DEFINE statement prevents the display of the HT and WT
columns, because they contain only missing or zero values.
* Using NOZERO on the DEFINE statement;
title1 'Refining REPORT Appearance';
title2 'Artificial Region 12 Data';
title3 'With the NOZERO Option on EDU HT and WT';
proc report data=reg12 nowd;
column region lname edu ht wt;
define region / group
format=$6.;
define lname / order;
define edu
/ analysis format=9.0 nozero;
define ht
/ analysis format=6.1 nozero;
define wt
/ analysis format=6.1 nozero;
run;
Chapter 6: Refining Our Understanding of the PROC REPORT Step 117
The NOZERO option has no effect on the EDU column, but the columns for HT and WT, which
are all zero or all missing, do not appear in the table.
Refining REPORT Appearance
Artificial Region 12 Data
With the NOZERO Option on EDU HT and WT
region
12
last name
Atwood
Haddock
Harbor
Ingram
Lawless
Leader
Little
Long
Maxwell
Mercy
Nabers
Saunders
Taber
years of
education
14
14
14
14
14
10
14
12
14
14
10
10
12
10
The NOZERO option has suppressed
the printing of both the WT and HT
columns.
The NOZERO option can also be used with summary statistics. In the next example, the MEAN
calculations for both HT and WT are suppressed.
* Using NOZERO on the DEFINE statement;
title1 'Refining REPORT Appearance';
title2 'Artificial Region 12 Data';
title3 'With the NOZERO Option on a Statistic';
proc report data=reg12 nowd;
column region clinname edu ht wt;
define region
/ group
format=$6.;
define clinname / group;
define edu
/ analysis mean format=9.0 nozero;
define ht
/ analysis mean format=6.1 nozero;
define wt
/ analysis mean format=6.1 nozero;
run;
Refining REPORT Appearance
Artificial Region 12 Data
With the NOZERO Option on a Statistic
region
12
clinic name
Boston National Medical
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
Vermont Treatment Center
years of
education
10
14
13
14
10
118 Carpenter’s Complete Guide to the SAS REPORT Procedure
The NOZERO option operates on page-sized chunks of the report, and therefore depends on the
destination’s definition of a page. For non-monospace destinations, this becomes problematic.
Starting in SAS 9.2, NOZERO will only be available for monospace destinations (e.g., LISTING),
and in any other destinations it will be ignored after producing a warning message in the log.
6.1.5 Using NOPRINT
It is sometimes necessary to include a variable on the COLUMN statement that is not to appear in
the report itself. For example, a variable might be needed as part of computations to be performed
in a compute block. This type of variable can be excluded from the report by specifying
NOPRINT on the DEFINE statement.
In the following example, the weight in kilograms is to be calculated from the weight in pounds
(which is not to be displayed).
* Masking a column with NOPRINT;
title1 'Refining REPORT Appearance';
title2 'Masking a Column with NOPRINT';
proc report data=rptdata.clinics nowd;
column lname sex (' Weight' wt wtkg);
define lname
/ order
width=18
'Last Name';
define sex
/ display width=6
'Gender';
define wt
/ analysis noprint;
define wtkg
/ computed format=9.2 'Kilograms';
compute wtkg;
wtkg = wt.sum / 2.2;
endcomp;
run;
Refining REPORT Appearance
Masking a Column with NOPRINT
Last Name
Adams
Adamson
Alexander
Antler
Atwood
Banner
Baron
Gender
F
F
M
M
M
M
M
Weight
Kilograms
70.45
71.82
79.55
109.09
47.73
79.55
72.73
The column WT is available
in the compute block (as
WT.SUM), but because of
the NOPRINT option, it is
not included in the report.
...Portions of the table are not shown...
MORE INFORMATION
The weight in pounds was also converted to kilograms in the examples in Sections 5.1 and 5.2. In
those examples, the computed value replaced the original value.
SEE ALSO
The PROC REPORT step in the online example Sample 770 uses the NOPRINT option to
suppress a column that is used to order the rows, but is not to be printed. Mitchell (2005) uses the
NOPRINT option to suppress everything except the titles.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 119
6.1.6 Identification Columns
Like the ID statement in the PROC PRINT step, the ID option is used to identify those columns
that should be repeated when the physical page is not wide enough and a logical row must wrap
across more than one physical row. The ID option is placed on the DEFINE statement associated
with the rightmost column that is to be repeated. This option is often used in conjunction with the
PAGE option (see Section 6.1.7).
In the following example, both REGION and the clinic name, CLINNAME, are to be used as row
identification columns. Since CLINNAME is the rightmost of these two columns, the ID option is
placed on its DEFINE statement.
* Using ID;
title1 'Refining REPORT Appearance';
title2 'Using ID';
proc report data=rptdata.clinics
(where=(region in('1','2')))
nowd;
columns region clinname (ht wt),(min max n mean median);
define region
/ group
width=6;
define clinname / group
id;
define ht
/ analysis;
define wt
/ analysis;
run;
The ID option specifies that both the clinic name (CLINNAME) column and any columns to its
left (REGION) will be repeated when the line wraps. In the following example, the statistics
based on WT appear on a second page, while the left two columns repeat.
Refining REPORT Appearance
Using ID
region
1
2
clinic name
Boston National Medical
Vermont Treatment Center
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
height in inches
n
mean
median
2
74
74
2
74
74
2
64
64
4
63.5
63.5
4
64
64
min
74
74
64
63
64
max
74
74
64
64
64
min
195
195
115
105
105
weight
max
195
195
115
109
105
Refining REPORT Appearance
Using ID
region
1
2
clinic name
Boston National Medical
Vermont Treatment Center
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
in pounds
n
mean
2
195
2
195
2
115
4
107
4
105
median
195
195
115
107
105
In this example we got lucky in that the logical line was broken for wrapping at a reasonable (in
this case perfect) place. Most folks are generally not going to be this fortunate, so we want to be
able to force where the break takes place. Section 6.1.7 discusses the PAGE option, which can
help with this issue.
120 Carpenter’s Complete Guide to the SAS REPORT Procedure
The ID option is most commonly used with output destinations, such as LISTING, that have
output with fixed page widths. However, it is appropriate whenever the logical line is too wide for
the physical space.
SEE ALSO
Albarran (2003) uses the ID option in an automated REPORT macro. The ID option is
demonstrated in the online sample program Sample 635.
6.1.7 Creating Vertical Page Breaks
The PAGE option can be used to create what is effectively a vertical page break between
columns. Because PROC REPORT lays down the columns from left to right, when PAGE is used
on a DEFINE statement, a new page is created when that column is written.
In Section 6.1.6, the ID option is discussed as a way to specify columns that should be repeated
when the report’s logical line spans more than one physical line. This option works well with the
PAGE option, and these two options are usually used together in a PROC REPORT step.
* Using PAGE;
title1 'Refining REPORT Appearance';
title2 'Using ID with PAGE';
proc report data=rptdata.clinics
where=(region in('1','2')))
nowd;
columns region clinname ht wt;
define region
/ group
width=6;
define clinname / group
id;
define ht
/ analysis mean;
define wt
/ analysis mean page;
run;
The use of the PAGE option forces WT to appear on a different page than HT.
Refining REPORT Appearance
Using ID with PAGE
region
1
2
clinic name
Boston National Medical
Vermont Treatment Center
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
Refining REPORT Appearance
Using ID with PAGE
region
1
2
clinic name
Boston National Medical
Vermont Treatment Center
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
1
height
in inches
74
74
64
63.5
64
2
weight
in pounds
195
195
115
107
105
The report has
five detail rows.
WT does not
appear on the
first page.
The same five
detail rows are
reprinted with
WT rather than
HT. The same
values of
REGION and
CLINNAME
from the first
page are
repeated.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 121
6.2 Using Variable Aliases
In the example in Section 6.1.6, six statistics are applied to each of the two analysis variables.
Here are the PROC and COLUMN statements:
proc report data=rptdata.clinics(where=(region in('1','2')))
nowd;
columns region clinname (ht wt),(min max n mean median);
Although nesting the statistics this way works fine, we have some limitations on the use of
DEFINE statements. Only one DEFINE statement can be applied to each of the analysis variables,
and the DEFINE attributes are applied to all of the statistics associated with that analysis variable.
This gives us only minimal control over the appearance of the individual columns associated with
each of the statistics. We can gain some control by associating additional DEFINE statements
with the statistics themselves (see Section 2.4.5). However, we can still have a problem, because
in effect we are attempting to use each analysis variable in multiple ways.
Fortunately, when we need to use one variable in more than one way, we can create an alias for
that variable. This alias can then have its own unique DEFINE statement. An alias is created in
the COLUMN statement by following a table variable with an equal sign and a valid variable
name that does not otherwise appear on the COLUMN statement. The syntax to create an alias of
HT might be as follows:
ht=htalias
As a result, the report items HT and HTALIAS would both be available for use on DEFINE
statements.
In the following example, we would like to create the same series of statistics for HT as were
specified in the preceding COLUMN statement, and we want to control the appearance of these
columns independently.
* Using Aliases;
title1 'Refining REPORT Appearance';
title2 'Using a Column Alias';
proc report data=rptdata.clinics
(where=(region in('1','2','3')))
nowd;
columns region ht ht=htmin ht=htmax ht=htmean ht=htmedian;
define region
/ group width=6;
define ht
/ analysis n
format=2. 'N';
define htmin
/ analysis min
format=4.1 'Min';
define htmax
/ analysis max
format=4.1 'Max';
define htmean
/ analysis mean
format=4.1 'Mean';
define htmedian / analysis median format=6.1 'Median';
run;
122 Carpenter’s Complete Guide to the SAS REPORT Procedure
The independent DEFINE statements enable us to tailor the characteristics of each of the columns.
Refining REPORT Appearance
Using a Column Alias
region
1
2
3
N
4
10
10
Min
74.0
63.0
64.0
Max
74.0
64.0
74.0
Mean
74.0
63.8
68.2
Median
74.0
64.0
67.0
Although only one analysis
variable is specified, aliases are
declared so that each column can
have its own DEFINE statement.
6.3 Nesting Variables
In several earlier examples (see Sections 2.3.2 and 6.1.6) statistics have been attached with a
comma to the analysis variables to which they are to be applied. Essentially the statistics have
been nested within the analysis variable. It is also possible to nest an analysis variable within an
ACROSS variable.
In the following example, SEX is defined as an ACROSS variable, and within each value of SEX
the mean height and weight is displayed. When no variable or statistic is nested within an
ACROSS variable, the number of observations is displayed (see Section 6.1.3). However, by
nesting an ANALYSIS variable under the ACROSS variable, we can explicitly specify the
statistic on the DEFINE statement. This allows us to choose what is to be displayed.
* Nesting Analysis Variables;
title1 'Refining REPORT Appearance';
title2 'Nesting Mean Weight and Height within Sex';
* Nesting variables;
proc report data=rptdata.clinics nowd;
column region sex,(wt=n wt ht); n
define region / group
width=6;
define sex
/ across
format=$2. 'Gender';
define n
/ analysis n
format=2.0 'N'; o
define wt
/ analysis mean format=6.2 'Weight';
define ht
/ analysis mean format=6.1 'Height';
run;
n The alias N is declared for WT. Two different statistics (N and MEAN) can now be calculated
for WT.
o A DEFINE statement appears for N, the alias of WT. Because of possible programmer
confusion with the N statistic, this alias name might not be the best of choices (WT_N might
be better). The PROC REPORT step, however, will not be confused.
Inspection of the table shows that for each value of SEX, there are three columns.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 123
Refining REPORT Appearance
Nesting Mean Weight and Height within Sex
Gender
region
1
10
2
3
4
5
6
7
8
9
N
.
2
6
5
4
5
4
.
4
2
F
Weight
.
163.00
109.67
127.80
143.00
146.20
187.00
.
160.00
177.00
Height
.
63.0
63.7
65.6
66.5
63.2
63.0
.
70.0
65.0
N
4
4
4
5
10
3
6
4
.
8
M
Weight
195.00
177.00
105.00
163.80
165.60
177.00
205.33
151.00
.
190.50
Height
74.0
69.0
64.0
70.8
70.0
70.7
69.0
66.0
.
68.0
In this table, the
value of N actually
counts the number of
nonmissing values of
WT. The count for
nonmissing values of
HT is not displayed
and could potentially
be different.
6.4 Taking Full Advantage of Formats
The use of formats can be very important to the PROC REPORT programmer. Both predefined
and user-defined formats can be used in a variety of ways in order to customize the appearance of
the report.
SEE ALSO
Extensive discussion of user-defined formats with numerous examples is provided by Chapman
(2003). Examples can also be found in Burlew (2005, e.g., p. 63).
6.4.1 User-Defined Formats
You can define formats that you can then use to further enhance the appearance of your report.
Like formats supplied with SAS, these user-defined formats are also designated either on the
DEFINE statement or through the use of the FORMAT statement.
In the following example, user-defined formats are used to solve two problems that have appeared
in a number of the previous examples. First, because REGION is a character variable, region ‘10’
sorts before region ‘2’ and second, the values of SEX (‘F’ and ‘M’) are a bit too terse. The $REG.
format places a blank space in front of each of the single digit regions, and $GENDER. maps ‘M’
and ‘F’ to ‘Male’ and ‘Female’ respectively.
These formats are generated using PROC FORMAT.
* Using user defined formats;
proc format;
value $reg '1'=' 1' '2'=' 2' '3'=' 3' '4'=' 4'
'5'=' 5' '6'=' 6' '7'=' 7' '8'=' 8'
'9'=' 9' '10'='10';
value $gender
'M'='Male' 'F'='Female';
run;
124 Carpenter’s Complete Guide to the SAS REPORT Procedure
* User Defined Formats;
title1 'Refining REPORT Appearance';
title2 'Using User Defined Formats';
proc report data=rptdata.clinics nofs;
column region sex,(wt=n wt ht);
define region / group
format=$reg6.;
define sex
/ across
format=$Gender.
define n
/ analysis n
format=2.0
define wt
/ analysis mean format=6.2
define ht
/ analysis mean format=6.1
run;
'Gender';
'N';
'Weight';
'Height';
Notice that the order of the regions has now changed. Unless the programmer specifies otherwise,
when a format is used on the DEFINE statement, there is an implied ORDER=FORMATTED
option.
Refining REPORT Appearance
Using User Defined Formats
region
1
2
3
4
5
6
7
8
9
10
N
.
6
5
4
5
4
.
4
2
2
Gender
Female
Weight Height
N
.
.
4
109.67
63.7
4
127.80
65.6
5
143.00
66.5 10
146.20
63.2
3
187.00
63.0
6
.
.
4
160.00
70.0
.
177.00
65.0
8
163.00
63.0
4
Male
Weight Height
195.00
74.0
105.00
64.0
163.80
70.8
165.60
70.0
177.00
70.7
205.33
69.0
151.00
66.0
.
.
190.50
68.0
177.00
69.0
SEX is an
ACROSS
variable with
three columns
under each
value.
In the LISTING destination, we can make the groups of columns that are nested under each value
of SEX more distinct by making a slight further modification to the $GENDER. format. Repeat
characters (see Section 4.3.1) are correctly applied to headers, even when they are used in
formatted values. Underscores work well as repeat characters in headers associated with ACROSS
variables. The VALUE statement in the PROC FORMAT step becomes the following:
value $gender
'M'='_Male_' 'F'='_Female_';
run;
Chapter 6: Refining Our Understanding of the PROC REPORT Step 125
Refining REPORT Appearance
Using User Defined Formats With Repeat Characters
region
1
2
3
4
5
6
7
8
9
10
Gender
______Female______ _______Male_______
N Weight Height
N Weight Height
.
.
.
4 195.00
74.0
6 109.67
63.7
4 105.00
64.0
5 127.80
65.6
5 163.80
70.8
4 143.00
66.5 10 165.60
70.0
5 146.20
63.2
3 177.00
70.7
4 187.00
63.0
6 205.33
69.0
.
.
.
4 151.00
66.0
4 160.00
70.0
.
.
.
2 177.00
65.0
8 190.50
68.0
2 163.00
63.0
4 177.00
69.0
The columns
associated with the
values of the
ACROSS variable
are more clearly
grouped.
In the following example, a formatted value is used along with the SPACING= option to
effectively concatenate two columns.
proc format; n
value $SYM
'01' = ':Sleepiness'
'02' = ':Coughing'
'03' = ':Limping '
'04' = ':Bleeding'
'05' = ':Weak'
'06' = ':Nausea'
'07' = ':Headache'
'08' = ':Cramps'
'09' = ':Spasms '
'10' = ':Shortness of Breath';
run;
* Define Statement spacing option;
title1 'Using Proc REPORT';
title2 'Define Statement SPACING Option';
title3 'Removing Spaces';
proc report data=rptdata.clinics nowd;
column ('Name' lname fname) symp symp=sympname; Y
define lname
/ order
'Last';
define fname
/ order
'First';
define symp
/ display format=$2. 'Sy'; Z
define sympname / display format=$sym. q
spacing=0
'mptom'; r
run;
n A user-defined format is created that associates symptom codes with the symptom label.
Y The variable SYMP is also assigned an alias, SYMPNAME.
Z The first instance of the variable SYMP is given just enough space ($2). Notice that the label
starts the word “Symptom.”
126 Carpenter’s Complete Guide to the SAS REPORT Procedure
q The $SYM. format is applied to the alias SYMPNAME, which is the second instance of the
variable SYMP.
r With SPACING= set to 0, there are no spaces between this column and the one to the left
(SYMP). The label finishes the word “Symptom.”
Using Proc REPORT
Define Statement SPACING Option
Removing Spaces
Name
Last
Masters
Maxim
Maxwell
Mercy
Moon
Most
Nabers
Nolan
Olsen
First
Martha
Kurt
Linda
Ronald
Rachel
Mat
David
Terrie
June
Symptom
02:Coughing
06:Nausea
04:Bleeding
10:Shortness of Breath
02:Coughing
04:Bleeding
04:Bleeding
10:Shortness of Breath
The column labeled
“Symptom” is actually
two columns with no
space (SPACING=0)
between them.
. . . Portions of the table are not shown . . .
The reader of the report sees only three apparent columns. The SPACING= option has little
utility outside of the LISTING destination. Creating a computed variable with a concatenated
value is generally a more robust solution, and one which works for destinations other than
LISTING (see Section 7.6.2 for an example that creates a character variable).
MORE INFORMATION
A user-defined format is used to form groups in the example in Section 7.4. Formats are used to
perform traffic lighting in Section 8.4.2 and to form links to other files in Section 8.5.8.
SEE ALSO
A tutorial on building and using user-defined formats is provided by Carpenter (2004b). Chapman
(2003) and DeAngelis (2005) both create and use several user-defined formats. Levin (2005)
creates user-defined formats to use with the STYLE= option. Izard and Chen (2005) also use the
SPACING= option to concatenate two columns.
6.4.2 Preloading Formats
When a level of a classification or grouping variable (define type GROUP, ORDER, or ACROSS)
is not included in the data, the associated row or column does not appear in the table. Even though
we might not always know ahead of time which values are not in the data, we might still want to
have all of the potential levels in the report.
One solution is to use a DATA step to build observations into the incoming data table. However,
through the process of preloading formats, we can accomplish the same thing without using the
DATA step.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 127
The options associated with this process are a bit confusing in that they work together, and it is
the combination of options that determines the ultimate effect. Some of these options appear only
on the PROC statement and others only on the DEFINE statement.
The following options are used with the DEFINE statement:
PRELOADFMT
EXCLUSIVE
loads the format levels prior to execution. This option is always present
when any of the others are also used.
specifies that only data levels that are included in the format definition are to
appear in the table.
The following PROC statement options work in combination with the DEFINE statement options:
COMPLETEROWS specifies that all rows representing format levels are to appear in the report.
COMPLETECOLS specifies that all columns representing format levels are to appear in the report.
Through the use of the DEFINE statement PRELOADFMT option and a format, we can cause
each individual level of a classification variable to appear on the table, regardless of whether it
exists in the data. The EXCLUSIVE option enables you to exclude those levels not on the format,
regardless of whether they are in the data table.
The examples in this section demonstrate the various combinations of options associated with the
PRELOADFMT option by referencing the following two user-defined formats. Each format
contains a level that is not in the data, and the $REGX. format specifies only some of the values
that do exist in the data.
* Using PRELOADFMT with user defined formats;
proc format;
value $regx '1'=' 1' '2'=' 2' 'X'=' X' ;
value $genderu
'M'='Male' 'F'='Female' 'U'='Unknown';
run;
If the PRELOADFMT option appears on the DEFINE statement, the documentation states that it
must be accompanied by a format specification and either an ORDER= DATA or the
EXCLUSIVE option. However, at least for some operating systems and versions of SAS, the
specification of PRELOADFMT along with a FORMAT= option does not necessarily also require
the ORDER= and EXCLUSIVE options.
In the following example, the EXCLUSIVE option is used to remove regions that are not on the
format.
* User Defined Formats;
title1 'Refining REPORT Appearance';
title2 'Using PRELOADFMT with EXCLUSIVE';
proc report data=rptdata.clinics nowd;
column region sex,(wt=n wt);
define region / group
format=$regx6.
preloadfmt exclusive;
define sex
/ across
format=$Genderu. 'Gender';
define n
/ analysis n
format=2.0 'N';
define wt
/ analysis mean format=6.2 'Weight';
run;
128 Carpenter’s Complete Guide to the SAS REPORT Procedure
Notice that, although there is no WHERE clause to eliminate data, only regions 1 and 2 appear on
the table.
Refining REPORT Appearance
Using PRELOADFMT with EXCLUSIVE
region
1
2
Gender
Female
Male
N Weight
N Weight
.
.
4 195.00
6 109.67
4 105.00
Since the format $REGX. contains
definitions only for regions 1 and 2 (X
isn’t in the data), the EXCLUSIVE
option eliminates all of the other
regions.
Simply replacing the EXCLUSIVE option with the ORDER= option (the PRELOADFMT option
usually expects either the EXCLUSIVE or the ORDER= options) does not cause the region X to
appear in the report. However, if we also use the PROC statement option COMPLETEROWS, we
get all the levels of the format (including region X, which has no data), as well as all of the levels
of REGION actually occurring in the data.
* Using PRELOADFMT;
title1 'Refining REPORT Appearance';
title2 'Using PRELOADFMT with';
title3 'ORDER= and COMPLETEROWS';
proc report data=rptdata.clinics
nowd
completerows;
column region sex,(wt=n wt);
define region / group
format=$regx6.
preloadfmt
order=data;
define sex
/ across
format=$Genderu. 'Gender';
define n
/ analysis n
format=2.0 'N';
define wt
/ analysis mean format=6.2 'Weight';
run;
Now, even though there is no data for region X, region X appears on the table.
Refining REPORT Appearance
Using PRELOADFMT with
ORDER= and COMPLETEROWS
region
1
2
X
3
7
10
9
4
5
6
8
Gender
Female
Male
N Weight
N Weight
0
.
4 195.00
6 109.67
4 105.00
0
.
0
.
5 127.80
5 163.80
0
.
4 151.00
2 163.00
4 177.00
2 177.00
8 190.50
4 143.00 10 165.60
5 146.20
3 177.00
4 187.00
6 205.33
4 160.00
0
.
Notice the values of REGION. The first
three rows are formatted (using
$REGX., which inserts a leading blank).
However, the remainder are
unformatted and in the same order as
the regions that first appear in the data
(ORDER=DATA).
Chapter 6: Refining Our Understanding of the PROC REPORT Step 129
As one might anticipate, using COMPLETEROWS along with EXCLUSIVE limits the table to
the levels of the format, but also includes any format levels not found in the data.
* Using PRELOADFMT;
title1 'Refining REPORT Appearance';
title2 'Using PRELOADFMT with EXCLUSIVE and COMPLETEROWS';
proc report data=rptdata.clinics
nowd completerows;
column region sex,(wt=n wt);
define region / group
format=$regx6.
preloadfmt exclusive;
define sex
/ across
format=$Genderu.
'Gender';
define n
/ analysis n
format=2.0
'N';
define wt
/ analysis mean format=6.2
'Weight';
run;
Remember that, generally, you should use either the EXCLUSIVE or ORDER= option along with
PRELOADFMT.
Refining REPORT Appearance
Using PRELOADFMT with EXCLUSIVE and COMPLETEROWS
Gender
region
1
2
X
Female
Weight
.
109.67
.
N
0
6
0
N
4
4
0
Male
Weight
195.00
105.00
.
EXCLUSIVE limits the
regions to 1 and 2,
whereas
COMPLETEROWS
brings in region X.
The previous examples in this section have to do with the control of rows. Since we are also
defining an ACROSS variable, we can preload formats for that variable as well.
* Using PRELOADFMT;
title1 'Refining REPORT Appearance';
title2 'Using PRELOADFMT with EXCLUSIVE';
title3 'as well as COMPLETEROWS and COMPLETECOLS';
proc report data=rptdata.clinics
nowd
completerows completecols;
column region sex,(wt=n wt);
define region / group
format=$regx6. preloadfmt exclusive;
define sex
/ across
format=$Genderu. 'Gender'
preloadfmt;
define n
/ analysis n
format=2.0 'N';
define wt
/ analysis mean
format=6.2 'Weight';
run;
Since the format $GENDERU. includes a level for unknown values of SEX, and since we have
included the COMPLETECOLS option on the PROC statement, that level now appears in the
table even though there are no observations with that value of SEX.
130 Carpenter’s Complete Guide to the SAS REPORT Procedure
Refining REPORT Appearance
Using PRELOADFMT with EXCLUSIVE
as well as COMPLETEROWS and COMPLETECOLS
region
1
2
X
Female
N Weight
0
.
6 109.67
0
.
Gender
Male
N Weight
4 195.00
4 105.00
0
.
Unknown
N Weight
0
.
0
.
0
.
The COMPLETEROWS and
COMPLETECOLS options
have caused the preloaded
formats to include both a row
and a column that do not
appear in the data.
In this example, PRELOADFMT is used without either an ORDER= or EXCLUSIVE option on
the DEFINE statement for SEX.
6.4.3 Order Based on Format Definition
Generally you will want to order the rows or columns of your report by using one of the four
primary values associated with the ORDER= option (see Section 2.4.3). Sometimes, however, you
might want to specify an order that does not fit any one of these types. A traditional approach has
been to create a nonprinted variable that can be used to force the desired order. This was done in
the first example in Section 6.4.1. Although effective, this technique can be a bit cumbersome. In
recent versions of SAS we have a better alternative.
Normally when a user-defined format is created, the format is internally placed into sorted order.
Thus it does not particularly matter what order the value/label pairings are specified in the value
statement. However, this reordering can be prevented through the use of the NOTSORTED option
on the VALUE statement. When this option is applied, the internal order of the format remains as
it is defined. In the following PROC FORMAT, the format $SYM is defined. However, the
researcher has placed the rows in the order that is to be used in the report.
proc format;
value $SYM (notsorted)
'01' = 'Sleepiness'
'02' = 'Coughing'
'10' = 'Shortness of Breath'
'05' = 'Weak'
'03' = 'Limping '
'07' = 'Headache'
'06' = 'Nausea'
'08' = 'Cramps'
'09' = 'Spasms '
'04' = 'Bleeding';
run;
When the NOTSORTED option is used on the VALUE statement, the order in which the item
pairs are defined in the VALUE statement is preserved.
title1 'Using Proc REPORT';
title2 'Using the Format Definition Order';
proc report data=rptdata.clinics
nowd completerows;
Chapter 6: Refining Our Understanding of the PROC REPORT Step 131
column symp n;
define symp / group
preloadfmt order=data
format=$sym. 'Symptom';
define n
/ 'N' ;
run;
We can take advantage of the nature of the format through the use of the PRELOADFMT and
ORDER=DATA options. Normally, the ORDER=DATA option would cause the table rows to be
data dependent. However, when it is used in conjunction with PRELOADFMT on a format that
was built using the NOTSORTED option, the rows are instead arranged according to the order of
the format definition. This is shown in the resulting table.
Using Proc REPORT
Using the Format Definition Order
Symptom
Sleepiness
Coughing
Shortness of Breath
Weak
Limping
Headache
Nausea
Cramps
Spasms
Bleeding
N
4
10
14
8
4
0
12
0
2
14
The order of the rows in the report are
now based on the order of the items in
the format definition.
Notice that, since the COMPLETEROWS option (see Section 6.4.2) is included on the PROC
statement, even symptoms with no records in the data set are included in the table.
For very large formats, there might be some performance access issues when you use the
NOTSORTED option.
SEE ALSO
In a paper not strictly directed at PROC REPORT, Stewart and Fecht (2002) give an example of
the use of these options.
6.5 Other PROC Statement Options
Most of the primary PROC REPORT statement options are detailed in other sections of this book.
A majority of these are introduced in Section 4.4. This section discusses some of the other, often
less commonly used, options associated with the PROC REPORT statement.
6.5.1 Removing Headers
Generally the ability to define and control the header text is a major advantage of the REPORT
procedure. However, for one reason or another you might sometimes want to turn off these
headers. You can use the NOHEADER option to suppress all of the text that appears in the header
section. This includes text specified in the COLUMN statement and on the DEFINE statement.
132 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Removing headers;
title1 'Refining REPORT Appearance';
title2 'Using NOHEADER';
proc report data=rptdata.clinics(where=(region in('1','2')))
noheader nowd;
columns region n ('Mean' ht wt);
define region / group width=6;
define n
/ 'N';
define ht
/ analysis mean
format=6.2 'Height';
define wt
/ analysis mean
format=6.2 'Weight';
run;
The NOHEADER option suppresses the header text, resulting in the following table:
Refining REPORT Appearance
Using NOHEADER
1
2
4
10
74.00
63.80
195.00
107.80
Obviously for a report of this type, turning off the headers would be an odd thing to do. However,
when the NOHEADER option is coupled with the options in the next two sections, advantages
emerge.
6.5.2 Using NAMED Output
The NAMED option writes the table in named format. This is similar to the NAMED input style
that can be used to read in data in the DATA step and the NAMED output style that can be used
on the PUT statement. In the report, the variable name is followed by an equals sign (=), which is
followed by the value itself.
Unless you have multiple columns that you want to stack, it is unlikely that you will find many
uses for this option. However, when coupled with the NOHEADER option in the LISTING
destination, it can have some value. The following example uses the same data as the example in
Section 4.4.8, and here we are using the NOHEADER and WRAP options along with the
NAMED option. The WRAP option (Section 4.4.8) applies only to the LISTING destination.
ods listing;
ods pdf file="&path\results\ch6_5_2.pdf";
* Using the WRAP option;
title1 'Refining REPORT Appearance';
title2 'Using WRAP with NOHEADER and NAMED';
data demog;
input name $ sex $ idnum score1-score10 / comment $65.;
format name $5. sex $3. idnum 5. score1-score10 6.
comment $65.;
datalines;
Russ M 123 1 9 3 8 4 7 5 8 6 3
Occasionally has difficulty with verbal communication
Kevin M 456 4 7 5 6 8 5 4 3 2 3
Chapter 6: Refining Our Understanding of the PROC REPORT Step 133
Is very particular about the placement of personal objects
Paige F 789 6 7 4 3 5 8 9 2 3 4
Gets very excited by the success of those people close to her
run;
proc report data=demog ls=80 nowd
noheader wrap named;
column name comment;
define name
/ group;
define comment / display;
break after name / skip;
run;
The NOHEADER option on the PROC statement removes the column headers, whereas the
WRAP option only affects the LISTING destination.
Refining REPORT Appearance
Using WRAP with NOHEADER and NAMED
name=Kevin
comment=Is very particular about the placement of personal objects
name=Paige
comment=Gets very excited by the success of those people close to her
name=Russ
comment=Occasionally has difficulty with verbal communication
Writing this table to the PDF destination results in the following:
Although this technique has had limited utility in the reports that I have generated, it is presented
here for the sake of completeness.
MORE INFORMATION
Section 4.5.3 discusses the DEFINE statement FLOW option.
SEE ALSO
Sample 637 discusses the wrapping of observations that are too long to fit on a report line. SAS
Technical Report P-258 (1993, pp. 183, 225) and SAS Guide to the REPORT Procedure,
Reference, Release 6.1 (1995, pp. 61–62) both introduce and use the NAMED option.
134 Carpenter’s Complete Guide to the SAS REPORT Procedure
6.5.3 Debugging with the LIST Option
The LIST option can be used to expand the PROC REPORT step in the SAS log. The expansion
includes the statements and options that have been left at defaults. The following simple PROC
REPORT step has three variables on the COLUMN statement, but only two DEFINE statements.
The output is directed only to the LISTING destination, and the programmer might want to know
which options are being applied as defaults.
title1 'Using Proc REPORT';
title2 'Using the LIST Option';
proc report data=rptdata.clinics
list nowd;
column region ht wt;
define region / group;
define ht
/ analysis mean 'HEIGHT';
run;
Since the LIST option has been specified, the SAS log includes the following expanded PROC
REPORT step code (the line breaks and code alignments are mine). The expanded code includes a
DEFINE statement for WT.
PROC REPORT DATA=RPTDATA.CLINICS
LS=126 PS=41 SPLIT="/" NOCENTER ;
COLUMN ( region ht wt );
DEFINE
DEFINE
DEFINE
region / GROUP FORMAT= $2.
WIDTH=2 SPACING=2 LEFT "region" ;
ht / MEAN FORMAT= BEST9.
WIDTH=9 SPACING=2 RIGHT "HEIGHT" ;
wt / SUM FORMAT= BEST9.
WIDTH=9 SPACING=2 RIGHT "weight in pounds" ;
RUN;
SEE ALSO
The LIST option is mentioned in SAS Technical Report P-258 (1993, pp. 182–183).
6.5.4 Including MISSING Classification Levels
By default, missing values for classification variables are not displayed in SAS programs. In
Section 6.1.3, the MISSING option is used on the DEFINE statement to designate that a missing
value is a valid level of the classification variable. You can also place the MISSING option on the
PROC statement to designate missing values as valid classification levels for all classification
variables.
In the following example, a few missing values have been artificially created for REGION and
CLINNAME in the RPTDATA.CLINICS data set, and written to WORK.REGMISS. Without a
MISSING option on either the DEFINE statement (see Section 6.1.3) or on the PROC statement,
missing values of these grouping variables are ignored.
* Using MISSING on the PROC statement;
title1 'Refining REPORT Appearance';
title2 'Artificial Missing Values';
title3 'Without the MISSING Option';
proc report data=regmiss(where=(region<'4'))
nowd;
Chapter 6: Refining Our Understanding of the PROC REPORT Step 135
column
define
define
define
define
run;
region clinname ('Height' ht ht=htmean);
region
/ group format=$6.;
clinname / group;
ht
/ analysis N
format=6.1 'N';
htmean
/ analysis Mean format=6.1 'Mean';
Refining REPORT Appearance
Artificial Missing Values
Without the MISSING Option
region
1
10
2
3
clinic name
Boston National Medical
Vermont Treatment Center
Portland General
Seattle Medical Complex
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
Bethesda Pioneer Hospital
Naval Memorial Hospital
Philadelphia Hospital
Height
N
Mean
2.0
74.0
2.0
74.0
2.0
69.0
4.0
66.0
2.0
64.0
3.0
63.7
3.0
64.0
4.0
72.5
4.0
65.5
2.0
65.0
Inclusion of the MISSING option on the PROC statement has the same effect as its inclusion on
the DEFINE statements for both REGION and CLINNAME. Because the MISSING option is on
the PROC statement, a missing value for REGION or CLINNAME is now a valid classification
level.
title1 'Refining REPORT Appearance';
title2 'Artificial Missing Values';
title3 'With the MISSING Option';
proc report data=regmiss(where=(region<'4'))
nowd missing;
column region clinname ('Height' ht ht=htmean);
define region
/ group format=$6.;
define clinname / group;
define ht
/ analysis N
format=6.1 'N';
define htmean
/ analysis Mean format=6.1 'Mean';
run;
The MISSING option has forced the inclusion of six rows that would have otherwise not been
written to the final report.
136 Carpenter’s Complete Guide to the SAS REPORT Procedure
Refining REPORT Appearance
Artificial Missing Values
With the MISSING Option
region
1
10
clinic name
Atlanta General Hospital
Dallas Memorial Hospital
Houston General
San Diego Memorial Hospital
Boston National Medical
Vermont Treatment Center
Portland General
Seattle Medical Complex
2
3
Geneva Memorial Hospital
New York General Hospital
New York Metro Medical Ctr
Bethesda Pioneer Hospital
Naval Memorial Hospital
Philadelphia Hospital
Height
N
Mean
1.0
70.0
2.0
66.0
2.0
63.0
2.0
69.0
1.0
70.0
2.0
74.0
2.0
74.0
2.0
69.0
4.0
66.0
2.0
63.5
2.0
64.0
3.0
63.7
3.0
64.0
4.0
72.5
4.0
65.5
2.0
65.0
As a general rule, I prefer to use the MISSING option on the DEFINE statement, as that location
offers more control and flexibility.
MORE INFORMATION
Missing classification levels can also be included through the use of the MISSING option on the
DEFINE statement (see Section 6.1.3).
SEE ALSO
SAS Technical Report P-258 (1993, pp. 226–227) has an example that uses the MISSING option.
6.6 BY-Group Processing
The usage of the BY statement with PROC REPORT is similar to its usage in other procedures.
The similarity extends to BY statement options, such as DESCENDING and NOTSORTED,
which are also supported and also behave the same as in other reporting procedures.
The BY statement can be used to form groups much as the GROUP option is used on the DEFINE
statement. The primary difference between the two is in the appearance of the report. By default,
the REPORT procedure creates a new page for each BY group.
MORE INFORMATION
Section 6.8 contains an additional brief discussion of paging issues.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 191–192) introduces the use of the BY statement.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 137
6.6.1 Using the BY Statement
As is the case whenever the BY statement is used (except with PROC SORT of course), the data
is expected to be in the order specified by the BY statement variables, or else the NOTSORTED
option must be present.
* Using the BY statement;
proc sort data=rptdata.clinics
out=clinics;
by region;
run;
title1 'Refining REPORT Appearance';
title2 'Using the BY Statement';
proc report data=clinics nowd ;
by region;
columns clinname ht wt;
define clinname / group;
define ht
/ analysis mean;
define wt
/ analysis mean;
run;
When the BY statement is used in the LISTING, RTF, or PDF destinations, a separate page is
created for each value of REGION.
Refining REPORT Appearance
Using the BY Statement
region=1
clinic name
Boston National Medical
Vermont Treatment Center
height
in inches
74
74
weight
in pounds
195
195
height
in inches
69
66
weight
in pounds
177
170
Refining REPORT Appearance
Using the BY Statement
region=10
clinic name
Portland General
Seattle Medical Complex
. . . Remaining regions are not shown . . .
138 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
The example in Section 8.6.2 uses the BY statement to build a table with separate pages for each
value of the BY statement, and includes the use of the TITLE statement option #BYVAL and the
NOBYLINE system option. These TITLE options are also discussed in Section 6.6.3.
6.6.2 Creating Breaks with BY Groups
Because the BY variable(s) are neither GROUP nor ORDER variables, you cannot use them in a
BREAK statement. Fortunately, the RBREAK statement does work with the BY statement, and it
creates a separate summary for each level of the BY variables.
* Using the BY statement;
proc sort data=rptdata.clinics
out=clinics;
by region;
run;
title1 'Refining REPORT Appearance';
title2 'Using the BY Statement with RBREAK';
proc report data=clinics nowd;
by region;
columns clinname ht wt;
define clinname / group;
define ht
/ analysis mean;
define wt
/ analysis mean;
rbreak after / dol summarize;
run;
When the BY statement is used, the RBREAK statement generates a summary line after each
page (BY variable combination), rather than once at the end of the report.
Refining REPORT Appearance
Using the BY Statement with RBREAK
region=1
clinic name
Boston National Medical
Vermont Treatment Center
height
in inches
74
74
=========
74
weight
in pounds
195
195
=========
195
Chapter 6: Refining Our Understanding of the PROC REPORT Step 139
Refining REPORT Appearance
Using the BY Statement with RBREAK
region=10
clinic name
Portland General
Seattle Medical Complex
height
in inches
69
66
=========
67
weight
in pounds
177
170
=========
172.33333
. . . Remaining regions are not shown . . .
Because of the BY statement, the RBREAK summary now appears for each BY level. There is no
longer a summary across the entire report. When you need the summary for each region and a
summary across the entire report, using the BY statement would not be your best approach.
Instead, REGION could be the highest level of group variable, and the _PAGE_ option (see
Section 5.4) could be used to create the page breaks for each region.
6.6.3 Using the #BYVAL and #BYVAR Options
Whenever the BY statement has been specified, the #BYVAR and #BYVAL TITLE and
FOOTNOTE options are available for use with the REPORT procedure.
th
#BYVARn is replaced with the name of the n BY variable in the BY list.
th
#BYVALn is replaced by the value of the n BY variable in the BY list.
These options are not at all limited to PROC REPORT steps, and can be widely used in SAS
programming.
* Using the BY statement;
proc sort data=rptdata.clinics
out=clinics;
by region;
run;
options nobyline; n
title1 'Refining REPORT Appearance';
title2 'Using the BY Statement with TITLE Options';
title3 '#byvar1 is #byval1'; o
proc report data=clinics nowd;
by region;
columns clinname ht wt;
define clinname / group;
define ht
/ analysis mean;
define wt
/ analysis mean;
run;
140 Carpenter’s Complete Guide to the SAS REPORT Procedure
n Because the value of the BY variable is in the title, the BYLINE option is no longer needed
and is suppressed by using the NOBYLINE system option.
Y In the resulting reports, #BYVAR1 is replaced by the name of the first variable in the list of
BY variables. In this case, there is only one BY variable, REGION. Similar action is taken for
#BYVAL1, except the value of the first BY variable replaces the option.
Refining REPORT Appearance
Using the BY Statement with TITLE Options
region is 1
clinic name
Boston National Medical
Vermont Treatment Center
height
in inches
74
74
weight
in pounds
195
195
Refining REPORT Appearance
Using the BY Statement with TITLE Options
region is 10
clinic name
Portland General
Seattle Medical Complex
height
in inches
69
66
weight
in pounds
177
170
. . . Remaining regions are not shown . . .
As an alternative to using the number of the BY variable, the name of the variable can also be
used by including it in parentheses. TITLE3 in the previous program becomes the following:
title3 '#byvar(region) is #byval(region)';
The #BYLINE option places the entire BYLINE into the title. It appears in the same form as the
BYLINE would otherwise appear in the body of the report. For the BY statement
by region sex;
a TITLE statement such as this one
title4 '#BYLINE';
would produce a title that might look like this:
region=9 patient sex=M
Chapter 6: Refining Our Understanding of the PROC REPORT Step 141
MORE INFORMATION
The #BYVAL option is used in Section 8.6.3 in an example that also puts page numbers in the
title.
SEE ALSO
Hamilton (2004) includes a discussion of the use of these options in a title. They are briefly
mentioned in SAS Technical Report P-258 (1993, p. 96) and are more fully described in the
documentation for Base SAS. Burlew (2005, pp. 20–25) uses these options in a PROC PRINT
example, and in a PROC REPORT example on page 149.
6.6.4 BY Groups and the Output Delivery System
In Section 6.6.1 it was stated that each level of the BY variables causes a new page to be
generated. This behavior is strictly true for the LISTING destination; however, the definition of
what constitutes a page changes for other ODS destinations.
If we rerun the example in Section 6.6.1 and include another destination, such as HTML, we see
that although the individual regions appear on separate reports, they are all on the same HTML
file and on the same virtual page.
ods html file="&path\results\ch6_6_4a.html";
* Using the BY statement;
proc sort data=rptdata.clinics
out=clinics;
by region;
run;
title1 'Refining REPORT Appearance';
title2 'Using the BY Statement';
title3 'Paging in HTML';
proc report data=clinics nowd;
by region;
columns clinname ht wt;
define clinname / group;
define ht
/ analysis mean;
define wt
/ analysis mean;
run;
ods html close;
142 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here are the first two regions shown in the HTML file:
You can also create separate files by using the NEWFILE= option on the ODS statement. This
option can take on the following values:
NONE
PROC
OUTPUT
PAGE
BYGROUP
(the default) results in a single page, even across PROC step boundaries.
creates a new file at each PROC step boundary.
writes each table for the OUTPUT destination into a separate table.
generates a new file each time a page is explicitly generated.
creates a new file for each level of the BY variables.
In the following example, the NEWFILE= option is set to BYGROUP, which will force a new file
for each level of the BY variables. When multiple files are created, a number will be appended to
the name of the file for each new file. If the name already contains a number or numbers, the
right-most number will be incremented. In this example, the first file is named
ch6_6_4b0.html, successive values of regions will be written into ch6_6_4b1.html,
ch6_6_4b2.html, etc.
ods html file="&path\results\ch6_6_4b0.html"
newfile=bygroup;
* Using the BY statement;
proc sort data=rptdata.clinics
out=clinics;
by region;
run;
title1 'Refining REPORT Appearance';
title2 'Using the BY Statement';
title3 'Paging in HTML with NEWFILE=BYGROUP';
Chapter 6: Refining Our Understanding of the PROC REPORT Step 143
proc report data=clinics nowd;
by region;
columns clinname ht wt;
define clinname / group;
define ht / analysis mean;
define wt / analysis mean;
run;
ods html close;
The file ch6_6_4b0.html contains the following:
The file ch6_6_4b1.html contains the following:
MORE INFORMATION
Section 6.8 contains an additional brief discussion of paging issues.
144 Carpenter’s Complete Guide to the SAS REPORT Procedure
6.7 Calculations Using the FREQ Statement
The FREQ statement enables you to specify a variable to be used when generating statistics based
on collapsed observations. This statement is essentially the same in PROC REPORT as it is in a
number of other procedures that make statistical calculations.
In the data table RPTDATA.CLINICS, there is one observation per patient. If we had received a
summary table that had been rolled up so that it had discrete values for HT, we would still be able
to calculate statistics for HT as long as we knew how many patients had each individual height.
When this frequency is stored in a variable, we can point to it with the FREQ statement.
This capability is demonstrated in the following example by first summarizing the data and then
calculating the mean and variance in PROC REPORT using only the summarized data. A PROC
MEANS can be used to count the number of observations with each distinct value of HT. For this
example, I summarize the data here only to demonstrate how you would generate summary
statistics if you received data that was already summarized. I hope it is apparent that it would be
more than a bit silly to summarize in PROC MEANS or PROC SUMMARY just so that the
FREQ option could be used in PROC REPORT.
title1 'Refining REPORT Appearance';
title2 'Using the FREQ Statement';
proc means data=rptdata.clinics(where=(region in('1','10')))
noprint;
class region ht;
var ht;
output out=clinics(keep=_type_ region ht count mean variance)
n=count mean=mean var=variance;
run;
The PROC MEANS output data (WORK.CLINICS) contains only one observation for each
distinct value of HT within a region. This is sufficient information to accurately calculate the
variance. For regions 1 and 10, the data that will be used by PROC REPORT looks like this:
Refining REPORT Appearance
Using the FREQ Statement
Summarized data used by REPORT
region
ht
count
1
10
10
74
63
69
4
2
4
Summarized data can still be used to
calculate statistics such as means and
variances. This data contains the counts or
frequency of each value of HT.
In region 1, all four patients had the same height. The data listing also indicates that there were
four observations with a height of 69 inches in the original data for region 10.
The PROC REPORT step uses the FREQ statement in order to calculate the mean and variance
correctly. The variable named on the FREQ statement does not usually appear on the COLUMN
statement. Effectively, when the statistics are calculated, each incoming observation is used the
number of times indicated by the frequency variable.
Chapter 6: Refining Our Understanding of the PROC REPORT Step 145
title3 'Mean and Variance Generated by REPORT';
proc report data=clinics nowd;
column region ht ht=htN ht=htvar;
define region / group format=$6.;
define htn
/ analysis n
'N';
define ht
/ analysis mean 'Mean';
define htvar / analysis var 'Variance';
freq count;
run;
Notice that COUNT does not appear on the COLUMN statement. The report generated by PROC
REPORT correctly calculates the means and variances by taking the frequency of each height.
Refining REPORT Appearance
Using the WEIGHT Statement
Mean and Variance Generated by REPORT
region
1
10
Mean
74
67
N
4
6
Variance
0
9.6
If you accidentally forgot to include the FREQ statement, the calculations would be done
incorrectly because the wrong N size would be used. The report would be as follows:
Refining REPORT Appearance
Using the WEIGHT Statement
Without including the FREQ Statement
region
1
10
Mean
74
66
N
1
2
Variance
.
18
MORE INFORMATION
The FREQ statement is used to calculate weighted means in Section 7.9.4.
6.8 A Further Comment on Paging Issues
Starting in SAS 9.2, PROC REPORT ignores the PAGESIZE and LINESIZE options in all nonmonospace destinations. Effectively, the page is treated as infinite in both dimensions. This means
that the only page boundaries are those introduced by a PAGE option on a BREAK, RBREAK
(see Section 3.2.4), or DEFINE statement (see Section 6.1.7).
Regardless of the ODS destination, the beginning of the report or the beginning of a BY group is
treated as a start-of-page, and the end of the report or the end of a BY group is treated as an endof-page.
If a report is directed to the LISTING destination and to a MARKUP destination such as HTML
(see Section 9.3) at the same time, any COMPUTE BEFORE/AFTER _PAGE_ blocks will be
executed at actual page boundaries in the LISTING output, but not in the MARKUP output.
146 Carpenter’s Complete Guide to the SAS REPORT Procedure
6.9 Chapter Exercises
1. Using the SASHELP.RETAIL quarterly sales data, for each year calculate the following sales
statistics: N, MEAN, SUM, and STDERR. Use aliases and a DEFINE statement for each
statistic. You might want to build on the results of Exercise 2 in Chapter 3.
2. Using the SASHELP.RETAIL quarterly sales data, for each year display the SALES amount
for each quarter (DATE) in a separate column (ACROSS). Why is a format such as QTR.
needed on DATE?
3. Building on the results of Exercise 2 in this section, add a spanning header for DATE and
modify the column labels. Is it necessary to nest SALES within DATE?
4. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Create a separate report BY YEAR that shows total PROFIT for each
PRODUCT_LINE and PRODUCT_CATEGORY.
Turn off the BYLINE option and place the year in the title using the #BYVAL title option.
C h a p t e r
7
Extending Compute Blocks
7.1 Understanding the Events of the Compute Block Process 149
7.1.1 Setup Phase: Generating the Computed Summary Information 150
7.1.2 Report Row Phase: Generating the Report 150
7.1.3 Process Example 151
7.2 Referencing Columns and Report Items in a Compute Block 154
7.2.1 Using Direct Variable Name References 156
7.2.2 Using Compound Variable Names 159
7.2.3 Using an Alias as a Column Reference 160
7.2.4 Using Absolute Column References: Referring to a Column by Its
Number 161
7.2.5 Using the Automatic Temporary Variable _BREAK_ 164
7.3 Using BEFORE and AFTER 166
7.4 Changing the Grouping Variable Values on Summary Lines 169
7.4.1 Specifying Text in a Compute Block 170
7.4.2 Using a Formatted Value 171
7.4.3 Creating a Dummy Column 173
7.5 Introducing the CALL DEFINE Routine 174
7.6 COMPUTE Statement Options and Switches 179
7.6.1 Justification of LINE Statement Text 179
7.6.2 Creating Character Variables with the CHARACTER and LENGTH=
Options 180
7.7 Using Logic and SAS Language Elements 182
7.7.1 Using the SUM Statement with Temporary Variables 183
148 Carpenter’s Complete Guide to the SAS REPORT Procedure
7.7.2 Repeating GROUP and ORDER Variables on Each Row 185
7.7.3 Counting Items across Page Breaks in the LISTING Destination 187
7.8 Doing More with the LINE Statement 191
7.8.1 Creating Group Summaries 192
7.8.2 Adding Repeated Characters 194
7.8.3 Understanding LINE Statement Execution 197
7.9 Examples of Common Tasks 199
7.9.1 Writing a Grand Total on Every Page 200
7.9.2 Combining Values into One Field or Column 202
7.9.3 Combining Values with Nested ACROSS Variables 204
7.9.4 Calculating a Weighted Mean 206
7.10 Chapter Exercises 209
Although understanding the intricacies of compute blocks is essential for advanced work with
PROC REPORT, it is also problematic. A number of issues make the successful use of compute
blocks difficult. This chapter discusses some of the issues surrounding their use.
As your compute blocks become more complex, understanding the compute block process
becomes even more essential. Often programmers who have trouble getting the compute block to
perform desired operations do not have a complete understanding of how the compute block
works and how it interacts with the various phases of the report generation process.
Unfortunately, the process itself is complex enough to provide ample opportunity for confusion.
Fortunately, the process can be broken down into components so that the source of the confusion
can be more easily explained. Several earlier sections of this book have provided increasingly
more complex overviews of the PROC REPORT step and specifically of the compute block
process. These sections include the following:
Section 1.4 provides an overview of the processing of the PROC REPORT step without
discussing compute blocks
Section 2.7 discusses the sequencing of step events with compute blocks. Emphasis is on the
timing issues associated with compute blocks that use the BEFORE and AFTER
locations.
Chapter 5 demonstrates a number of compute block examples and includes comments on the
compute block process.
The information in this chapter assumes that the reader has a reasonably good understanding of
the material in Section 1.4, Section 2.7, and Chapter 5. In order to get a better idea of what is
going on behind the scenes, Sections 7.1 and 7.2 provide views of the processing of the compute
block from different perspectives.
Section 7.1 looks at the events and process timing issues that take place during the processing of
the step.
Section 7.2 covers a number of report item naming issues.
The remaining sections of this chapter expand on the compute block introductions provided in
Section 2.6 and in Chapter 5. This is accomplished by the introduction of additional options,
Chapter 7: Extending Compute Blocks 149
techniques, and capabilities of the compute block. Each of these examples extend our
understanding of how the compute block is processed.
MORE INFORMATION
A detailed presentation of the processing of the PROC REPORT step is provided in Chapter 11,
“Details of the PROC REPORT Process.”
SEE ALSO
Chapman (2002) has a number of very instructive examples that highlight the use of the compute
block. Russ Lavery’s “An Animated Guide to the SAS REPORT Procedure,” which is included
on the CD that accompanies this book, highlights in detail the compute block’s relationship to the
report generation process.
7.1 Understanding the Events of the Compute
Block Process
As the PROC REPORT step is executed, a series of phased events takes place. Looking at the
process in terms of these phases can often prove to be helpful.
The following is a simplified diagram of the PROC REPORT process when compute blocks are
present.
The first phase of the process is the evaluation phase. It is during the execution of this phase that
determinations are made about the resources and summarizations that are needed during the setup
phase. The setup phase is the first of the two phases that are of primary interest to the SAS
programmer who desires a deeper understanding of the PROC REPORT process. The discussion
of these two primary phases includes the following sections:
150 Carpenter’s Complete Guide to the SAS REPORT Procedure
Section 7.1.1
Section 7.1.2
Section 7.1.3
The setup phase creates the computed summary information used in the
report row phase.
The report row phase processes the computed summary information one
row at a time.
Compute blocks are processed during the report row phase.
The order and position of the compute blocks within the PROC REPORT step is not what drives
this process. The process is driven by a combination of the order of the report items on the
COLUMN statement, presence of BREAK/RBREAK statements, and the compute blocks
themselves. Basically, the COLUMN statement defines the report columns and their order, while
the COMPUTE statements with BEFORE or AFTER and the BREAK or RBREAK statements
define how the summary rows should be constructed in the final report.
SEE ALSO
SAS Guide to the REPORT Procedure, Reference, Release 6.11 (Chapter 5) discusses event
timing in a fairly simple report step that contains a compute block.
7.1.1 Setup Phase: Generating the Computed Summary
Information
During the setup phase, the incoming data is read and, if necessary, the columns that are used for
summarizing are sent to the MEANS/SUMMARY engine where the summarization takes place. If
the data is not in sorted order, and the report needs to be ordered or grouped, then the
MEANS/SUMMARY engine also does the sorting.
If any variables have a define usage of ACROSS, then it is at this point that the number of levels
for the ACROSS variable is determined. This number enables PROC REPORT to calculate the
overall number of columns. After the total number of columns on the report is determined, the
absolute column names (_C1_, _C2_, etc) can be assigned. Usually you do not need to worry
about these internal absolute column names. However, they can be very useful in some situations
(see examples in Sections 7.2.4, 7.9.3, and 10.1.4).
The MEANS/SUMMARY engine results are saved in the computed summary information. This
data summarization is stored in memory and is available for use during the report row phase.
7.1.2 Report Row Phase: Generating the Report
The actual rows of the report are generated during the report row phase. The report is created one
row at a time, and for each row the report items are processed from left to right in the same order
as they are included on the COLUMN statement. As the items of the report row are processed,
any associated compute blocks are executed. Since PROC REPORT assigns values to the columns
in a row of a report from left to right, you cannot base the calculation of a computed variable on
any column that appears to its right in the report.
Compute blocks that are tied to a specific report item are executed for each row of the report,
while compute blocks that are defined with BEFORE or AFTER on the COMPUTE statement are
executed at the time that the specified breakpoint occurs.
It is during this phase that summarized information that is used in the final report is drawn from
the computed summary information (which was generated during the setup phase).
Chapter 7: Extending Compute Blocks 151
7.1.3 Process Example
If REGION is a grouping variable, the compute block defined by the following COMPUTE
statement executes only before each new REGION is encountered in the report.
compute before region;
Depending on the type of report being constructed during the report row phase (detail or
summary), the report rows are populated either directly from the input data set or from the
computed summary information held in active memory. When a compute block executes for a
particular report item, any summary information that is needed for the compute block statements
is retrieved from this memory area.
As the report rows are constructed, each report row is populated from left to right, corresponding
to the report items on the COLUMN statement. The report rows are constructed on the basis of
the statements that were set aside during the setup phase. For a simple report (such as a detail
report with no analysis variables or compute blocks) there might be no summary information for
the report. For these simple detail reports without compute blocks, the values for the report rows
might not change from what is found in the input data set.
The following step presents the number of visits per region and the percentage of the total number
of patient visits represented by each region. This requires a series of PROC REPORT step events.
Here is the final table that we want to generate:
Regional Patient Visits
region
Mid West
No. East
So. East
Western
Number
of Visits
26
24
14
16
80
percnt
32.50%
30.00%
17.50%
20.00%
100.00%
o
The PROC REPORT step utilizes the user-defined $REGNAME. format to group regions. The
incoming data set RPTDATA.CLINICS contains 10 distinct regions (REGION is a $2 variable).
The format $REGNAME is used to consolidate these 10 regions into 4 areas or super regions. The
percentages, which are held in the computed variable PERCNT, are based on the totals of these
larger areas. One of the beauties of using a format like this to define groups is that we do not need
to first create a separate variable in a separate step. This gives us a great deal of flexibility.
* Percentage of visits;
proc format;
value $regname
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
run;
title1 'Regional Patient Visits';
proc report data=rptdata.clinics
out=regout nowd;
column region n percnt; n
define region / group
format=$regname8.;
define n
/ 'Number of Visits';
152 Carpenter’s Complete Guide to the SAS REPORT Procedure
define percnt / computed format=percent9.2;
rbreak after
/ summarize suppress;
o
compute before ; p
totvisits = n;
endcomp;
compute percnt; q
percnt = n/totvisits;
endcomp;
run;
n The COLUMN statement defines the order for the variables in the report. Because we want to
use N to calculate PERCNT, N must appear to the left of PERCNT on the COLUMN
statement.
o The RBREAK statement is executed only once—at the end of the report. It makes no
difference that the RBREAK statement appears in the code before the compute blocks; it still
executes last. The SUPPRESS option is ignored when it appears on the RBREAK statement, as
it does here.
p This compute block assigns the current value of N to the temporary variable TOTVISITS. This
statement is executed only once (at the top of the report), when N contains the total number of
visits across the entire report. In this usage, N is the count of the observations, so as long as
every observation represents a visit, this is the correct statistic to use here. However, when we
need to count something other than just observations (perhaps observations with a nonmissing
value of an analysis variable) N would probably not be the appropriate statistic.
q The compute block for PERCNT is executed once for each report row. The percentage is
calculated as the ratio of N and the temporary variable TOTVISITS.
This report would fail completely if we simply reversed the order of the computed variable
PERCNT and the statistic N in the COLUMN statement.
column region percnt n;
If we use this COLUMN statement, the compute block that calculates the percentage now relies
on a report variable N that is on the right side of the computed variable PERCNT. When the
compute block for PERCNT executes, the value of N is not yet available, and the assignment
statement returns a missing value.
In the preceding PROC REPORT step, the OUT= option is used to generate a data set named
REGOUT. This table can often give us a fairly good idea of what has gone on during the
processing of the step.
Chapter 7: Extending Compute Blocks 153
Regional Patient Visits
Obs
region
1
2
3
4
5
n
80
percnt
5
1
4
10
6
_BREAK_
.
26
24
14
16
0.325
0.300
0.175
0.200
80
1.000
_RBREAK_
p
_RBREAK_
o
q
Notice that the value for PERCNT is missing in the first row of this data set. This does not matter
for our table, because this row is not going to appear in our final report. However, we need to look
at this a bit more closely if we truly want to understand the timing of the events.
Remember that compute blocks that do not have a BEFORE or AFTER are executed for each
report row. Also remember that not all report rows are written to the final report. This means that
the COMPUTE PERCNT q block is executed 6 times (because there are 6 report rows—which is
not necessarily the same as the number of rows in the report). Compute blocks that do have either
a BEFORE or AFTER execute only on selected report rows. For the first row (Obs 1) two
compute blocks execute (p and q).
When more than one compute block executes on a row, it becomes important to understand how
they affect the final report rows. If we add some silly code into the compute block for PERCNT
we can see what happens.
compute percnt; q
percnt = n/totvisits;
* Add two silly assignment statements to test order of
* events. Check the output data to see what happens;
if totvisits = . then percnt=22; r
else if _break_ = '_RBREAK_' then percnt=4; s
endcomp;
Here is the resulting output data set:
Regional Patient Visits
Obs
region
1
2
3
4
5
5
1
4
10
6
n
percnt
_BREAK_
80
26
24
14
16
22.000
0.325
0.300
0.175
0.200
r _RBREAK_
80
4.000
s _RBREAK_
When the compute block for PERCNT q executes, the value for TOTVISITS is still missing.
(Our silly code notes this by setting PERCNT to 22 r. It would have been set to 4.0, as it was at
s, if TOTVISITS had a nonmissing value). This outcome means that the BEFORE report
compute block p has not yet executed. The conclusion is important! For a given report row,
compute blocks associated with report items are executed before the compute blocks that contain
154 Carpenter’s Complete Guide to the SAS REPORT Procedure
BEFORE and AFTER. This would also be true if there were a COMPUTE AFTER statement; it
would execute last.
Because the report is processed one row at a time from left to right and from top to bottom, you
need to stage events so that information is available when you need it. Primarily this means that if
you need to use a value, that value has to be available in the computed summary information or to
the left on the current row. In fact, this is the sole purpose of the COMPUTE BEFORE p block.
In our example, on every report row, we need to know the overall number of patient visits in order
to calculate the percentage. This compute block makes that number available, and we retain it by
assigning it to a temporary variable (TOTVISITS).
As in this example, if you need to retain a value from one row to the next, use a temporary
variable. Because temporary variables are initialized to missing only once (at the start of the
report row phase) and are not cleared from row to row as the table is processed, these variables
are perfect for retaining information as the rows of the table are processed.
SEE ALSO
Pass and McNeil (2003) discuss timing issues and show a number of examples with compute
blocks. Chapman (2002) provides some extensive examples.
Although SAS Technical Report P-258 (1993, Chapter 10) contains a great deal of good
information and discusses the sequencing of events in detail, it is documenting pre-ODS behavior.
PROC REPORT has been extensively modified to work with ODS and to continually improve its
processing efficiency. Therefore, although the overall processing concepts remain the same, the
actual behind-the-scenes processing that is being performed has changed.
7.2 Referencing Columns and Report Items in a
Compute Block
In Section 2.6.3, the example includes the following compute block, which writes the value of the
variable REGION using the format $REGNAME. This compute block is referencing the report
item REGION.
compute before region;
line @3 region $regname8.;
endcomp;
You can reference any report item that forms a column, even columns that are not printed, and
there are four ways to reference a variable in a compute block. In the LINE statement of the
preceding compute block, the variable REGION, which is also a report item and a GROUP
variable, is referenced explicitly by name.
In a compute block you can reference report items in these ways:
explicitly by name
by using a compound name
by specifying an alias
directly by using the report column number
Chapter 7: Extending Compute Blocks 155
These four methods of referencing the report item are discussed specifically in the following
sections of this book:
Section 7.2.1
The variable name can be used directly, as in this example, when the
variable has a define type of GROUP, ORDER, COMPUTED, or
DISPLAY. Temporary variables, such as those that are created and used in a
compute block, are always addressed explicitly by variable name.
Section 7.2.2
Compound variable names are needed when an analysis variable has been
used to calculate a statistic. The association between analysis variable and
statistic can be established on the DEFINE statement or through nesting on
the COLUMN statement. The compound name, which was introduced in
Section 5.1, is a combination of the variable name and the statistic that it has
been used to calculate. The general form is variablename.statistic, as shown
in the following example:
wt.mean
Section 7.2.3
An alias can be specified in the COLUMN statement. Aliases are created
when you want to use an analysis variable in more than one way—generally
to calculate more than one statistic. The creation of aliases was introduced in
Section 6.2, and the following COLUMN statement generates a series of
aliases for the HT analysis variable.
columns region ht
ht=htmin ht=htmax
ht=htmean ht=htmedian;
Section 7.2.4
Sometimes, as the report is constructed, a given column might not have a
specific name. This is especially the case when a variable with the define
type of ACROSS creates a series of columns. These, and indeed any column
in the report, can be referenced by using the column number as an absolute
column reference. This absolute column name is always of the following
form:
_Cxx_
(The xx is the column number as read from left to right on the report.)
The column count includes any columns that ultimately do not print, e.g.,
those columns defined with NOPRINT or NOZERO.
Section 7.2.5
This section introduces the automatic temporary variable _BREAK_.
SEE ALSO
Carpenter (2006b) and SAS Technical Report P-258 (1993, pp. 134–135) both discuss how to
reference report items in a compute block.
156 Carpenter’s Complete Guide to the SAS REPORT Procedure
7.2.1 Using Direct Variable Name References
When a variable has the define type of DISPLAY, that variable’s name is used explicitly in
compute blocks. This is shown in the following example, which calculates a computed column.
The Body Mass Index, BMI, is a rough measure of health, and for most adults a BMI between
18.5 and 24.9 is generally considered to be in the normal range. The following example calculates
the BMI for the students in the SASHELP.CLASS data set.
title1 'Extending Compute Blocks';
title2 'Using Variable Names';
proc report data=sashelp.class nowd;
column name weight height BMI;
define name
/ display;
define weight / display;
define height / display;
define bmi
/ computed format=4.1 'BMI';
compute bmi;
bmi = weight / (height*height) * 703;
endcomp;
run;
Extending Compute Blocks
Using Variable Names
Name
Alfred
Alice
Barbara
Carol
Henry
James
Jane
Janet
Jeffrey
John
Joyce
Judy
Louise
Mary
Philip
Robert
Ronald
Thomas
William
Weight
112.5
84
98
102.5
102.5
83
84.5
112.5
84
99.5
50.5
90
77
112
150
128
133
85
112
Height
69
56.5
65.3
62.8
63.5
57.3
59.8
62.5
62.5
59
51.3
64.3
56.3
66.5
72
64.8
67
57.5
66.5
BMI
16.6
18.5
16.2
18.3
17.9
17.8
16.6
20.2
15.1
20.1
13.5
15.3
17.1
17.8
20.3
21.4
20.8
18.1
17.8
The computed variable BMI is
calculated using two display
variables, HEIGHT and
WEIGHT. Both of these
variables are addressed directly
by name in the compute block.
In this example, the two variables HEIGHT and WEIGHT have a define type of DISPLAY. This
means that they are not automatically set up to be used to calculate statistics. When DISPLAY
variables are used in a compute block, they are referenced directly, as is the computed variable
BMI, which is also addressed explicitly.
If these same variables (HEIGHT and WEIGHT) had been designated with a define type of
ANALYSIS (which is typical, and the default for numeric variables without DEFINE statements),
Chapter 7: Extending Compute Blocks 157
the previous compute block would not have worked (see the same example in Section 7.2.2,
where these same variables are given a define usage of ANALYSIS). The SAS log would contain
the following message, which seems to be wrong and is therefore confusing.
NOTE: Variable height is uninitialized.
NOTE: Variable weight is uninitialized.
NOTE: Division by zero detected at line 1 column 15.
The misleading error
messages result from
our misnaming of the
analysis variables in
the compute block.
Of course, the variables HEIGHT and WEIGHT do exist and both are initialized, but they have different
names in the compute block when they have a define type of ANALYSIS (see Section 7.2.2).
The following example converts the weight of each class member from pounds to kilograms (see
Section 5.1 for a similar example). Notice that the compute block is associated with a variable
with a define type of DISPLAY.
title1 'Extending Compute Blocks';
title2 'Using Variable Names';
title3 'Converting Pounds to Kilograms';
proc report data=sashelp.class nowd;
column name sex ('Weight' weight);
define name
/ order
'Name';
define sex
/ display
'Sex';
define weight / display format=6. 'Kg';
compute weight;
weight = weight / 2.2;
endcomp;
run;
The compute block executes for each row of the report, and since WEIGHT has a define type of
DISPLAY, a direct variable reference is used in the compute block to name the variable to be
converted.
Extending Compute Blocks
Using Variable Names
Converting Pounds to Kilograms
S
e Weight
Name
x
Kg
Alfred
M
51
Alice
F
38
Barbara
F
45
Carol
F
47
Henry
M
47
James
M
38
Jane
F
38
Janet
F
51
. . . Portions of the report are not shown . . .
The variable WEIGHT is on
the COLUMN statement and is
in the incoming data table.
This variable is therefore not a
temporary variable.
Temporary variables are created in compute blocks and are always addressed explicitly using the
variable name. A temporary variable is created when a variable name that is not on the COLUMN
statement is declared in a compute block using a statement such as an assignment or a SUM
158 Carpenter’s Complete Guide to the SAS REPORT Procedure
statement. This is demonstrated in the following example, which counts observations using the
temporary variable CNT.
One of the default features of PROC PRINT is the OBS column; however, this is not even an
option in PROC REPORT. The following example adds this observation counter to output
generated by a PROC REPORT step. In the compute block where the OBS column is computed,
two variables, OBS and CNT, are used. Because the variable CNT does not appear on the
COLUMN statement, it is a temporary variable. As is the case in the DATA step, temporary
variables that appear on a SUM statement are automatically initialized to 0 by PROC REPORT.
And because temporary values will be retained from row to row, the SUM statement is especially
useful n. The value of CNT, which contains the current row number, is assigned to the computed
variable OBS o.
title1 'Extending Compute Blocks';
title2 'Using Variable Names';
title3 'Creating an OBS Column';
proc report data=sashelp.class nowd;
column obs name sex weight;
define obs
/ computed 'Obs'
format=3.;
define name
/ order
'Name';
define sex
/ display 'Sex';
define weight / display 'Weight' format=6.2 ;
compute obs;
cnt + 1; n
obs = cnt; o
endcomp;
run;
Here is the resulting table:
Extending Compute Blocks
Using Variable Names
Creating an OBS Column
S
e
x
M
F
F
F
Obs Name
Weight
1 Alfred
112.50
2 Alice
84.00
3 Barbara
98.00
4 Carol
102.50
. . . Portions of the report are not shown . . .
Because values of report items
are not retained from row to
row, the variable OBS could
not be used in the SUM
statement. Temporary
variables such as CNT are
automatically retained, but
cannot be displayed. Hence we
need two variables, CNT and
OBS, in the compute block.
It is actually very common to use temporary variables, such as CNT, to retain values from one
row to the next during the report row phase. A common mistake is to attempt to use report item
variables for this purpose (an approach that could be successfully used in the DATA step).
However, the values in report items are cleared from one row to the next and the values of report
items cannot be retained.
MORE INFORMATION
The use of the SUM statement in the compute block is discussed in more detail in Section 7.7.1.
Chapter 7: Extending Compute Blocks 159
SEE ALSO
In a paper with a number of interesting techniques, Chapman (2003) includes an example of a
compute block that directly references report variables. The problem of generation of the OBS
column is specifically addressed by Pass (2000).
7.2.2 Using Compound Variable Names
Compound names have two parts: the name of the variable on the COLUMN statement, and the
name of a statistic, such as SUM, MEAN, STD. The two parts are joined with a period.
In the first example of Section 7.2.1, BMI is calculated for the students in SASHELP.CLASS. In
that example, the variables WEIGHT and HEIGHT receive a define usage of DISPLAY. When
the define usage of ANALYSIS is used, the variable cannot be addressed directly; instead the use
of compound names is required.
title1 'Extending Compute Blocks';
title2 'Using Compound Variable Names';
proc report data=sashelp.class nowd;
column name weight height BMI;
define name
/ display;
define weight / analysis;
define height / analysis;
define bmi
/ computed format=4.1 'BMI';
compute bmi;
bmi = weight.sum / (height.sum*height.sum) * 703;
endcomp;
run;
Extending Compute Blocks
Using Compound Variable Names
Name
Weight
Height
BMI
Alfred
112.5
69 16.6
Alice
84
56.5 18.5
Barbara
98
65.3 16.2
Carol
102.5
62.8 18.3
Henry
102.5
63.5 17.9
. . . Portions of the report are not shown . . .
The variables WEIGHT
and HEIGHT have both
been defined as ANALYSIS
variables. Direct reference
to these variables, without
using compound names,
results in errors.
A compound name is always a combination of the variable name and a statistic. When a statistic
has not been directly specified, SUM is the default statistic.
Because HEIGHT and WEIGHT are analysis variables, compound names are required in the
compute block. Whereas the following assignment statement worked in the first example of
Section 7.2.1, it would not work here because compound names are not used.
compute bmi;
bmi = weight / (height*height) * 703;
endcomp;
When statistics are specified for analysis variables, they must also appear as part of the compound
name whenever that variable is used in a compute block. The following example converts the
mean weight of class members from pounds to kilograms. Notice that the compute block is
associated with a variable with a define type of ANALYSIS, and this means that compound
160 Carpenter’s Complete Guide to the SAS REPORT Procedure
variable names are used in the compute block. In this case, because the requested statistic is
MEAN, .mean is appended to the analysis variable to form the compound name.
title1 'Extending Compute Blocks';
title2 'Using Compound Variable Names';
title3 'Converting Pounds to Kilograms';
proc report data=sashelp.class nowd;
column sex ('Weight' weight);
define sex
/ group 'Sex' format=$3.;
define weight / analysis mean format=6.1 'Kg';
compute weight;
weight.mean = weight.mean / 2.2;
endcomp;
run;
Notice that the name of the compute block itself is not a compound name. Only the variables
within a compute block require a compound name.
SEE ALSO
Cochran (2005) and Chapman (2002) both show several examples of the use of compound
variable names.
7.2.3 Using an Alias as a Column Reference
When aliases are used in a compute block, they are addressed by using the alias explicitly. In the
following example, the alias WTMAX, which is created with the specification WT=WTMAX in
the COLUMN statement, allows you to use the report item WT in two different DEFINE
statements. This alias can then be used directly in the compute block.
title1 'Extending Compute Blocks';
title2 'Using an Alias';
proc report data=rptdata.clinics nofs;
column region ('Weight in Pounds' wt wt=wtmax wt_range);
define region
/ group format=$6.;
define wt
/ analysis min 'Min'; n
define wtmax
/ analysis max 'Max';
define wt_range / computed
'Range';
compute wt_range;
wt_range = wtmax - wt.min; o
endcomp;
run;
n The statistic MIN is specified for the variable WT.
o The alias WTMAX and the computed variable WT_RANGE are both referenced directly,
whereas the analysis variable WT is referenced as a compound name that includes the statistic
(MIN).
Chapter 7: Extending Compute Blocks 161
Extending Compute Blocks
Using an Alias
region
1
10
2
3
4
5
6
7
8
9
Weight in Pounds
Min
Max
195
195
163
177
105
115
105
195
131
201
98
215
175
240
147
155
158
162
147
215
Range
0
14
10
90
70
117
65
8
4
68
Notice the use of the compound names in the compute block assignment statement o. Because
WTMAX is an alias, the following assignment statement would not work:
wt_range = wtmax.max - wt.min;
Nor could we ignore the compound name for the analysis variable WT. The following assignment
statement would also fail.
wt_range = wtmax - wt;
In the early versions of PROC REPORT, variable aliases could not be used in compute blocks.
Fortunately, this is no longer true. However, occasionally you might encounter legacy code that
contains workarounds that are no longer necessary.
7.2.4 Using Absolute Column References: Referring to a
Column by Its Number
Derived columns, such as those created by using the ACROSS define usage, do not have a
column name on the COLUMN statement. When these columns are used in a compute block, an
absolute column reference is required, although any column on the report can be referenced by
using its absolute column reference. These absolute column names have the form _Cxx_, where xx
is the column number.
Determining the column number
Sometimes you need to address the column of the report by number. These numbers are known as
the absolute column numbers, and sometimes they are also called indirect column references.
These column numbers are determined in the setup phase for PROC REPORT, which takes into
account a number of factors, such as the number of levels for the ACROSS variable and any
ORDER= option that might have an effect on the final report row column order.
For the programmer, the column number itself is usually determined by inspection. Start by
counting the variables in the COLUMN statement (from left to right). Columns that do not appear
in the report (NOPRINT columns and columns eliminated with the NOZERO option) are included
in the count. With the exception of ACROSS variables, each variable/alias/statistic combination
results in a single column. ACROSS variables result in one column per value of the ACROSS
variable (not knowing the number of levels taken on by the ACROSS variable makes things more
interesting). When variables or statistics are nested within an ACROSS variable, the number of
162 Carpenter’s Complete Guide to the SAS REPORT Procedure
columns can increase drastically. As the table becomes more complex (and if it is not complex
then you will generally not need the absolute column number anyway), you can use the OUT=
option to take a look at the structure of the report. This should help with the counting process.
The following example is similar to the one used in Section 6.3. It nests three statistics for WT
under the ACROSS variable SEX.
title1 'Extending Compute Blocks';
title2 'Using Indirect Column References';
title3 'Determining Column Counts';
* Nesting statistics within an ACROSS ;
proc report data=rptdata.clinics nowd;
column region sex,('_Weight_' wt wt=wtmean wt=wtstd );
define region / group width=6;
define sex
/ across
format=$2. '_Gender_';
define wt
/ analysis n
format=2.0 'N';
define wtmean / analysis mean format=6.2 'Mean';
define wtstd / analysis mean format=6.1 'STD';
run;
Here is the resulting table:
Extending Compute Blocks
Using Indirect Column References
Determining Column Counts
region
1
10
2
3
4
5
6
7
8
9
_c1_
________________Gender________________
F
M
______Weight______ ______Weight______
N
Mean
STD
N
Mean
STD
.
.
.
4 195.00
195.0
2 163.00
163.0
4 177.00
177.0
6 109.67
109.7
4 105.00
105.0
5 127.80
127.8
5 163.80
163.8
4 143.00
143.0 10 165.60
165.6
5 146.20
146.2
3 177.00
177.0
4 187.00
187.0
6 205.33
205.3
.
.
.
4 151.00
151.0
4 160.00
160.0
.
.
.
2 177.00
177.0
8 190.50
190.5
_c2_
_c3_
_c4_
_c5_
_c6_
There is one column
for REGION and then
a column for each
combination of SEX
and the nested
variables (2 X 3 = 6
columns).
Column numbers have
been added.
_c7_
If we wanted to address the mean weight of males in a compute block, we would use _C6_ as the
indirect reference to that column.
In the following example, the ratio of the mean weight of the two genders is calculated in a
compute block. It is our responsibility to determine whether column _C2_ will contain the mean
weight for the males or the females. In this example, the DEFINE statement for SEX has an
ORDER=DATA option n, and this option can affect which gender will be in _C2_.
We can inspect the data, or the first draft of the report, to determine that the mean weight of males
is in the second column. Because we want the weight of the females to be the numerator, we use
_c3_ / _c2_ as the ratio o.
Chapter 7: Extending Compute Blocks 163
* Using indirect column references;
proc format;
value $regname
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
value $gender
'M' = 'Male'
'F' = 'Female';
run;
title1 'Extending Compute Blocks';
title2 'Using Indirect Column References';
proc report data=rptdata.clinics nowd split='*';
column region ('Mean Weight*in Pounds' sex,wt ratio);
define region / group width=10 'Region'
format=$regname. order=formatted;
define sex
/ across 'Gender'
format=$gender. order=data; n
define wt
/ analysis mean format=6. ' ';
define ratio
/ computed format=6.3 'Ratio*F/M ';
rbreak after
/ dol summarize;
compute ratio;
ratio = _c3_ / _c2_; o
endcomp;
run;
We verify by inspection that the weight for males is in the second column (_C2_) and that the
weight of the females is in the third column (_C3_). The compute block therefore calculates the
correct ratio.
Extending Compute Blocks
Using Indirect Column References
Region
Mid West
No. East
So. East
Western
Mean Weight
in Pounds
Gender
Male Female
Ratio
F/M
182
163
0.895
155
118
0.759
166
143
0.864
186
170
0.914
====== ====== ======
172
146
0.846
It is not always immediately obvious
what the order of the levels of the
ACROSS variable will be.
Be careful—especially when an
ORDER= option is present.
Because the column numbers of the table are used in the compute block, the programmer must
have a certain knowledge of the final table before selecting absolute column numbers. In the
previous example, we can observe that in the incoming data table the first observation is for a
male patient and that the ORDER= option has been set to DATA. Changing the DEFINE
statement for SEX to use ORDER=FORMATTED
define sex / across 'Gender'
format=$gender. order=formatted;
164 Carpenter’s Complete Guide to the SAS REPORT Procedure
would place FEMALES in column 2 (“F” sorts before “M”). The assignment statement for
RATIO would then be as follows:
compute ratio;
ratio = _c2_ / _c3_;
endcomp;
MORE INFORMATION
The example in Section 7.9.3 adds another level of nesting and a computed variable to the list of
nested columns.
SEE ALSO
SAS Technical Report P-258 (1993, pp 82–83) and Burlew (2005, pp. 42–49) both include
examples of the use of indirect column names.
7.2.5 Using the Automatic Temporary Variable _BREAK_
As PROC REPORT executes the report row phase, the report rows are constructed and, when
break or summary rows are needed, the computed summary information is retrieved from the
memory area created in the setup phase. The automatic temporary variable _BREAK_ is used to
identify lines in the computed summary information that were generated as a result of grouping
variables, BREAK statements, or RBREAK statements.
Adding an OUT= option to the PROC statement allows us to save the final report rows in the
form of a SAS data table. In the following example, an RBREAK statement creates a summary
line that summarizes across regions.
proc format;
value $regname
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
value $gender
'M' = 'Male'
'F' = 'Female';
run;
title1 'Extending Compute Blocks';
title2 'Examining the _BREAK_ Variable';
proc report data=rptdata.clinics nowd split='*'
out=temptbl;
column region wt wt=wtmean;
define region / group width=10 'Region'
order=formatted format=$regname. ;
define wt
/ analysis n
format=2.;
define wtmean / analysis mean
format=6.2;
rbreak after
run;
/ dol summarize;
proc print data=temptbl;
run;
Chapter 7: Extending Compute Blocks 165
The following display shows the data table (WORK.TEMPTBL) resulting from the OUT= option.
The variable _BREAK_ has missing (blank) values for the detail rows of the report. However, for
the summary line that is generated for the report break, the variable _BREAK_ takes on the value
of ‘_RBREAK_’. As we should anticipate, REGION is appropriately blank on the summary line,
because the numbers on this line represent all the regions.
Extending Compute Blocks
Examining the _BREAK_ Variable
The variable _BREAK_
has missing values on
detail rows. For
summary lines, it
indicates the type of
summary.
Obs
region
wt
wtmean
_BREAK_
1
2
3
4
5
5
1
4
10
26
24
14
16
80
172.538
138.167
159.143
182.000
161.775
_RBREAK_
When a BREAK statement is added to this report, its associated summary lines are indicated by
the _BREAK_ variable. The following example repeats the previous example, but adds a second
group variable (SEX) and a BREAK statement.
. . . Portions of the code are not shown . . .
proc report data=rptdata.clinics nowd split='*'
out=temptbl;
column region sex wt, (n mean);
define region / group width=10 'Region'
order=formatted format=$regname. ;
define sex
/ group
format=$gender7.;
define wt
/ analysis n
format=2.;
define wtmean / analysis mean
format=6.2;
break after region / summarize skip suppress;
rbreak after
/ dol summarize;
run;
. . . Portions of the code are not shown . . .
Because they were generated from the BREAK AFTER REGION statement the _BREAK_
variable indicates these summary lines with the inclusion of the value of REGION. The data set
generated by the OUT= option shows the _BREAK_ variable and its values.
Extending Compute Blocks
Examining the _BREAK_ Variable
Obs
1
2
3
4
5
6
7
8
9
10
11
12
13
region
sex
wt
wtmean
5
5
5
1
1
1
4
4
4
10
10
10
F
M
13
13
26
11
13
24
4
10
14
4
12
16
80
163.000
182.077
172.538
117.909
155.308
138.167
143.000
165.600
159.143
170.000
186.000
182.000
161.775
F
M
F
M
F
M
_BREAK_
region
region
region
region
_RBREAK_
166 Carpenter’s Complete Guide to the SAS REPORT Procedure
The value contained by the _BREAK_ variable can be essential when performing logical
processing that must detect group or report boundaries. Several examples in this book use
_BREAK_ in logical expressions (see Sections 7.4.3, 7.8.1, and 7.8.2).
MORE INFORMATION
Various values of _BREAK_ are shown and used in Section 7.3.
SEE ALSO
Chapman (2002) includes examples that use the _BREAK_ variable.
7.3 Using BEFORE and AFTER
You can use the BEFORE and AFTER location options on the compute statement to determine
when the compute block is to be executed. These timing locations can be relative to both grouping
variables and to the overall report. In the following example, we want to calculate the percentage
of patients falling within each weight group both by and across gender. We want to generate the
following report.
Extending Compute Blocks
Using BEFORE and AFTER
Gender
Female
Male
Weight
< 100
100-< 200
200-< 300
100-< 200
200-< 300
N
2
29
1
==
32
Percent
by Sex
6%
91%
3%
========
100%
Percent
Total
3%
36%
1%
========
40%
37
11
==
48
77%
23%
========
100%
46%
14%
========
60%
==
80
========
.
========
100%
Percentage calculations
are always a bit more
interesting, because they
require a current value
and a group total.
This example has two
types of percentage
calculations and therefore
requires two types of
totals.
The variable WT is used as a grouping variable, and an alias of WT (WTN n) is used to collect
the number of patients. The format PNDS. will be used to form the groups of interest.
proc format;
value pnds
low-<100
100-<200
200-<300
300-high
value $gender
'M'='Male'
'F'='Female';
run;
=
=
=
=
' < 100'
'100-< 200'
'200-< 300'
'300 and over';
Chapter 7: Extending Compute Blocks 167
title1 'Extending Compute Blocks';
title2 'Using BEFORE and AFTER';
proc report data=rptdata.clinics nowd
out=outpct split='*';
column sex wt wt=wtn percnt tpercnt;
define sex
/ group format=$gender6.
'Gender' ;
define wt
/ group format=pnds.
order=formatted
'Weight';
define wtn
/ analysis n
format=2.
'N'; n
define percnt / computed
format=percent8.
'Percent*by Sex';
define tpercnt / computed
format=percent8.
'Percent*Total';
compute before; q
* Total number of patients;
totcount = wtn;
endcomp;
o
p
compute before sex; r
* Total number within gender group;
count = wtn;
endcomp;
compute percnt; s
* percent within weight group;
percnt= wtn/count;
endcomp;
compute tpercnt; t
* Total percent within weight group;
tpercnt= wtn/totcount;
endcomp;
* Percent count for summary after SEX;
compute after sex; _
percnt= wtn/count;
tpercnt= wtn/totcount;
endcomp;
break after sex / suppress summarize dol skip;
* Percent count for summary after the report
* (across SEX);
compute after; `
percnt= .;
tpercnt= wtn/totcount;
endcomp;
rbreak after / summarize dol;
run;
168 Carpenter’s Complete Guide to the SAS REPORT Procedure
In each of the compute blocks above, notice the usage of the variable WTN. It has a very different
meaning in each of the compute blocks used in this PROC step.
n The alias WTN is used to collect the number of patients with nonmissing values of WT
(which is being used as a grouping variable).
o PERCNT is a computed variable that holds the percentage of patients within each gender
and weight group.
p The report item TPERCNT is a computed variable that holds the percentage of patients across
genders.
q During the execution of this compute block, WTN contains the total number of nonmissing
values of WT. (Assuming that there are no missing values for WT, this count is also the total
number of patients.) In order to retain this total, it is saved in the temporary variable
TOTCOUNT. This is the denominator for all percentage calculations based on the overall
patient count. This compute block is executed only once for the entire report.
r During the execution of this compute block, WTN contains the number of nonmissing values
of WT within a gender. Because the timing location is specified as BEFORE, just before the
value of SEX changes, the number of patients (with a weight) is saved in the temporary
variable COUNT. Since the value of a temporary variable is automatically retained, this value
of COUNT is available for all the detail lines associated with this value of SEX. COUNT
becomes the denominator for all percentage calculations within a gender. This compute block
executes only once for each value of SEX.
s The value of PERCNT is calculated for the detail lines of the report. During the execution of
this compute block, WTN contains the number of nonmissing values of WT within a specific
weight group. The assignment statement in this compute block calculates a computed variable
(PERCNT) using the temporary variable COUNT and a report item variable (WTN).
t The value of TPERCNT is calculated during the execution of this compute block for the detail
lines of the report. Although PERCNT and TPERCNT are both calculated for each detail line,
both the PERCNT and the TPERCNT compute blocks must appear. This is because for detail
lines, the report item’s computed value is only transferred to the report row from its own
compute block. During the execution of this compute block, WTN contains the number of
nonmissing values of WT within a specific weight group.
_ This compute block and the BREAK statement both have the location of AFTER SEX,
so it is here that the values for PERCNT and TPERCNT are calculated for the group
summaries for each gender. During the execution of this compute block, WTN contains the
number of nonmissing values of WT within this value of gender.
` Since no variable is on this COMPUTE statement, it is executed after the entire report, along
with the RBREAK statement. It is at this time that the values for PERCNT and TPERCNT are
calculated for the final summary line of the report. During the execution of this compute block,
WTN contains the number of nonmissing values of WT across the entire table. Of course,
WTN and TOTCOUNT will have the same value at this point. This compute block is executed
only once, at the end of the report.
Sometimes it can help to understand the relationship of the compute blocks and the report item
variables by looking at the data set created by using the OUT= option on the PROC REPORT
statement. The data set associated with the example in this section has been printed here.
title3 'Final Report Rows';
proc print data=outpct;
run;
Chapter 7: Extending Compute Blocks 169
Extending Compute Blocks
Using BEFORE and AFTER
Final Report Rows
Obs
sex
1
wt
wtn
.
80
percnt
.
.
2
F
.
32
3
F
98
2
0.06250
0.0250
4
F
105
29
0.90625
0.3625
5
F
201
1
0.03125
0.0125
6
F
.
32
1.00000
0.4000
7
M
.
48
1.50000
0.6000
8
M
105
37
0.77083
0.4625
9
M
201
11
0.22917
0.1375
10
M
.
48
1.00000
0.6000
.
80
11
.
tpercnt
.
0.4000
1.0000
_BREAK_
q
sex
r
s
s
s
sex
_
sex
r
s
s
sex
_
_RBREAK_ `
_RBREAK_
t
t
t
t
t
SEE ALSO
Chapman (2002) and Cochran (2005) both show several good examples of the use of the
BEFORE and AFTER location options. SAS Technical Report P-258 (1993, Chapter 6) discusses
the use of compute blocks with summaries.
7.4 Changing the Grouping Variable Values on
Summary Lines
The BREAK and RBREAK statements both create summary rows in the computed summary
information, and if the SUMMARIZE option is specified, the summaries appear in the final report
as well. By default, when the SUMMARIZE option is used, the resulting summary row contains
the value of the grouping variable that is being summarized.
Often we want to replace the value of the grouping variable with text that we supply, but there is
no option to do this. As close as we can come to directly controlling the text through options is
with the use of the SUPPRESS option, which merely removes the grouping variable value.
Fortunately, although text cannot be directly specified through options, it is possible for us to
supply this text ourselves.
In the following example, regions are used to form groups based on the user-defined format
$REGNAME n. In addition, a request for an overall summary line has been made through the
use of the RBREAK statement Y. Remember that SUPPRESS is not used with the RBREAK
statement, as there is no value of the grouping variable to suppress Z.
170 Carpenter’s Complete Guide to the SAS REPORT Procedure
proc format;
value $regname n
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
run;
title1 'Extending Compute Blocks';
title2 'RBREAK Does not Create Summary Text';
proc report data=rptdata.clinics nowd split='*';
column region edu ht wt;
define region / group width=10 'Region' n
format=$regname. order=formatted;
define edu
/ analysis mean 'Years of*Education'
format=9.2 ;
define ht
/ analysis mean format=6.2 'Height';
define wt
/ analysis mean format=6.2 'Weight';
rbreak after
/ summarize dol; Y
run;
The resulting output table shows that for the summary line, the REGION is blank. Z
Extending Compute Blocks
RBREAK Does not Create Summary Text
Region
Mid West
No. East
So. East
Western
Z
Years of
Education
14.31
13.25
15.00
12.63
=========
Height
66.85
67.33
69.00
67.25
======
Weight
172.54
138.17
159.14
182.00
======
13.78
67.45
161.78
By default there is no text Z
under the grouping variable
for the RBREAK summary
line.
The remaining examples in Section 7.4 show some techniques that can be used to supply text for
this and other summary rows.
7.4.1 Specifying Text in a Compute Block
This is simplest method for adding text to the summary line; however, it does have limitations. If
we expand the PROC REPORT step in the previous section by adding the following compute
block, we can change the value of REGION, which otherwise has a value of MISSING.
* Text for the report summary line;
compute after;
region = 'Combined';
endcomp;
After adding the compute block, the following report is generated.
Chapter 7: Extending Compute Blocks 171
Extending Compute Blocks
Using COMPUTE to Supply Summary Text
Region
Mid West
No. East
So. East
Western
==========
Co
Years of
Education
14.31
13.25
15.00
12.63
=========
13.78
Height
66.85
67.33
69.00
67.25
======
67.45
Weight
172.54
138.17
159.14
182.00
======
161.78
The text ‘Combined’ has
become truncated because
REGION is a character
variable with a length of 2
spaces.
Notice that only the first two letters of the word ‘Combined’ are displayed. This is because
REGION is a $2 variable, and although a width of 10 has been specified, only the first two letters
will fit into the variable. This problem is addressed in Section 7.4.2.
A problem can also occur if you try to assign a value of the wrong type to the grouping variable in
the compute block. You cannot assign text to a numeric variable or numbers to a character
variable.
The following two sections of this book demonstrate alternative approaches to this problem.
SEE ALSO
This approach is successful in an example in SAS Technical Report P-258 (1993, pp. 146–147).
7.4.2 Using a Formatted Value
In the example in Section 7.4.1, only two of the letters of the text ‘Combined’ were displayed.
Because REGION has an associated format ($REGNAME.), the format is automatically applied
to the value stored in REGION by the compute block. Of course ‘Co’ is not on the format, so the
value is displayed unformatted.
We can easily solve this problem by including a line in the format definition that will
accommodate the summary text n.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
other
=
run;
'No. East'
'So. East'
'Mid West'
'Western'
'Combined';
n
title1 'Extending Compute Blocks';
title2 'Using a Format to Rename Summary Text';
proc report data=rptdata.clinics nowd split='*';
column region edu ht wt;
define region / group width=10 'Region'
format=$regname. order=formatted;
define edu
/ analysis mean 'Years of*Education'
format=9.2 ;
define ht
/ analysis mean format=6.2 'Height';
172 Carpenter’s Complete Guide to the SAS REPORT Procedure
define wt
/ analysis mean format=6.2 'Weight';
rbreak after
/ summarize dol;
* Text for the report summary line;
compute after;
region = 'x'; o
endcomp;
run;
n Values other than MISSING and the 10 region numbers are displayed as ‘Combined’. The
only time this happens (in this particular data) is when a value is assigned to the variable
REGION, and this is done in the compute block o.
o REGION is assigned a value other than MISSING or one of the acceptable region numbers.
Through the use of the $REGNAME. format, this value is displayed as ‘Combined’.
Extending Compute Blocks
Using a Format to Rename Summary Text
Region
Mid West
No. East
So. East
Western
==========
Combined
Years of
Education
14.31
13.25
15.00
12.63
=========
13.78
Height
66.85
67.33
69.00
67.25
======
67.45
Weight
172.54
138.17
159.14
182.00
======
161.78
‘Combined’ is not truncated
here as it was in Section 7.4.1,
because it is actually a
formatted value.
In the DATA step, a missing value is picked up by the OTHER in the format definition. That is
not the case in the PROC REPORT step when you are formatting a value on a summary line. This
means that you have to assign a nonmissing value to REGION to make this technique work.
If you want to use the previous technique, but for one reason or another you do not have
permission to modify the $REGNAME. format, consider the use of a two-step format. The PROC
FORMAT step becomes the following:
proc format;
value $regoth
'1' - '9'
=
other
=
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
[$regname.]
'Combined';
'No. East'
'So. East'
'Mid West'
'Western';
The FORMAT= option now points to the outer of the two formats (the one you do have
permission to modify). The DEFINE statement is now the following:
define region
/ group width=10 'Region'
format=$regoth. order=formatted;
For all the standard regions (1 through 9) the value is handed off to the $REGNAME. format and
all other values are given a formatted value of ‘Combined’. Actually I have been a bit carefree
here, because alternate unacceptable values of REGION, such as 1Q, are also passed to
$REGNAME.
Chapter 7: Extending Compute Blocks 173
7.4.3 Creating a Dummy Column
In the examples in Sections 7.4.1 and 7.4.2, text is assigned to the summary column. Another
approach to adding the text is to build a computed column specifically designed to hold the text of
interest.
In this example, we are back to using the original format to map the values of REGION to the
appropriate name, and a new computed column, REGNAME, has been added to the COLUMN
statement n.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Using COMPUTE to Create a Text Column';
proc report data=rptdata.clinics nowd split='*';
column region regname edu ht wt; n
define region / group noprint format=$regname.; o
define regname / computed 'Region'; p
define edu
/ analysis mean 'Years of*Education'
format=9.2 ;
define ht
/ analysis mean format=6.2 'Height';
define wt
/ analysis mean format=6.2 'Weight';
rbreak after
/ summarize dol;
* Determine the region name;
compute regname/char length=8;
if _break_='_RBREAK_' then regname = 'Combined';
else regname = put(region,$regname.); r
endcomp;
run;
q
n Both REGION and REGNAME appear on the COLUMN statement.
o The column REGION is used for grouping, but is not printed. Notice that the format is still
required, as it is used to form the appropriate groups.
p Define the computed column REGNAME to hold the text (row identifiers).
q Since this is a compute block associated with a computed variable, it is executed once for
each report row. We need to be able to distinguish the type of row, and we do this by using the
automatic variable _BREAK_. When the summary line generated by the RBREAK statement
is being processed (_BREAK_=‘_RBREAK_’), we supply the desired text (‘Combined’) r.
r Use the PUT function to assign the text to REGNAME using the $REGNAME. format for each
line other than the final summary line.
174 Carpenter’s Complete Guide to the SAS REPORT Procedure
The resultant table is similar to that in Section 7.4.2.
Extending Compute Blocks
Using COMPUTE to Create a Text Column
Region
Mid West
No. East
So. East
Western
========
Combined
Years of
Education
14.31
13.25
15.00
12.63
=========
13.78
Height
66.85
67.33
69.00
67.25
======
67.45
Weight
172.54
138.17
159.14
182.00
======
161.78
Although not apparent from an
inspection of the table, the column
labeled REGION is actually a
computed character variable.
MORE INFORMATION
The values taken on by the _BREAK_ variable are discussed in Section 7.2.5. The CHARACTER
(CHAR) and LENGTH options on the COMPUTE statement are discussed in Section 7.6.2. The
example in Section 7.7.2 also creates a computed region name.
SEE ALSO
An example in SAS Technical Report P-258 (1993, pp. 78–79) uses a NOPRINT option to hide
the variables used to order the table and a computed variable to provide a variation of the order
variables. Also in P-285, a summary line text value is changed through the generation of a new
computed value (pp. 109–110).
7.5 Introducing the CALL DEFINE Routine
The CALL DEFINE routine is used to set the attributes of a value in a particular cell. This routine
has three arguments that specify the column, attribute, and the value for the attribute.
call define(columnid, attribute, attributevalue);
columnid
identifies the column for which the attribute is to be applied. This can be a
column name (literal or a character expression) or a column number
(including a numeric literal, name of the form _Cxx_, or a numeric
expression).
The column identifier can include two special keywords (as keywords, these
are not quoted):
_COL_ refers to the current column.
_ROW_ refers to the entire row rather than the specific column.
attribute
specifies the attribute. Attributes have specific names, some of which are
described in the following section.
associates a specific attribute value with the attribute in the second
argument. Many attribute values are specific to particular attributes.
attributevalue
Chapter 7: Extending Compute Blocks 175
Only one attribute is specified per CALL DEFINE statement. However, multiple CALL DEFINE
statements can be specified.
The primary attributes that can be specified include the following:
FORMAT
STYLE
URL
URLBP
URLP
specifies a column format.
overwrites the ODS style attributes.
specifies that the contents of the cell form a link.
specifies that the contents of the cell form a link.
specifies that the contents of the cell form a link.
Although the FORMAT attribute can be used with any ODS destination, including the output
window or LISTING destination, the other attributes are best used with ODS destinations other
than LISTING. The STYLE attribute is described in Section 8.3, and the attributes used to form
links (URL, URLBP, URLP) are described in Section 8.5.
The following simple PROC REPORT example calculates the N and MEAN for the variable WT,
which is nested within the ACROSS variable SEX. Consequently the report will have 5 columns.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
value $gender
'M'='Male'
'F'='Female';
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Mean Weight';
title3 'Unformatted Analysis Columns';
proc report data=rptdata.clinics nowd;
column region sex,wt,(n mean);
define region / group
format=$regname.;
define sex
/ across format=$gender6. 'Gender' ;
define wt
/ analysis ' ';
run;
176 Carpenter’s Complete Guide to the SAS REPORT Procedure
Extending Compute Blocks
Mean Weight
Unformatted Analysis Columns
Gender
Female
region
Mid West
No. East
So. East
Western
n
13
11
4
4
Male
mean
163
117.90909
143
170
n
13
13
10
12
mean
182.07692
155.30769
165.6
186
A format for MEAN
would improve the
appearance of this
report.
We would like to format the values for N and MEAN, and one approach might be to add a
FORMAT= option to the DEFINE statement.
define wt
/ analysis ' ' format=5.1;
This does not really give us what we would like. Here is the resulting report:
Extending Compute Blocks
Mean Weight
DEFINE Statement Format
Gender
Female
region
Mid West
No. East
So. East
Western
n
13.0
11.0
4.0
4.0
Male
mean
163.0
117.9
143.0
170.0
n
13.0
13.0
10.0
12.0
mean
182.1
155.3
165.6
186.0
A format on the
DEFINE statement
for WT is applied
to both N and
MEAN.
We see that the format has been applied to all the columns associated with weight. This would not
be a problem for computed variables, because they have individual DEFINE statements. But when
columns have been generated by an ACROSS operation, the individual columns are no longer
associated with individual DEFINE statements.
We could also declare DEFINE statements for the individual statistics.
define wt
define n
define mean
/ analysis ' ';
/ format=3.;
/ format=5.1;
In this example this actually gives us what we want, but we still do not have control over the
formatting in the individual columns (both N columns and both MEAN columns must be
formatted the same).
Chapter 7: Extending Compute Blocks 177
Extending Compute Blocks
Mean Weight
DEFINE Statements for Individual Statistics
Gender
Female
region
Mid West
No. East
So. East
Western
n
13
11
4
4
Male
mean
163.0
117.9
143.0
170.0
n
13
13
10
12
mean
182.1
155.3
165.6
186.0
The statistics are formatted
individually; however, we still do
not have control of the individual
columns within a statistic.
Fortunately, the CALL DEFINE statement provides the control that we are seeking, and it can be
used to apply formats to individual columns even when they have been generated with an
ACROSS. A compute block for the variable WT has been added to the previous example, and a
series of CALL DEFINE statements have been written, one for each of the columns generated by
the ACROSS operation. The PROC REPORT step becomes the following:
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
value $gender
'M'='Male'
'F'='Female';
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Mean Weight';
title3 'Using CALL DEFINE to Format a column';
proc report data=rptdata.clinics nowd;
column region sex,wt,(n mean);
define region / group
format=$regname.;
define sex
/ across format=$gender6. 'Gender' ;
define wt
/ analysis
' ';
compute wt;
call define('_c2_','format','2.');
call define('_c3_','format','5.1');
call define('_c4_','format','2.');
call define('_c5_','format','5.1');
endcomp;
run;
178 Carpenter’s Complete Guide to the SAS REPORT Procedure
Extending Compute Blocks
Mean Weight
Using CALL DEFINE to Format a column
Gender
Female
region
Mid West
No. East
So. East
Western
n
13
11
4
4
Male
mean
163.0
117.9
143.0
170.0
n
13
13
10
12
mean
182.1
155.3
165.6
186.0
Formats have now
been applied to
each statistic
individually.
Four columns have been derived from WT. Because these are the columns that are to be
formatted, the appropriate report item to use on the COMPUTE statement is WT. The compute
block for WT is called for each statistic associated with WT, and each time it is called, all four
assignment statements are executed. This is a bit of inefficiency, but this small level of
inefficiency is generally of minor concern. Even this minor inefficiency can be eliminated when
the variable that we need to format is not the furthest to the right on the COLUMN statement.
The following example repeats the previous example, but it adds one more variable to the
COLUMN statement. In this case, HT is to the right of WT on the COLUMN statement, and
therefore the report item on the COMPUTE statement can be HT.
. . . Portions of the code are not shown . . .
proc report data=rptdata.clinics nowd;
column region sex,wt,(n mean) ht;
define region / group format=$regname.;
define sex / across format=$gender6. 'Gender' ;
define wt / analysis ' ';
define ht / analysis mean format=6.2 'Height';
compute ht;
call define('_c2_','format','2.');
call define('_c3_','format','5.1');
call define('_c4_','format','2.');
call define('_c5_','format','5.1');
endcomp;
run;
The compute block for HT is executed only once for each row, rather than the four times when
WT was used as the report item on the COMPUTE statement. When you want to use this
technique, but do not want to display another variable, you can use a computed variable that is not
printed to accomplish the same result.
. . . Portions of the code are not shown . . .
proc report data=rptdata.clinics nowd;
column region sex,wt,(n mean) dummy;
define region / group
format=$regname.;
define sex
/ across format=$gender6. 'Gender' ;
define wt
/ analysis
' ';
define dummy
/ computed noprint;
compute dummy;
call define('_c2_','format','2.');
call define('_c3_','format','5.1');
Chapter 7: Extending Compute Blocks 179
call define('_c4_','format','2.');
call define('_c5_','format','5.1');
endcomp;
run;
MORE INFORMATION
CALL DEFINE is discussed in greater detail, relative to the ODS environment, in Section 8.3. A
discussion on determining the indirect column numbers can be found in Section 7.2.4.
SEE ALSO
The online sample program Sample 608 extends the example in this section by placing a
computed variable under an ACROSS variable with an unknown number of columns. Molter
(2006) uses the CALL DEFINE routine to set display attributes. In a very interesting paper, Poppe
(2005) uses the CALL DEFINE routine to bring GRSEG entries into the report.
7.6 COMPUTE Statement Options and Switches
The COMPUTE statement supports only a few options, and these are not needed on a regular
basis. However, you do need to be aware that they exist.
7.6.1 Justification of LINE Statement Text
When the _PAGE_ target is used with a compute block, justification options can also be included
on the COMPUTE statement. These options include RIGHT, CENTER (the default), or LEFT,
and will affect any text that is generated by a LINE statement in that compute block.
In the following example, text is generated in a compute block and added to the report between
the title and the report header. Notice that the text has been left-justified n (to match the title,
since the NOCENTER system option has also been specified).
option nocenter;
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
value $gender
'M'='Male'
'F'='Female';
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Using Justification Options';
proc report data=rptdata.clinics nowd;
column region sex ('Weight' wt wt=wtmean);
define region / group
format=$regname8.;
define sex
/ group
format=$gender.
'Gender';
define wt
/ analysis format=3.
n 'N' ;
define wtmean / analysis format=5.1
180 Carpenter’s Complete Guide to the SAS REPORT Procedure
mean 'Mean';
compute after region;
line ' ';
endcomp;
compute before _page_ / left;
line 'Weight taken during';
line 'the entrance exam.';
endcomp;
run;
n
The compute block is executed at the start of each page, and the LINE statements place the text
between the titles and the headers.
Extending Compute Blocks
Using Justification Options
Weight taken during
the entrance exam.
n
region
Mid West
Gender
Female
Male
Weight
N
Mean
13 163.0
13 182.1
No. East
Female
Male
11
13
117.9
155.3
So. East
Female
Male
4
10
143.0
165.6
Western
Female
Male
4
12
170.0
186.0
The BEFORE _PAGE_ causes the
LINE statements in the compute block
to execute at the top of each page before
the headers are written.
This justification option (the default is CENTER) is independent of the NOCENTER system
option.
7.6.2 Creating Character Variables with the CHARACTER and
LENGTH= Options
When you create a character variable in the DATA step, the attributes of the variable are
determined either indirectly by its usage or directly through declarative statements. In a compute
block that creates a character variable, either a report item or a temporary variable, the process is
similar to that in the DATA step. In addition, the COMPUTE statement supports the
CHARACTER and LENGTH= options, which can be used to assist in this declaration.
The CHARACTER option can be abbreviated as CHAR. The following example creates a
computed variable that contains the concatenated first and last names of the patient n.
Chapter 7: Extending Compute Blocks 181
title1 'Extending Compute Blocks';
title2 'Defining Character Columns';
proc report data=rptdata.clinics(where=(region='2'))
nowd split='*';
column lname fname name wt ht edu;
define lname / order noprint;
define fname / order noprint;
define name / computed 'Patient Name';
define wt
/ display 'Weight';
define ht
/ display 'Height';
define edu
/ display 'Years*Ed.';
compute name / character length=17; n
name = trim(fname) || ' ' ||lname;
endcomp;
run;
Extending Compute Blocks
Defining Character Columns
Patient Name
Teddy Atwood
Linda Haddock
Samuel Harbor
Weight
105
105
105
Height
64
64
64
Years
Ed.
14
14
14
Marcia Ingram
Zac Leader
Sandra Little
Margot Long
Linda Maxwell
Liz Saunders
105
115
105
109
115
105
109
64
64
64
63
64
64
63
14
14
14
12
14
14
12
Y
The attributes of the
computed variable
NAME have been
specified as options on
the compute statement.
n The computed variable NAME is declared to be CHARACTER with a LENGTH of 17. The
default variable type is numeric, and the default length for character variables is 9.
Y Samuel Harbor has two visits, and because last and first names are order variables, the compute
block is executed only for the first occurrence. When this behavior is not desirable, the
example in Section 7.7.2 discusses the process for adding repeated text for ORDER and
GROUP report items.
When you specify a character format using the FORMAT= option on the DEFINE statement
associated with a computed variable, you must also use one or both of these two COMPUTE
statement options. If you fail to do so, the SAS log will show an “Invalid numeric data” note.
Although using both options together is recommended, either is sufficient when a character
format is present on the DEFINE statement. If both a format and a LENGTH= option are present,
the length will be taken from the LENGTH= option.
182 Carpenter’s Complete Guide to the SAS REPORT Procedure
The previous code could have been written as follows:
title1 'Extending Compute Blocks';
title2 'Defining Character Columns';
proc report data=rptdata.clinics(where=(region='2'))
nowd split='*';
column lname fname name wt ht edu;
define lname / order noprint;
define fname / order noprint;
define name / computed format=$17. 'Patient Name';
define wt
/ display 'Weight';
define ht
/ display 'Height';
define edu
/ display 'Years*Ed.';
compute name / char;
name = trim(fname) || ' ' ||lname;
endcomp;
run;
In both examples NAME will have a length of 17 characters.
MORE INFORMATION
The example in Section 7.4.3 creates a computed column using the CHAR and LENGTH=
options.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 85–86) includes an example that shows the use of the
CHARACTER and LENGTH= options, and also includes an example that uses the LENGTH
statement (pp. 125–127).
7.7 Using Logic and SAS Language Elements
Within the compute block we can use most of the power of the DATA step. Functions,
assignment statements, SUM statements, and other executable statements are available to us. Most
of the elements that are not available would not logically be needed in the compute block anyway.
Some of the statements that are not available in the compute block include those that do the
following:
read or write data—e.g., DATA, SET, MERGE, and UPDATE
affect the PDV—e.g., KEEP, DROP, and RETAIN
Many other SAS Language elements are available in the compute block. Of these, some of the
more commonly used statements include the following:
ARRAY
CALL routines and functions
comments (all types are recognized)
Chapter 7: Extending Compute Blocks 183
DO block, iterative DO, DO WHILE, DO UNTIL
(as in the DATA step, all DO loops and DO blocks expect the END statement)
IF-THEN/ELSE and SELECT
%INCLUDE
SUM and assignment
Within these statements you can use SAS language functions in the same manner as you would in
the DATA step.
The use of macro variables and macro calls is the same in the compute block as it is in other
locations within SAS programs. There are some minor resolution differences when a macro
variable is used inside of single quotation marks. However, these will not be of general interest
(see Section 10.2.2).
MORE INFORMATION
Section 2.6.4 introduces the topic of SAS language elements in the compute block. Using the
LINE statement with other SAS language elements can be problematic in a compute block (see
Section 7.8.3).
SEE ALSO
Cody (2004) provides an excellent discussion of SAS language functions. In SAS Technical
Report P-258 (1993, pp. 128–130) there is an example that includes logic in the compute block.
This same report (pp. 133–135) discusses the SAS language elements available in the compute
block. Burlew (2005, p. 45) and Molter (2006) both have an example that uses the IF statement.
7.7.1 Using the SUM Statement with Temporary Variables
The SUM statement can be used as a counter or an accumulator. The form of the statement is as
follows:
variable_name + expression;
The expression can be any valid arithmetic expression, but is typically a constant.
The SUM statement is as handy in the compute block as it is in the DATA step, and as in the
DATA step, it can be used to count the number of items within a group.
In Section 7.2.1, the SUM statement was used to create an OBS column for the entire report. In
the following example, we would like to have a consecutive counter for items within a group.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Using the SUM Statement';
184 Carpenter’s Complete Guide to the SAS REPORT Procedure
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd split='*';
column region cnt clinnum ('
Patient' wt=wtn wt);
define region / group format=$regname.;
define cnt
/ computed ' '; n
define clinnum/ group 'Clinic*Number';
define wtn
/ analysis n 'N';
define wt
/ analysis mean
format=6. 'Mean*Weight';
compute before region;
clincount=0; o
endcomp;
compute before clinnum;
clincount+1; p
endcomp;
compute cnt;
cnt=clincount; q
endcomp;
break after region/ suppress skip;
run;
n The computed variable CNT is used to display the item numbers.
o CLINCOUNT is initialized to 0 before each new REGION. This temporary variable is used
to accumulate the counts. Because CNT is a report item, its value is cleared before each row
is written to the report, and consequently it cannot be used to hold values from one row to the
next.
p The SUM statement is used to accumulate the count of the clinics in the temporary variable
CLINCOUNT.
q The cumulative count in the temporary variable CLINCOUNT is written to the report item
variable CNT so that it can appear on the report.
Extending Compute Blocks
Using the SUM Statement
Patient
region
No. East
So. East
1
2
3
4
5
6
7
8
1
2
3
4
Clinic
Number
011234
014321
023910
024477
026789
031234
036321
038362
033476
043320
046789
049060
N
2
2
4
4
2
4
4
2
4
4
2
4
Mean
Weight
195
195
105
107
115
179
130
112
156
178
160
143
A counter has been
added for each
detail row (within
REGION) of the
report.
Chapter 7: Extending Compute Blocks 185
As is shown in this example, accumulated values are stored in a temporary variable. It is this
variable that is used in the SUM statement, and it is this variable that is used to assign the value to
the report item variable at the appropriate time.
In this example, we could have simplified the code a bit by eliminating the compute block for
BEFORE CLINNUM. Because this compute block p and the compute block for CNT q both
execute once for each detail row (CLINNUM has a usage of GROUP), the SUM statement can be
moved into the second compute block. The new compute block becomes the following:
compute before region;
clincount=0;
endcomp;
compute cnt;
clincount+1;
cnt=clincount;
endcomp;
SEE ALSO
Pass (2000) adds an observation counter to a report in a very succinct explanation of the
difference between report items and temporary variables.
7.7.2 Repeating GROUP and ORDER Variables on Each Row
One of the characteristics of GROUP and ORDER variables is that only the first item of a series
of repeated values is shown. In the example in Section 7.7.1, the value of REGION is shown only
once, on the first report line for each region. Usually this is a preferred behavior; however, you
might want to show the group value on each row.
The discussion of the example that follows touches on some fairly detailed aspects of the compute
block process (especially at p). (This is discussed even more fully in Section 7.7.3) A full
understanding of this process is not required to simply modify PROC REPORT steps. However,
as you write more complex steps you will need to understand this process in greater depth. This
may require that you reread this particular section, and other sections throughout the book that
deal with the PROC REPORT step process. And you might need to reread them multiple times;
the PROC REPORT step process is very much worth studying and understanding.
In the following example, the GROUP variable’s value is repeated on every line. This is
accomplished by the creation of a computed variable Y to hold the name of the region n.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Repeating a Group Name';
proc report data=rptdata.clinics
(where=(region in('1' '2' '3' '4')))
nowd split='*';
column region regname clinnum
('Patient Weight' wt=wtn wt);
define region / group format=$regname. noprint; n
186 Carpenter’s Complete Guide to the SAS REPORT Procedure
define regname/ computed 'Region'; o
define clinnum/ group
'Clinic*Number';
define wtn
/ analysis n
format=5. 'N';
define wt
/ analysis mean
format=4. 'Mean';
compute before region;
* Load the formatted region into a temporary variable;
rname = put(region,$regname.); p
endcomp;
compute regname / character length=12;
* Move region from temporary variable to a report item;
regname = rname; q
endcomp;
break after region/ suppress skip;
run;
n REGION is used to form the groups, but it is not printed.
o The value of the repeated grouping variable is actually displayed through the use of a
computed variable, REGNAME.
p The current formatted value of the grouping variable REGION is stored in the temporary
variable RNAME. The assignment has to be made when the value of REGION is available, and
the easiest time to do this is in a BEFORE REGION compute block. This compute block is
executed on the first detail line of each region. This is important since REGION is a grouping
variable, and thus its value is only available for use in the computed summary information
on this the first row of each group. Placing the assignment statement in the same compute
block as the REGNAME assignment statement q would not work, because as a grouping
variable, REGION is not on every row of the computed summary information; it is only
available for use in an assignment statement when it appears on the actual report row—in a
GROUP or ORDER usage. This means that the value is available only the first time the
GROUP or ORDER value appears.
q The value of the temporary variable RNAME is assigned to the report item REGNAME.
Notice the use of the CHARACTER and LENGTH= options on the COMPUTE statement.
Extending Compute Blocks
Repeating a Group Name
Region
No. East
No. East
No. East
No. East
No. East
No. East
No. East
No. East
So. East
So. East
So. East
So. East
. . . Portions
Clinic
Number
011234
014321
023910
024477
026789
031234
036321
038362
Patient Weight
N Mean
2
195
2
195
4
105
4
107
2
115
4
179
4
130
2
112
The value of REGION is actually the
computed variable REGNAME.
033476
4
156
043320
4
178
046789
2
160
049060
4
143
of the table are not shown . . .
Chapter 7: Extending Compute Blocks 187
SEE ALSO
The online sample program Sample 611 gives a similar but simpler example. The two papers by
Chapman (2002 and 2003) each show several examples of compute blocks that demonstrate
similar techniques. Chung and Dunn (2005) use similar, although more sophisticated, techniques
when they discuss how to create a page x of y counter. Molter (2006) discusses the problem of
repeating group headings when some rows are not displayed. Russ Lavery’s “An Animated Guide
to the SAS REPORT Procedure,” which is on the CD that accompanies this book, discusses these
issues in more detail.
7.7.3 Counting Items across Page Breaks in the LISTING
Destination
Unless you are using BY groups or writing to a monospace destination, such as LISTING,
physical page breaks have limited meaning. This is because the concepts of paging and page
breaks work differently for ODS destinations, such as HTML, RTF, and PDF. In the following
example, the PS= option is used to control page breaks. However, PS= is ignored for other ODS
destinations, so you should not expect this example to work anyplace except for the LISTING
destination. Consequently, the example in this section has little practical value. However, even in
this limited context, this example does illustrate and reinforce concepts related to the timing issues
surrounding the execution of compute blocks.
This example illustrates how to conditionally perform an operation within a compute block at the
start of a page. This operation might include the counting of the page itself or the generation of
page-specific titles or footnotes. More importantly, it demonstrates the issues associated with the
use of _PAGE_ as the target on the COMPUTE statement.
In the example in Section 7.7.1, the SUM statement is used to count items within a group. In the
following example, the SUM statement is used to count items within a group as well as on a page.
The counters are used to demonstrate the timing and execution of compute blocks when the
_PAGE_ target is used. Ostensibly this PROC REPORT repeats the value of the grouping variable
(REGION) at the top of each page, even if the page break occurs within a group. Of course, this
happens automatically with grouping variables, but then where would be the fun of this example?
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Counting Group and Page Breaks';
options ps=15;
proc report data=rptdata.clinics
out=outreg
nowd split='*';
column region regname clinnum
('Patient*Weight' wt=wtn wt) rcnt pgcnt;
define region / group format=$regname. noprint;
define regname/ computed 'Region';
define clinnum/ group 'Clinic*Number';
define wtn
/ analysis n
n
188 Carpenter’s Complete Guide to the SAS REPORT Procedure
define wt
format=1. 'N';
/ analysis mean
format=4. 'Mean';
compute before region;
rname = put(region,$regname.);
* Initialize within region line counter;
rcounter=0; o
endcomp;
compute after _page_; p
* Initialize within page line counter;
pcounter=0; q
endcomp;
compute regname / character length=12; r
rcounter+1;
pcounter+1;
* Create region name at the start of region
* and/or page;
if rcounter=1 or pcounter=1 then regname = rname;
endcomp;
compute rcnt; t
rcnt = rcounter;
endcomp;
compute pgcnt; _
pgcnt = pcounter;
endcomp;
break after region/ suppress skip;
run;
options ps=56;
proc print data=outreg;
run;
s
n The two computed variables RCNT and PGCNT are added to the COLUMN statement.
Normally we would probably not show these columns (although the group counter is used in
Section 7.7.1), but they are included here so that we can see how their values change.
o The counter for the individual lines within REGION is initialized.
p When PROC REPORT determines the need for a page break (in destinations that support page
breaking via PS=), this compute block is executed, and the counter that counts lines on a page
is initialized.
q The page counter is reset to zero after the page is filled, and is then incremented r in the same
compute block where the assignment is made for REGNAME.
r The compute block for REGNAME is executed for each report row. It is here that we
increment the two counters for the group and page. We also assign the region name to
REGNAME.
s The output data set (WORK.OUTREG) created from the report rows is printed to help us see
what is going on row by row.
Chapter 7: Extending Compute Blocks 189
t The value of the temporary variable RCOUNTER is transferred from memory to the report
item RCNT.
_ The value of the temporary variable PCOUNTER is transferred from memory to the report
item PGCNT.
Because the PAGESIZE= option (PS=15) has been set, a new page is written after every 15 lines.
Here is the first report page:
Extending Compute Blocks
Counting Group and Page Breaks
Region
Mid West
Clinic
Number
051345
054367
057312
059372
063742
063901
065742
066789
082287
Patient
Weight
N Mean
2
215
2
160
2
158
2
98
4
221
4
187
4
151
2
175
2
158
rcnt
1
2
3
4
5
6
7
8
9
pgcnt
2
3
4
5
6
7
8
9
10
Notice that while RCNT is 1 for the first row, PGCNT is 2. Because these two counters are
always incremented together, we might expect them to have the same value on the first line.
Examining the output data set created from the report rows might help to explain why this is not
the case.
Extending Compute Blocks
Counting Group and Page Breaks
Obs region
1
5
2
5
3
5
4
5
5
5
6
5
7
5
8
5
9
5
10
5
11
5
regname
clinnum
Mid West 051345
054367
057312
059372
063742
063901
065742
066789
082287
082287
wtn
26
2
2
2
2
4
4
4
2
2
2
wt
rcnt pgcnt
172.538 1
1
215.000 1
2
160.000 2
3
158.000 3
4
98.000 4
5
220.500 5
6
187.000 6
7
151.000 7
8
175.000 8
9
158.000 9
10
158.000 9
10
_BREAK_
region
_PAGE_
As we go through the processing steps, remember that for a given report row, compute blocks
associated with report items are executed before compute blocks containing BEFORE or AFTER.
190 Carpenter’s Complete Guide to the SAS REPORT Procedure
OBS=1 COMPUTE BEFORE REGION summary line
The COMPUTE BEFORE REGION compute block causes this line to appear, and this is
the summary line calculated BEFORE the region. Each of the non-summary compute
blocks (r, t, and _) are executed. Although RCOUNTER and PCOUNTER are not
explicitly set to 0 except at o and q, they each appear in a SUM statement and are
therefore automatically initialized to 0. After execution of the REGNAME compute block
r both counters contain a 1, as do the report items RCNT and PGCNT. Finally, the
summary line compute block o is executed, and RCOUNTER is reset to 0 (we cannot
see this on the first line of the output data set, but it is reflected in OBS=2).
OBS=2
This is the first detail row of the report for this value of REGION. First the report item
compute blocks are executed, incrementing the two counters r, and the resulting values
are placed in RCNT t and PGCNT _.
OBS=10
This is the last row of the report that will fit on the current page in the LISTING
destination. However, this is not the last row for this region, so we want the next row in
the table (first row on the second page) to have a value for REGNAME.
OBS=11
Notice that _BREAK_ takes on the value of _PAGE_. This row in the output data set is
generated because we have used the _PAGE_ target on a compute block p, and as a
result, the compute block for AFTER _PAGE_ is executed on this row. Again, the
compute blocks associated with the report items are executed first, incrementing the
counters and setting values for RCNT and PGCNT. After PGCNT has been assigned a
value _, the compute block for AFTER _PAGE_p executes for the first time, and the
PCOUNTER temporary variable is reset to 0 q. It is of course too late for this to be
reflected in the report item variable PGCNT, because its compute block _executes first
(before q).
Obs region regname
11
5
12
5
Mid West
13
5
14
1
15
1
No. East
16
1
17
1
18
1
19
1
20
1
21
1
22
1
23
1
No. East
clinnum
082287
084890
011234
014321
023910
024477
026789
031234
036321
036321
038362
wtn
2
2
26
24
2
2
4
4
2
4
4
4
2
wt rcnt pgcnt
158.000
9
10
162.000
10
1
172.538
11
2
138.167
12
3
195.000
1
4
195.000
2
5
105.000
3
6
107.000
4
7
115.000
5
8
178.500
6
9
130.000
7
10
130.000
7
10
112.000
8
1
_BREAK_
_PAGE_
region
region
_PAGE_
OBS=12
This is the first observation on the second page. When the REGNAME compute block r
executes, PCOUNTER is increased to 1. The IF statement is true, and a value is written
to REGNAME.
Chapter 7: Extending Compute Blocks 191
OBS=13 and 14
These are the AFTER and BEFORE summary rows for REGION. Notice the value of
_BREAK_. OBS=13 is generated by the BREAK statement, which has a target of
AFTER REGION, and OBS=14 is generated by the BEFORE REGION compute block
o. After the values for RCNT and PGCNT have been determined in OBS= 14, the
BEFORE REGION compute block executes and resets the value of RCOUNTER to 0.
OBS=15
RCOUNTER is incremented to 1, and this value is written to the report item RCNT.
Here is the second page of the report:
Extending Compute Blocks
Counting Group and Page Breaks
Region
Mid West
No. East
Clinic
Number
084890
Patient
Weight
N Mean
2
162
011234
014321
023910
024477
026789
031234
036321
2
2
4
4
2
4
4
195
195
105
107
115
179
130
rcnt
10
pgcnt
1
1
2
3
4
5
6
7
4
5
6
7
8
9
10
Here are the two major lessons from this section:
1. For a given report row, report item compute blocks execute first, and compute blocks
containing BEFORE and AFTER execute second.
2. The output data set reflects the end result of all of the compute block execution process.
Well actually, perhaps a third lesson is that it sometimes can be a bit difficult to determine exactly
what did take place and in what order.
SEE ALSO
Issues associated with controlling page breaks are also discussed by Guttadauro (2003). Variables
created in a DATA step are used by Humphreys (2006) to control page breaks. Russ Lavery’s “An
Animated Guide to the SAS REPORT Procedure, which is included on the CD that accompanies
this book, also includes a number of examples that cover compute block timing issues.
7.8 Doing More with the LINE Statement
The LINE statement is unique to the REPORT procedure step. When it is used in the compute
block with a location designation (BEFORE or AFTER), it is roughly analogous to the PUT
statement in the DATA step. In Section 2.6.1, the LINE statement is introduced, and it is used to
write constant text in Section 2.6.2. Justification options are applied to LINE statement text in
Section 7.6.1. The LINE statement can also be used to write computed values.
192 Carpenter’s Complete Guide to the SAS REPORT Procedure
SEE ALSO
Guttadauro (2003) uses the LINE statement and preprocessing to control page numbering when
variables wrap. Dunn (2004) adds text lines to augment the number of titles.
7.8.1 Creating Group Summaries
Because the LINE statement is capable of writing both text and the values of variables, it is
commonly used to write summary information on the final report output.
In the clinical study that we have been following, the manager of each of the four primary areas
(or groups of regions) is responsible for recruiting eight clinics with an average of five patients for
each clinic. The following report assesses the status for each area as the study progresses.
The LINE statement is used to write out the statistics associated with each area. The statistics
themselves are calculated in compute blocks.
Extending Compute Blocks
Using LINE for Group Totals
Clinic Patient
Number
Count
051345
2
054367
2
057312
2
059372
2
063742
4
063901
4
065742
4
066789
2
082287
2
084890
2
Total of 10 clinics is 125.0% of target
Patient enrollment is
26
Per clinic this is
52.0% of target
region
Mid West
No. East
011234
2
014321
2
. . . Portions of the report are not shown . . .
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Using LINE for Group Totals';
proc report data=rptdata.clinics
nowd split='*';
column region clinnum n; n
define region / group
format=$regname8.;
The summary following
each set of clinics within
a regional area is written
using LINE statements in
a COMPUTE AFTER
block.
Chapter 7: Extending Compute Blocks 193
define clinnum / group
'Clinic*Number';
define n
/ width=7 'Patient*Count';
compute before region;
clincnt = 0;
patcnt = 0;
endcomp;
o
compute n; p
if _break_= ' ' then do; q
* Within region patient count;
patcnt + n; r
* Clinic count;
clincnt + 1; s
end;
endcomp;
compute after region;
* Fraction of target (8);
clinpct = clincnt/8; t
* Fraction of target (5);
patpct = patcnt/clincnt/5; _
line @5 'Total of ' clincnt 3. ' clinics is ' `
clinpct percent8.1 ' of target';
line @5 'Patient enrollment is ' patcnt 4.;
line @5 'Per clinic this is ' patpct percent8.1 ' of target';
line ' ';
line ' ';
endcomp;
run;
n The N statistic is added to the COLUMN statement. We can now associate DEFINE and
COMPUTE statements p with it.
o Before processing each region, the counters are reset to 0.
p The compute block that is tied to the N statistic is used to execute the SUM statement
accumulators.
q This compute block is executed for each report row, including summary rows.
Fortunately, through the use of the _BREAK_ variable, it is fairly easy to distinguish between
detail and summary rows. _BREAK_ is nonmissing for summary rows, and has a missing
value for detail rows. Because we want to increment our counters only for detail rows, we need
to execute the SUM statements (r and s) on report rows where _BREAK_ is missing.
r Because the table has been summarized to the clinic level (CLINNUM has a define type of
GROUP), each row represents the number of patient visits within the clinic, and this number is
stored in the report item N.
s Each row in the final report represents one clinic.
t CLINCNT is the total number of clinics within the region. The ratio of this count to the target
value (8 clinics) is calculated here.
_ The fraction of patients per clinic relative to the target value (5 patients) is calculated.
` The LINE statement is used to write out the calculated values.
194 Carpenter’s Complete Guide to the SAS REPORT Procedure
Although you can use the LINE statement to write out a value contained in either a temporary
variable or a report item variable, you cannot specify the calculations themselves in the LINE
statement. In this example, the calculations of CLINPCT t and PATPCT _ could not have been
placed on the LINE statements.
You might also notice that when variables are used on the LINE statement, they are always
followed by a format. Although a format is not required on the PUT statement, it is required on
the LINE statement.
The compute block at p is specified as
compute n;
If instead it had been specified as
compute clinnum;
the SUM statement for PATCNT r would have failed. This is because N is to the right of
CLINNUM on the COLUMN statement. If the table were to have N to the left of the clinic
number (column region n clinnum; n) then either compute block specification would
have worked.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 111–113) and Burlew (2005, pp. 118–122) both have
examples that use the LINE statement to build summaries between groups.
7.8.2 Adding Repeated Characters
The HEADLINE option (Section 4.1) and the ability to place repeating characters in spanning
headers (Section 4.3.1) are only available for the LISTING destination. This functionality is not
generally needed in the other destinations; however, when you need it, you can use the LINE
statement to simulate some of this functionality for other destinations.
To create a string of 30 dashes is as simple as specifying the LINE statement as follows:
line @2 '------------------------------------------';
The same syntax that is used in the PUT statement to create repeated strings can also be used in
the LINE statement. The previous LINE statement could be simplified as follows:
line @2 30*'-';
When using this syntax, whatever quoted text follows the asterisk (*) is repeated the specified
number of times. Although the REPEAT function cannot be used in the LINE statement, it can be
used in a compute block. Consequently, another way of specifying a series of dashes could be
something like the following, which first creates the temporary variable STR to hold the 30
dashes (notice that the REPEAT function expects one less than the total number):
str = repeat('-',29);
line @2 str $30.;
In the LINE statement, variable names must always be followed by a format. This is not the case
for constant text.
Chapter 7: Extending Compute Blocks 195
The following example, which builds on the example in Section 7.8.1, adds repeated text after the
summary of each region and also simulates the HEADLINE o and HEADSKIP p options.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Using LINE to Add Repeated Text';
proc report data=rptdata.clinics
nowd split='*';
column region clinnum n;
define region / group
format=$regname8.;
define clinnum / group
'Clinic*Number';
define n
/ width=7 'Patient*Count';
compute before; n
line @3 8*'______';
line ' '; p
endcomp;
o
compute before region;
clincnt = 0;
patcnt = 0;
endcomp;
compute n;
if _break_= ' ' then do;
patcnt + n;
clincnt + 1;
end;
endcomp;
compute after region;
clinpct = clincnt/8;
patpct = patcnt/clincnt/5;
line @5 'Total of ' clincnt 3. ' clinics is '
clinpct percent8.1 ' of target';
line @5 'Patient enrollment is ' patcnt 4.;
line @5 'Per clinic this is ' patpct percent8.1 ' of target';
line @3 8*'------'; q
line ' ';
endcomp;
run;
ods html close;
n The COMPUTE BEFORE block executes after the headers have been written and before
any of the report rows.
o This line of underscores (8 * 6 underscores) simulates the HEADLINE option. Notice that no
format follows the constant text in the LINE statement.
196 Carpenter’s Complete Guide to the SAS REPORT Procedure
p This blank line simulates the HEADSKIP option.
q A line of 48 (8 * 6) dashes also follows the summary text.
Extending Compute Blocks
Using LINE for Group Totals
Clinic
Number
region
Patient
Count
________________________________________________
Mid West 051345
2
054367
2
057312
2
059372
2
063742
4
063901
4
065742
4
066789
2
082287
2
084890
2
Total of 10 clinics is 125.0% of target
Patient enrollment is
26
Per clinic this is
52.0% of target
o
------------------------------------------------
q
No. East
011234
2
014321
2
. . . Portions of the table are not shown . . .
In the previous example, the programmer already knows that the width of each of the lines is to be
48 (8 * 6) characters. When you automate the step, you sometimes might not know this number
ahead of time. In that case, the length of the line could be based on the width of the page itself. In
the LISTING destination, the width of the page is held in the LINESIZE= system option, which
can be abbreviated as LS. This value can be retrieved using the GETOPTION function, and this
means that you can create a set of repeated characters that span the page.
The following code creates a string of characters (stored in STR) that is 4 characters shorter than
the width of the page (LS).
str=repeat('_',getoption('LS') - 5);
line @3 str $;
We could also create a line of the same width by storing the number of times to repeat the
character in a macro variable (&PGWIDTH).
%let pgwidth = %eval(%sysfunc(getoption(LS)) - 4);
The LINE statement could then be written without creating the temporary variable or the
REPEAT function:
line @3 &pgwidth*'_';
SEE ALSO
SAS Technical Report P-258 (1993, p. 216) has an example of the use of repeated characters in
the LINE statement.
Chapter 7: Extending Compute Blocks 197
7.8.3 Understanding LINE Statement Execution
Although the LINE statement initially seems to be very similar to the DATA step PUT statement,
there are important differences that can, at the very least, cause consternation on the part of the
programmer.
The primary difference is in the way that PROC REPORT executes the LINE statements. The
SAS language elements (statements, functions, etc.) are executed in sequence within the compute
block in essentially the same way as they are in the DATA step. The LINE statements are not
executed in sequence with the other statements. In fact, PROC REPORT separates out the LINE
statements from the SAS language elements, which are then executed. After all of the SAS
language elements in the compute block have been executed, PROC REPORT then executes all
the LINE statements in the order in which they appear in the compute block.
This timing means that since all of the SAS language elements are executed before any of the
LINE statements, you can never conditionally execute a LINE statement with an IF-THEN/ELSE
statement, nor can you use a LINE statement inside of a DO loop.
As if all of this were not confusing enough, things can get a bit more convoluted when a LINE
statement is to write the value of a report item or of a temporary variable. The general rule, as
stated previously, is that assignment statements, even those that follow the LINE statement in the
compute block, are executed first. The exception is for temporary character variables. When
temporary character variables appear on the LINE statement, the variables are evaluated when the
LINE statement is encountered, even though the LINE statement itself is actually executed later.
The following silly report demonstrates these rules.
title1 'Extending Compute Blocks';
title2 'Understanding the LINE Statement';
title3 'Just to show the LINE Execution';
proc report data=sashelp.class nowd;
column age sex,height;
define age
/ group ;
define sex
/ across ;
define height / mean;
rbreak after / summarize;
compute after;
line ' ';
value = 'DEF'; n
count = 5678; o
line
line
line
line
'Show letters ' value $3.; p
'Show count ' count 5.1; q
'Show M HT ' _c3_ 5.1; r
' ';
do cnt = 1 to 5;
line 'Loop counter is ' cnt 3.;
end;
s
198 Carpenter’s Complete Guide to the SAS REPORT Procedure
if 1 = 2 then line 'This is true: 1 = 2';
value = 'ABC'; _
count = 1234; v
_c3_ = 11.11; w
endcomp;
run;
t
n The temporary character variable VALUE is given the value of ‘DEF’. This is changed later to
‘ABC’ _, presumably before the LINE statement p is executed.
o The numeric temporary variable COUNT is initialized. This value is later changed to 1234 v,
and it is that value that shows in the report, even though the LINE statement q appears earlier.
p This LINE statement displays the value of ‘DEF’ instead of ‘ABC’.
q The value assigned in v is displayed, even though the assignment statement is after the LINE
statement.
r The report item value has been changed at w. This is reflected in the report and in the result of
the LINE statement.
sThe DO loop executes five times, leaving the temporary variable CNT with a value of 6. The
LINE statement executes only once (outside of the loop).
t The IF expression is clearly false; however, the LINE statement executes anyway. You cannot
conditionally execute LINE statements.
_ Although this assignment statement executes before the LINE statement p that uses VALUE,
the variable value on the LINE statement has already been resolved.
v The value for COUNT is redefined. This is the value that is used by the LINE statement q.
w The value of a report item (mean male height) is changed. This new value appears twice in the
report.
Extending Compute Blocks
Understanding the LINE Statement
Just to show the LINE Execution
Age
11
12
13
14
15
16
Sex
F
M
Height
Height
51.3
57.5
58.05 60.366667
60.9
62.5
63.55
66.25
64.5
66.75
.
72
60.588889
11.11
Show letters DEF
Show count 1234
Show M HT
11.1
n p
q v
r w
Loop counter is
6
This is true: 1 = 2
s
t
w
Chapter 7: Extending Compute Blocks 199
It is generally considered a good programming practice to always put all of the LINE statements
at the end of the compute block as a visual reminder of their order of execution. The previous
compute block could/should be rewritten as:
compute after;
do cnt = 1 to 5;
end;
if 1 = 2 then ;
value = 'ABC';
count = 5678;
count = 1234;
_c3_ = 11.11;
value = 'DEF';
line ' ';
line 'Show
line 'Show
line 'Show
line ' ';
line 'Loop
line 'This
endcomp;
letters ' value $3.;
count ' count 5.1;
M HT ' _c3_ 5.1;
counter is ' cnt 3.;
is true: 1 = 2';
Obviously the compute block is now full of silly stuff (actually it was there all along, but now we
can see it). The point is that LINE statements should appear at the end of the compute block. If
you are always in the habit of putting them last in the compute block, you will not get caught up
in the odd behaviors and rules demonstrated in this example.
7.9 Examples of Common Tasks
Although the types of reports in this section are fairly common, it is unlikely that they will be
“just what you need.” They are presented here as additional examples of various uses of compute
blocks. Understanding how and why these examples work is much more important than exactly
what they do or produce.
These examples take advantage of a variety of the techniques that are discussed in the previous
sections of this chapter.
Section 7.9.1
Section 7.9.2
Section 7.9.3
In a report of multiple pages, a common value (a grand total) is written
on each page as part of a header.
At times we need to combine multiple values into a single column or field.
In this example we combine the mean and standard deviation into one
column that takes the form of
mean (standard_deviation).
Combining numeric values as is done in Section 7.9.2 is even more
interesting when they are nested within an ACROSS variable. In a situation
such as the one shown in this example, columns must be addressed using
absolute column numbers (indirect variable names of the form _Cxx_).
200 Carpenter’s Complete Guide to the SAS REPORT Procedure
7.9.1 Writing a Grand Total on Every Page
In this report, we want to summarize the data so that we will have patients summarized within
clinics and clinics within regions with one page per region. Nested summaries like this are easy to
do with BREAK BEFORE and BREAK AFTER statements; however, we would also like to show
the overall study summary (RBREAK) on each page, not just before or after the report t. Just to
make things a bit more interesting, we would like the two grouping variables (REGION and
CLINNUM) to appear stacked in the same column. _
The target final report will look something like this:
Extending Compute Blocks
Repeating Report Wide Totals
( N, MEAN )
Study Weights (80,161.78)
Study Heights (80,67.45)
_
Western
093785
094789
095277
107211
108531
t
Patient
Weight
Height
N
Mean
N Mean
16
2
2
6
2
4
182.0
177.0
185.0
192.3
177.0
170.0
16
2
2
6
2
4
67.3
65.0
70.0
67.3
69.0
66.0
q
The study-wide statistics for height and weight are each written as a single value, which includes
parentheses, text, and converted numeric values t. These text values are constructed in a
compute block s with the use of the PUT function to convert numeric values to characters, and
the CATS function to do the actual concatenation. This multiple-value construction process is
described in more detail in Section 7.9.2.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Repeating Report Wide Totals';
proc report data=rptdata.clinics
nowd;
column region clinnum name n
('Patient'
('Weight' wt=wtn o wt)
('Height' ht=htn o ht));
define region / group format=$regname. noprint;
define clinnum/ group noprint;
define name
/ computed ' ' right; p
define wtn
/ analysis n
format=2. 'N';
Chapter 7: Extending Compute Blocks 201
define wt
define htn
define ht
/ analysis mean
format=5.1 'Mean';
/ analysis n
format=2. 'N';
/ analysis mean
format=4.1 'Mean';
break before region / summarize;
break after region / page; r
q
* Compute the Report totals;
compute before; s
* Hold values of interest in DATA step variables;
allwtn = put(wtn,3.);
allwt = put(wt.mean,6.2);
linevalwt = cats('Study Weights (',allwtn,',',allwt,')');
allhtn = put(htn,3.);
allht = put(ht.mean,6.2);
linevalht= cats('Study Heights (',allhtn,',',allht,')');
endcomp;
* Write the overall statistics;
compute before _page_; t
line @15 '( N, MEAN )';
line @1 linevalwt $35.;
line @1 linevalht $35.;
line ' ';
endcomp;
* Determine the line header;
compute name / char length=8; _
if clinnum = ' ' then name=put(region,$regname.);
else name = clinnum;
endcomp;
run;
v
n The variable NAME must appear to the right of both REGION and CLINNUM in the
COLUMN statement, or the assignment statements in its compute block u will not work.
o An alias (WTN and HTN) is declared to hold the count (N) for the two analysis variables.
p The NAME column is used to display both the region name and the clinic number in the final
table _.
q For this report, we would like the region’s summary line to appear before the detail lines of
that same region.
r The page break is inserted after each region has been completed. Because the SUMMARIZE
option does not appear on this statement, no summary line is generated. Using two BREAK
statements, one before and one after region, gives us quite a bit of added flexibility.
s The overall report-wide statistics are saved in temporary variables for use on each page t.
These values are created only once for the entire report, but they are used on each page.
tThe saved report-wide values s are written out at the start of each new page.
202 Carpenter’s Complete Guide to the SAS REPORT Procedure
_ The value for NAME is assigned from either REGION or CLINNUM, depending on the type
of summary or detail row.
v The value of CLINNUM is missing for the row generated by the BREAK BEFORE
REGION statement.
The output page for the Western region is shown below. Notice that the report-wide (across all
regions) information appears in the title area t. The region name (Western) appears in the same
column as the individual clinic numbers. This takes less horizontal space than the report in
Section 7.7.3.
Extending Compute Blocks
Repeating Report Wide Totals
( N, MEAN )
Study Weights (80,161.78)
Study Heights (80,67.45)
Western
093785
094789
095277
107211
108531
t
Patient
Weight
Height
N
Mean
N Mean
16 182.0 16 67.3
2 177.0
2 65.0
2 185.0
2 70.0
6 192.3
6 67.3
2 177.0
2 69.0
4 170.0
4 66.0
Keep in mind that this COMPUTE BEFORE _PAGE_ technique might not work in destinations
other than LISTING. For example, in the RTF destination, you have much less control over
vertical pagination; therefore, there is no guarantee that this technique would work for the RTF
destination. For other destinations, a related potential solution would be to use BY variables to
force the page breaks (see Section 6.6.2).
7.9.2 Combining Values into One Field or Column
Sometimes for convenience we would like to display two (or more) values in the same field.
Although not described in detail, this was done in the example in Section 7.9.1. In the following
example, we would like to display the mean along with its standard deviation in parentheses
within the body of the report.
proc format;
value $regname
'1','2','3' =
'4'
=
'5' - '8'
=
'9', '10'
=
run;
'No. East'
'So. East'
'Mid West'
'Western';
title1 'Extending Compute Blocks';
title2 'Combining Values in One Field';
proc report data=rptdata.clinics
nowd;
Chapter 7: Extending Compute Blocks 203
column region n
('Patient'
('
Weight' wt=wtmean wt wtval) n
(' Height' ht=htmean ht htval));
define region / group format=$regname. 'Area';
define n
/ format=3. ' N'; o
define wtmean / analysis mean noprint; p
define wt
/ analysis std noprint; p
define wtval / computed ' Mean (SD)'; q
define htmean / analysis mean noprint;
define ht
/ analysis std noprint;
define htval / computed 'Mean (SD)';
* Combine the WT values; r
compute wtval / char length=15;
wtval = cats(put(wtmean,5.1),' (',
put(wt.std,7.1),')');
endcomp;
* Combine the HT values;
compute htval / char length=15;
htval = cats(put(htmean,5.1),' (',
put(ht.std,7.1),')');
endcomp;
run;
n The analysis variable, WT, is specified along with an alias, WTMEAN, and a computed
variable that will contain the combined values.
o The N statistic is requested directly. This value counts observations and reflects the number
used to calculate the statistics, assuming that there are no missing values.
p The MEAN and STD are calculated but are not printed. Instead their values are loaded into the
computed variable WTVAL (q and r).
q The computed variable that will hold the combined values is defined.
r The CATS function is used to concatenate the mean (WTMEAN), standard deviation
(WT.STD) and the parentheses. WTVAL is a character variable, and the COMPUTE statement
includes the use of the CHAR and LENGTH= options.
Extending Compute Blocks
Combining Values in One Field
Patient
Area
Mid West
No. East
So. East
Western
N
26
24
14
16
Weight
Mean (SD)
172.5(34.4)
138.2(38.5)
159.1(23.7)
182.0(22.7)
Height
Mean (SD)
66.8(3.5)
67.3(4.4)
69.0(2.6)
67.3(2.2)
This technique gives you a great deal of flexibility when building the value that is to be displayed.
204 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
The techniques associated with combining variable values are also discussed in Sections 7.9.3 and
10.1.3.
7.9.3 Combining Values with Nested ACROSS Variables
The ACROSS define type is used when we want values of classification variables to be next to
each other horizontally. In this example, we want to nest ACROSS variables, and we want to
create a concatenated column as we did in Section 7.9.2. Because of the nested ACROSS
variables, we need to use absolute column numbers, which use the _Cxx_ column designations.
proc format;
value $regname
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
value $gender
'm', 'M' = 'Male'
'f', 'F' = 'Female';
value birthgrp
'01jan1945'd - '31dec1959'd = '
Boomer' n
other
= 'Non-Boomer';
run;
title1 'Extending Compute Blocks';
title2 'Combining Values in ACROSS Columns';
proc report data=rptdata.clinics
out=outreg o
nowd;
column region n sex,dob, p
(wt=wtmean wt wtval);
define region / group format=$regname. 'Area';
define n
/ format=3. ' N';
define sex
/ across format=$gender. ' ';
define dob
/ across format=birthgrp. ' ';
define wtmean / analysis mean noprint; q
define wt
/ analysis std noprint;
define wtval / computed ' Mean (SD)';
* Combine the WT values;
compute wtval / char length=12; r
_c5_ = cats(put(_c3_,5.1),' (', s
put(_c4_,7.1),')'); t
_c8_ = cats(put(_c6_,5.1),' (',
put(_c7_,7.1),')');
_c11_ = cats(put(_c9_,5.1),' (',
put(_c10_,7.1),')');
_c14_ = cats(put(_c12_,5.1),' (',
put(_c13_,7.1),')');
endcomp;
run;
proc print data=outreg;
run;
Chapter 7: Extending Compute Blocks 205
n Leading spaces are added to help center the text.
oWhen working with columns during the program development phase, it is sometimes helpful to
print out the output data set of the report so that the column numbers can be accurately
determined.
pThe comma is used to nest the three weight variables within date of birth (DOB), which
is nested within SEX. Notice that parentheses are placed around the three weight
variables to form a group that can be nested within DOB.
q The statistics for weight (WT and WTMEAN) are calculated, but the NOPRINT option
prevents their display.
r Notice that although we use the compute block for WTVAL, this variable is never actually
created. Instead, because it is nested under ACROSS classification variables, we assign the
computed values directly into the appropriate columns using the absolute column numbers.
sThe mean, _C3_, is added to the computed character string that will contain the statistics. We
cannot address the mean using the alias WTMEAN because of the nesting across age groups.
This solution only works because we know what is contained in each of the columns. Notice
that although WTMEAN is not displayed on the final report (because it has been defined with
the NOPRINT option q), it still occupies columns (_C3 _, _C6 _, _C9 _, and _C12 _) on the
output data set.
tThe standard deviation, _C4_, is added to the computed value. The variable WT.STD would
not be available for use.
The output data set WORK.OUTREG, can provide a snapshot of the end result of the processing
of the compute blocks during the report row phase, and it can also be used to help determine the
column numbers that are used in the WTVAL compute block. The absolute column numbers
themselves are available for use because they have already been determined during the setup
phase.
Extending Compute Blocks
Combining Values in ACROSS Columns
Obs region
1
2
3
4
5
1
4
10
n
24
24
14
16
_C3_
_C4_
140.667
119.222
139.000
177.000
36.9504
20.6505
13.8564
0.0000
Obs
_C9_
_C10_
1
2
3
4
161.000
121.286
170.667
187.800
20.2287
27.8132
27.6381
37.2451
_C5_
140.7(37.0)
119.2(20.7)
139.0(13.9)
177.0(0.0)
_C11_
161.0(20.2)
121.3(27.8)
170.7(27.6)
187.8(37.2)
_C12_
200.143
195.000
158.000
184.714
_C6_
_C7_
_C8_
172.625
112.000
155.000
163.000
34.8668
0.0000
.
0.0000
172.6(34.9)
112.0(0.0)
155.0(.)
163.0(0.0)
_C13_
33.4187
0.0000
18.9209
13.8770
_C14_
200.1(33.4)
195.0(0.0)
158.0(18.9)
184.7(13.9)
_BREAK_
206 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here is the final REPORT table:
Extending Compute Blocks
Combining Values in ACROSS Columns
Female
Area
Mid West
No. East
So. East
Western
N
24
24
14
16
Boomer
Mean (SD)
140.7(37.0)
119.2(20.7)
139.0(13.9)
177.0(0.0)
Non-Boomer
Mean (SD)
172.6(34.9)
112.0(0.0)
155.0(.)
163.0(0.0)
Male
Boomer
Mean (SD)
161.0(20.2)
121.3(27.8)
170.7(27.6)
187.8(37.2)
Non-Boomer
Mean (SD)
200.1(33.4)
195.0(0.0)
158.0(18.9)
184.7(13.9)
This final report shows how PROC REPORT can provide some of the same functionality and
ability to create cross-tabular reports as PROC TABULATE (albeit with different syntax).
MORE INFORMATION
The alignment of decimal points in derived composite columns such as these can be problematic;
see Section 10.1.3 for a further discussion on this topic.
SEE ALSO
SAS Technical Report P-258 (1993, pp. 83–84) includes an example of computed variables that
share columns with nested variables. Burlew (2005, pp. 128–130) uses the ARRAY statement
with a DO loop to calculate percentages using indirect column references.
7.9.4 Calculating a Weighted Mean
The FREQ statement (see Section 6.7) can be used to specify a variable that contains a row
frequency. This frequency is then applied to all the analysis variables for that row. Sometimes we
have a value that we would like to apply as a weight, but to only one other variable. To do this we
need to perform the calculations ourselves.
In the following example we would like to use the number of items sold (QUANTITY) and the
unit price for that quantity (PRICE) to calculate the weighted average and the total revenue. The
data has been simplified so that we can more easily see what is going on and what is going wrong.
The data shows that one unit was sold at $1.50, ten for $1.00, and 100 units were sold for $0.50.
data temp;
input product $ quantity price;
label product = 'Product Code'
quantity = 'Number Sold'
price
= 'Unit price';
datalines;
a 1 1.5
a 10 1.0
a 100 .5
;
run;
The first attempt to determine the total revenue and average price uses the FREQ statement (see
Section 6.7).
Chapter 7: Extending Compute Blocks 207
title1 'Extending Compute Blocks';
title2 'Calculating a Weighted Mean';
title3 'Using the FREQ statement';
proc report data=temp nowd;
columns product quantity price price=revenue; n
define product / display;
define quantity / analysis sum;
define price / analysis mean format=dollar5.2 width=7;
define revenue / sum 'Total Revenue' format=dollar6.2 width=7;
freq quantity; o
rbreak after / summarize dol;
run;
n REVENUE is defined as an alias of PRICE.
o The FREQ statement is used to identify the variable that contains the frequency.
The total revenue and weighted average unit price has been calculated correctly; however, since
we are also displaying the QUANTITY (the FREQ variable) that value is now incorrectly
displayed.
Extending Compute Blocks
Calculating a Weighted Mean
Using the FREQ statement
Product
Code
a
a
a
Number
Sold
1
100
10000
=========
10101
Unit
price
$1.50
$1.00
$0.50
=======
$0.55
The FREQ variable is applied
to all analysis variables
including itself
(QUANTITY**2). The
weighted average price and
total revenue are correct.
Total
Revenue
$1.50
$10.00
$50.00
=======
$61.50
We can take control of the weighting of the calculations by using compute blocks to calculate the
revenue.
title3 'Computed Values';
proc report data=temp nowd;
columns product quantity price revenue p;
define product / display;
define quantity / analysis sum;
define price / analysis mean format=dollar5.2 width=7;
define revenue / computed 'Total Revenue' format=dollar6.2 width=7;
compute revenue;
revenue = price.mean * quantity.sum;
endcomp;
rbreak after / summarize dol;
run;
q
p Revenue is defined as a computed variable.
q Revenue is the product of the quantity and the unit price.
208 Carpenter’s Complete Guide to the SAS REPORT Procedure
Inspection of the resulting table shows that the values are correct everywhere except on the
summary line generated by the RBREAK statement. The overall mean unit price is not weighted
consequently the total revenue is incorrect.
Extending Compute Blocks
Calculating a Weighted Mean
Computed Values
Product
Code
a
a
a
Number
Sold
1
10
100
=========
111
Unit
price
$1.50
$1.00
$0.50
=======
$1.00
Total
Revenue
$1.50
$10.00
$50.00
=======
111.00
The detail lines of the report are
correct, however on the summary
line only the total quantity value is
correct. The mean price is
unweighted, and this unweighted
average is used to calculate the total
revenue.
The problem is that we have failed to calculate the weighted average unit price on the report
summary line. We can do this in a COMPUTE AFTER block, however, because of the order that
compute blocks are executed we must be careful to control the process. Because compute blocks
associated with report items are executed before compute blocks associated with summary lines
(COMPUTE AFTER in this example), the value of the weighted average unit price and the total
revenue must both be recalculated.
title3 'Computed Values Utilizing the _BREAK_';
proc report data=temp nowd;
columns product quantity price revenue;
define product / display;
define quantity / analysis sum;
define price / analysis mean format=dollar5.2 width=7;
define revenue / computed 'Total Revenue' format=dollar6.2 width=7;
compute revenue;
revenue = price.mean * quantity.sum;
if _break_ = ' ' then _sum_revenue + revenue;
endcomp;
compute after;
price.mean = _sum_revenue / quantity.sum;
revenue = _sum_revenue; t
endcomp;
r
s
rbreak after / summarize dol;
run;
r For detail rows (_break_ = ' ') we need to accumulate the total revenue, which we will
use for the summary row calculations.
s The weighted average unit is defined as the total revenue divided by the total number of units
sold.
t The total revenue for this summary line was already calculated incorrectly in the compute
revenue block. Here it is recalculated using the total that we saved.
Chapter 7: Extending Compute Blocks 209
Extending Compute Blocks
Calculating a Weighted Mean
Computed Values Utilizing the _BREAK_
Product
Code
a
a
a
Number
Sold
1
10
100
=========
111
Unit
price
$1.50
$1.00
$0.50
=======
$0.55
Total
Revenue
$1.50
$10.00
$50.00
=======
$61.50
MORE INFORMATION
Section 6.7 introduces the FREQ statement and Section 11.3 discusses the order of compute block
processing.
7.10 Chapter Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports
clothing and equipment store. Create a report that shows total PROFIT and percentage of
annual sales for each PRODUCT_LINE within each year. You might want to build on the
results of Exercise 1 in Chapter 5.
Include a summary line (BREAK) for each product line and a total summary line (RBREAK).
Use a compute block to change the percentage on the report summary to missing.
2. Building on the results of Exercise 1 in Chapter 7, add ‘Total’ as text in the YEAR column
for each of the annual summary lines, and ‘Overall’ on the report summary line.
Use the OUT= option to examine the output data set, and note the values taken on by the
_BREAK_ variable.
3. Instead of using the RBREAK statement to calculate the overall total profit, calculate the
value in a compute block and display it using a LINE statement.
210 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
8
Using PROC REPORT with ODS
8.1 Introduction to the STYLE= Option 213
8.2 Using STYLE= to Change Attributes 216
8.2.1 Changing Text and Cell Attributes 216
8.2.2 Adding a Logo to Your Report 219
8.2.3 Controlling Report Size 223
8.2.4 Adding Horizontal and Vertical Spaces to Separate Data 223
8.3 Using CALL DEFINE to Change Style Attributes 227
8.3.1 Using CALL DEFINE in a Simple Report 228
8.3.2 Creating Shaded Rows 229
8.3.3 Conditional Assignment of Attributes 231
8.4 Creating Trafficlighting Effects 232
8.4.1 Building Trafficlighting Formats 233
8.4.2 Using Formats with the STYLE= Option 234
8.4.3 Controlling Trafficlighting with CALL DEFINE 236
8.4.4 Trafficlighting in the Presence of Computed Variables and Summary
Lines 236
8.4.5 Trafficlighting When Differentiating between Columns 240
8.4.6 Differentiating between Columns on Group Summary Rows 242
8.4.7 Trafficlighting on the REPORT Summary Row 245
8.4.8 A Few Things to Remember When Using Formats for Trafficlighting
249
8.5 Embedding Hyperlinks within Your Table 249
8.5.1 Linking Titles and Footnotes Using HTML Anchor Tags and the LINK=
Option 250
212 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.5.2 HTML Anchor Tags as Data Values
255
8.5.3 Establishing Links Using CALL DEFINE
8.5.4 Forming Links Using STYLE=
257
260
8.5.5 Creating Links in a PDF Document
262
8.5.6 Creating Links in an RTF Document
265
8.5.7 Automation Using the Macro Language
8.5.8 Using Formats to Build a Link
266
268
8.6 Using the Escape Character for In-Line Formatting
8.6.1 Controlling Superscripts and Subscripts
8.6.2 Displaying Page Numbers
270
271
273
8.6.3 Generating a Dagger 277
8.6.4 Using the Escape Character with S={ } and {STYLE} to Change Style
Attributes 279
8.6.5 Line Breaks and Wrapping
282
8.6.6 Passing Raw Destination-Specific Codes
289
8.7 Using TITLE and FOOTNOTE Statement Options
292
8.8 Creating Tip or “Flyover” Text for HTML and PDF
8.8.1 Using CALL DEFINE
293
293
8.8.2 Placing Tip Text Using STYLE=
8.8.3 Placing Tip Text Using ~S={ }
295
297
8.9 Specifying Multiple Columns for RTF and PDF
8.10 Adding Text through the TEXT= Option
298
300
8.11 RIGHTMARGIN: Aligning Numbers When Using CELLWIDTH
8.12 Chapter Exercises
301
304
A number of sources contain information on the Output Delivery System that is both more
complete and more detailed than can be provided in this book. Since it is not possible to present
all aspects of ODS here, this book includes only those topics that either directly relate to PROC
REPORT or are commonly associated with the reporting process. The reader is encouraged to
investigate some of the sources highlighted in the various SEE ALSO sections.
For the most part, the topics included in Chapter 8 are fairly independent of the ODS destination.
Related topics that apply only to specific destinations are presented in Chapter 9, “Reporting
Specifics for ODS Destinations.”
Although this usage is not discussed in this book, the Output Delivery System can be used to
generate data sets through the use of the OUTPUT destination (see Smith (2003) for detailed
examples). For PROC REPORT, the OUT= option is generally sufficient if we need to “see” the
resultant table as data. As a matter of fact, for the current releases of SAS, PROC REPORT does
not even support the OUTPUT destination; consequently, this ODS destination has little utility for
the PROC REPORT programmer.
Chapter 8: Using PROC REPORT with ODS 213
MORE INFORMATION
Chapter 9 also discusses ODS-related topics that are more destination-specific.
SEE ALSO
®
McNeill (2002) discusses a number of things that deal with ODS that are new to SAS 9 and in the
process provides a nice introduction to a number of items that deal with reporting. The books by
Haworth (2001a) and Gupta (2003) are “must reads” for the Output Delivery System.
8.1 Introduction to the STYLE= Option
When ODS is used, the basic formatting of display attributes, such as font style, font size,
foreground color and background color, are controlled through an ODS style selected by the user.
PROC REPORT can override a majority of these display attributes through the use of the
STYLE= option. This option is distinct from the STYLE= option on the ODS statement, which is
used to select the overall ODS style.
The STYLE= option discussed in this section can be used on most PROC REPORT statements to
control attributes of the table such as color, font, and text size. The STYLE= option can be used
on the following PROC REPORT statements:
PROC REPORT
DEFINE
COMPUTE
BREAK
RBREAK
CALL DEFINE (The form of the option is slightly different; see Section 8.3.)
Here is the general form of the STYLE= option:
style(component) = {attribute=value}
component identifies what is to receive the style.
attribute
identifies the style attribute(s) that is to receive the value.
Notice that the attribute specification (to the right of the equals sign) is enclosed in curly braces
{ }. Although other brackets can usually be used, the curly braces are recommended.
Because PROC REPORT controls all aspects of the construction of the report table, and because a
given statement can sometimes modify several different aspects of the table, we must have a
mechanism to identify which portion of the table is to be modified by a given application of the
STYLE= option. This is accomplished through a combination of the component portion of the
option, along with the statement on which the option appears.
214 Carpenter’s Complete Guide to the SAS REPORT Procedure
There is a default component for each PROC REPORT statement that supports the use of the
STYLE= option. If you know the default component and you want the option to apply to that
component, you can omit that portion of the option. The form of the option becomes as follows:
style = {attribute=value}
Obviously, because not all portions of the table are modified by all the statements, all of the
components cannot be used with all of the PROC REPORT statements. The following table shows
which components can be used with which statements.
PROC REPORT
Statement
Supported
Components
PROC REPORT
CALLDEF
Cells identified by the CALL
DEFINE routine
COLUMN
Cells of all columns
HEADER | HDR
Column and spanning
headers
LINES
Line statements
REPORT *
Structural part of the report
SUMMARY
Default summary lines
COLUMN*
Cells of this column
HEADER | HDR *
Column headers
CALL DEFINE
CALLDEF *
Cells identified by the CALL
DEFINE routine
COMPUTE
LINES *
Line statements
BREAK
SUMMARY *
Summary lines
RBREAK
SUMMARY *
Summary lines
DEFINE
Portion of Report Affected
Although I think it is always a good idea, you are not required to specify a component when using
the STYLE= option. In the previous table, the * indicates the default component if the component
is not explicitly specified.
Since the purpose of the STYLE= option is to change the attributes of various portions of the
report, there is also an association between the components and the attributes. By implication, this
means that not all attributes are available for all components.
The following table shows a small subset of available attributes. The full list of attributes comes
from the ODS STYLE. Consult the ODS documentation or see Lund (2005), which has an
extensive list of attributes.
Chapter 8: Using PROC REPORT with ODS 215
Selected
Components
Selected
Attributes
What the Attribute Does
REPORT
JUST
Controls justification (left, right, center)
CELLPADDING
Specifies the amount of white space
surrounding the text
BORDERCOLOR
Specifies the border color
HEADER
BACKGROUND
Specifies the background color
COLUMN
CELLHEIGHT
Specifies the cell height
SUMMARY
CELLWIDTH
Specifies the cell width
FOREGROUND
Specifies the foreground color
FONT_WEIGHT
Specifies the text thickness, e.g., light,
medium, bold
FONT_FACE
Specifies the text type face, e.g., Arial, Times
FONT_SIZE
Specifies the text size
FONT_STYLE
Specifies the text style, e.g., italic, roman,
slant
RIGHTMARGIN
Specifies the margin width within a cell
For each statement and component combination there is also a default attribute. The following
table is taken from the documentation and shows these default attributes. However, in the long run
it is generally more efficient, from the coder’s perspective, to specify the component and
attributes rather than to rely on the defaults.
PROC REPORT
Statement
Component
Default Attribute
REPORT
REPORT, COLUMN, HEADER|HDR,
SUMMARY, LINES, CALLDEF
TABLE
BREAK
SUMMARY, LINES
DATAEMPHASIS
CALL DEFINE
CALLDEF
DATA
COMPUTE
LINES
NOTECONTENT
DEFINE
COLUMN
DATA
HEADER
HEADER
SUMMARY, LINES
DATAEMPHASIS
RBREAK
216 Carpenter’s Complete Guide to the SAS REPORT Procedure
SEE ALSO
Stroupe (2002) introduces the STYLE= option, and Pass and McNeil (2003) provide extensive
examples. Haworth (2001b) provides a more complete enumeration of available style attributes,
as does Lund (2005). Tables similar to those presented in this section can also be found in Burlew
(2005, p. 191).
8.2 Using STYLE= to Change Attributes
The syntax of the STYLE= option seems a bit odd at first. Be sure to remember that you must
specify both the attribute(s) and, either explicitly or implicitly by using the default, the component
to which the attributes are to be applied. Although the STYLE= option can be used in most of the
statements in the PROC REPORT step, it is the combination of statement, attribute, and
component that determines precisely the effect of the option.
8.2.1 Changing Text and Cell Attributes
It is easiest to describe the effects of various combinations of components, attributes, and
statements through an example that shows most of these combinations. First consider the
following REPORT procedure, which does not use the STYLE= option.
options center;
ods listing close;
ods html style=default
path="&path\results"
body='ch8_2_1a.html';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
break after region / summarize suppress;
rbreak after
/ summarize;
run;
ods html close;
Chapter 8: Using PROC REPORT with ODS 217
Using the DEFAULT ODS style, this PROC REPORT step produces the following HTML table.
Destination: HTML Style: DEFAULT
The following REPORT procedure produces exactly the same report, except it uses various
STYLE= options.
options center;
ods listing close;
ods html style=default
path="&path\results"
body='ch8_2_1b.html';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(report)={just=right} n
style(header)={background=yellow o
font_weight=bold p
foreground=green} q
style(column)={foreground=red r
background=white};
column region country product,actual totalsales;
define region / group
style(header)={background=cyan} s;
define country / group
style(column)={background=pink} t
style(header)={font_face=times};
define product / across;
define actual / analysis sum format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
218 Carpenter’s Complete Guide to the SAS REPORT Procedure
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
break after region/summarize suppress
style(summary)={font_weight=bold} u;
rbreak after
/summarize
style(summary)={font_weight=bold v
font_size=4};
run;
ods html close;
Clearly this procedure creates a fairly ugly table (actually it is worse in color). However, it does
serve to demonstrate the effects of various combinations of statements and components.
X
o
p
s
r
q
t
u
v
Destination: HTML Style: DEFAULT
n The entire table has been right-justified. This makes the title, which is centered, appear to be
left-justified.
o The background color for the entire header area is yellow, with bolded pgreen q letters.
r All of the cells of the table are given a white background with red letters (foreground).
s The header area for this particular column is given a cyan background. This overrides the color
specified in the PROC REPORT statement o.
Chapter 8: Using PROC REPORT with ODS 219
t The column background color is changed to pink, overriding the white background specified
in r. Notice that the font for the header has been changed to Times, and that two different
STYLE= options with two different components have been specified in the same
statement.
u The summary line has been bolded.
v The report summary line has been bolded and the size has been increased. Nominally the size
is in points. However, actual size seems to be dependent on destination and potentially on font.
You might need to experiment a bit.
Although generally not useful and, in my opinion, a bit too much of a shortcut, it is possible to
combine components when the specified attributes are the same. At t in the previous example,
two style options are applied in the same DEFINE statement.
define country / group
style(column)={background=pink}
style(header)={font_face=times};
If they had both been establishing the same attributes e.g.,
define country / group
style(column)={background=pink
font_face=times}
style(header)={background=pink
font_face=times};
the two options could have been combined into one option with two components:
define country / group
style(header column)=
{background=pink
font_face=times};
SEE ALSO
Levin (2005) and Baroud, Senner, and Johnson (2006) both use the STYLE= option, as does
Sample 609, which shows how to form indented text in column 1. Smoak (2004) uses the
STYLE= option in RTF examples that generate tables to be exported to Microsoft Word. Girgis
(2006) uses the STYLE= option extensively in a series of examples.
8.2.2 Adding a Logo to Your Report
The PREIMAGE style attribute can be used to add pictures or images to your report. The
following example adds not only text (using LINE statements), but also an image.
ods listing close;
ods html style=default
path="&path\results"
body='ch8_2_2a.html';
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
220 Carpenter’s Complete Guide to the SAS REPORT Procedure
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
compute before/style(lines)={preimage="&path\magic.gif"
font_weight=bold
font_face=arial
font_size=6};
line 'Magic Mystery, Inc.';
line 'Sales Summary';
endcomp;
break after region / summarize suppress;
rbreak after
/ summarize;
run;
ods html close;
Although the image has been imported correctly and the STYLE= attributes have been applied,
the lines have been added to the table after the header section. This is due to the order that the
PROC REPORT events are processed (see Section 1.4.2, Section 7.1 and Chapter 11, “Details of
the PROC REPORT Process”).
Chapter 8: Using PROC REPORT with ODS 221
We can change when the lines from the COMPUTE block are added to the report by including the
_PAGE_ option in the COMPUTE statement.
ods listing close;
ods html style=default
path="&path\results"
body='ch8_2_2b.html';
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
compute before _page_ /
left
style={preimage="&path\magic.gif"
font_weight=bold
font_face=arial
font_size=6};
line 'Magic Mystery, Inc.';
line 'Sales Summary';
endcomp;
break after region / summarize suppress;
rbreak after / summarize;
run;
ods html close;
When the _PAGE_ option is used on the COMPUTE statement, the results of the LINE
statements are laid down on the page first. Notice that, unlike all the other uses of this option in
this PROC REPORT step, the STYLE = option does not have a component associated with it.
Since the style is being applied before PROC REPORT takes control of the page, a REPORT
component is not appropriate.
222 Carpenter’s Complete Guide to the SAS REPORT Procedure
The LEFT option causes the title to be left-justified. However, the justification options on the
COMPUTE statement (LEFT, CENTER, RIGHT) were originally designed to work with the
LISTING destination, and their continued functionality for non-LISTING destinations may
change in future releases of SAS. For these destinations, using the JUST= style attribute modifier
on the STYLE= option or the CALL DEFINE routine (see Section 8.3) is more appropriate. The
compute block becomes the following:
compute before _page_ /
style={preimage="&path\magic.gif"
just=left
font_weight=bold
font_face=arial
font_size=6};
line 'Magic Mystery, Inc.';
line 'Sales Summary';
endcomp;
MORE INFORMATION
Section 5.4 introduces the _PAGE_ location for the COMPUTE statement. The POSTIMAGE=
attribute modifier is used in Section 10.3.1.
SEE ALSO
Haworth (2001b) places a logo on the page using STYLE=, and Hadden (2006) uses CALL
DEFINE. Also using STYLE=, Burlew (2005, pp. 212–215) places images on the report using the
PREIMAGE= and POSTIMAGE= attribute options.
Chapter 8: Using PROC REPORT with ODS 223
8.2.3 Controlling Report Size
There are two things that you need to take into consideration if you want to control the overall
size of the report. For instance, if you would like to decrease the overall size of the report,
reducing the size of the font would be insufficient for most destinations. This is because the size
of the individual cell depends on several factors, only one of which is the size of the font. For
most destinations, reducing the size of the font merely results in the smaller text residing within a
cell of the same size.
Fortunately, we can control the cell size (spacing between rows) and the size of the font
independently through the use of the STYLE= option on the PROC REPORT statement. In the
following PROC REPORT statement, both the REPORT and COLUMN components are used.
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(report)={font_size=9pt} n
style(column)={font=('times new roman', 9pt)} o;
n When applied with the REPORT component, the attribute FONT_SIZE resizes the cells by
specifying how much space is needed between the lines of text. Although the REPORT
component has less utility than most of the other components, it can be very helpful when you
need to set report-wide attribute characteristics.
o When used with the COLUMN component, the FONT= attribute, which combines both
FONT_FACE and FONT_SIZE, specifies the actual font size.
Notice also that the font name is enclosed in quotation marks in this example, whereas in the other
examples the FONT_FACE was specified without quotation marks. Generally, the quotation
marks will only be needed when you specify a font with multiple words, e.g., 'times new
roman'.
8.2.4 Adding Horizontal and Vertical Spaces to Separate Data
It is often desirable to add either horizontal or vertical spaces to highlight certain portions of your
report. Vertical lines can be added by creating a computed variable (DUMMY in the following
example) that contains no text. As we have seen elsewhere in this book, the LINE statement can
be used to add blank horizontal spaces. In both cases, we might want to control the attributes,
especially width and height, of the space.
The STYLE= option can be used to control both the vertical column’s width and the horizontal
row’s height. In this example, we add a vertical space to the left of the TOTALSALES column
and blank horizontal spaces after each region.
ods html style=default
path="&path\results"
body='ch8_2_4a.html';
ods pdf style=printer
file="&path\results\ch8_2_4a.pdf";
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual dummy totalsales;
define region / group;
define country / group;
define product / across;
224 Carpenter’s Complete Guide to the SAS REPORT Procedure
define actual
/ analysis sum
format=dollar8.
'Sales';
define dummy / computed ' ' n
style(column) = {cellwidth=2pt o
background=gray99 p
bordercolor=gray99};
define totalsales / computed format=dollar10.
'Total Sales';
break after region / summarize suppress;
rbreak after
/ summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
compute dummy /char length=2;
dummy=' ';
endcomp;
compute after region/style(lines) = {font_size=2pt q
background=grayaa
line ' ' ; s
endcomp;
run;
r;
n The computed column DUMMY is empty and does not have a header. This forms a vertical
space.
o The CELLWIDTH attribute determines the width of the column. The exact width varies
depending on the destination.
p The background color of the vertical space is specified as a dark gray. Gray scale colors vary
from GRAY00 (black) to GRAYFF (white). There are 256 shades of gray, and the last two
digits are the hex codes for these shades.
q LINE statements executed within this compute block are written after each region. Because
there is also a BREAK AFTER REGION summary, the line appears after the summary line.
The height of the space can be controlled by the specification of the font size.
r The background color is set to a dark gray (a few shades lighter than the vertical space).
s The LINE statement creates the blank line with the attributes specified in the STYLE=
option (q and r).
Chapter 8: Using PROC REPORT with ODS 225
Although it appears that the individual cells have borders in the dummy column, what we are
actually seeing as borders is the space between the cells. Because the vertical space is formed by a
computed variable (DUMMY), its individual cells are surrounded by space, and this can itself be
annoying. The width of this space can be controlled by the CELLSPACING attribute modifier.
Here this attribute is applied to all the cells of the table.
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd style={cellspacing=0};
column region country product,actual dummy totalsales;
. . . Portions of the code are not shown . . .
Because the component has not been specified, the default component is REPORT. Here is the
new HTML table:
226 Carpenter’s Complete Guide to the SAS REPORT Procedure
The blank lines are
given a light gray
color through the
use of the
BACKGROUND=
attribute modifier.
Unfortunately, CELLSPACING does not work equally well for all destinations.
If we want the dummy spaces to blend into the background we need to know the color of the
background. For STYLE=DEFAULT shown above, inspection of the style definition shows that
the default color for the data area background is the light gray cxD3D3D3. Using this value as the
background color causes these lines to blend into the background.
define dummy
/
computed ' '
style(column) = {cellwidth=1mm
background=cxD3D3D3};
Destination: HTML Style: DEFAULT
Chapter 8: Using PROC REPORT with ODS 227
Learning how to tease information such as the data area background color from the style
definition is outside the scope of this book. Often, however, a visual inspection of the PROC
TEMPLATE step that defines the style (the definition is stored in SASHELP.TMPLMST) can be
sufficient, although there can often be some guessing and a bit of trial and error.
MORE INFORMATION
The example in Section 8.3.1 uses CALL DEFINE to change attributes of a summary line.
SEE ALSO
Lund (2005) includes examples that create both blank columns and blank rows in PDF tables.
Carpenter (2004a, p. 176) has an example of a macro that generates uniformly spaced gray scale
PATTERN statements. Cell and table borders can be further controlled through the use of the
FRAME= and RULES= attributes; see Sevick (2006).
8.3 Using CALL DEFINE to Change Style
Attributes
The CALL DEFINE routine can also be used to set style attributes. Its usage is similar to that of
the STYLE= option. As a matter of fact, when CALL DEFINE is used to set style attributes, the
second argument is ‘STYLE’ and the third argument is essentially the STYLE= option discussed
in Sections 8.1 and 8.2.
Because the CALL DEFINE routine is an executable statement, it can be conditionally executed
in a compute block. This gives the user a finer degree of control than the STYLE= option, and this
is one of the primary advantages of CALL DEFINE.
Here is the CALL DEFINE syntax:
call define(location, 'style', style<(component)>={attributespecification(s)});
Multiple calls to the CALL DEFINE routine are permitted, and a given call can have multiple
attribute specifications.
MORE INFORMATION
The CALL DEFINE routine was introduced in Section 7.5 and is also used for linking table
elements in Section 8.5.
SEE ALSO
Hadden (2006) uses the CALL DEFINE statement to add a company logo to a report.
228 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.3.1 Using CALL DEFINE in a Simple Report
We would like to emphasize the summary lines for each region. We can do this by changing the
summary line attributes. In this example, a compute block n has been added for the sole purpose
of allowing us to execute a CALL DEFINE.
ods listing close;
ods html style=default
path="&path\results"
body='ch8_3_1.html';
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
compute after region; n
call define(_row_, o
'style',
'style={font_weight=bold
background=white
font_face=arial}');
endcomp;
break after region / summarize suppress;
rbreak after
/ summarize;
run;
ods html close;
n The compute block is executed after each region. Consequently, the style designated in the
CALL DEFINE statement applies to the summary row generated by the BREAK AFTER
REGION statement.
o The special location designator _ROW_ is used to apply the result of CALL DEFINE to the
entire row. Notice that _ROW_ is not enclosed in quotation marks.
Chapter 8: Using PROC REPORT with ODS 229
The summary
row is assigned
attributes for a
bolded Arial
text with a
white
background.
Destination: HTML Style: DEFAULT
8.3.2 Creating Shaded Rows
The previous example changes the background color for the summary lines. It is not unusual to
need to change the background color for the detail lines as well. In this example, we would like
the detail lines to alternate shades of gray, the summary lines for REGION to be white
(GRAYFF), and the overall summary line a bit darker (GRAY00 is black) than the others.
ods listing close;
ods html style=default
path="&path\results"
body='ch8_3_2.html';
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
break after region / summarize suppress; n
rbreak after
/ summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
cnt+1; o
if mod(cnt,2) then call define(_row_,
'style',
p
230 Carpenter’s Complete Guide to the SAS REPORT Procedure
'style={background=graydd}');
else call define(_row_, q
'style',
'style={background=graycc}');
endcomp;
compute after region;
call define(_row_, r
'style',
'style={font_weight=bold
background=grayff
font_face=arial}');
cnt=0; s
endcomp;
compute after;
call define(_row_, t
'style',
'style={font_weight=bold
background=graybb
font_face=arial}');
endcomp;
run;
ods html close;
n The BREAK and RBREAK statements are needed to create the summaries.
o The TOTALSALES compute block is executed once for each row. A counter (CNT) is created
so that we can determine odd vs. even rows.
p The MOD function returns a 1 for odd-numbered rows. Since 1 is true, this CALL DEFINE
statement is executed when CNT is odd.
q This CALL DEFINE statement is executed for even-numbered rows.
r The row attributes are set for the region summary.
s The counter is reset to zero so that the first line for each region will always have the same
shade.
t The report-wide summary line attributes are set in this CALL DEFINE statement.
Chapter 8: Using PROC REPORT with ODS 231
Destination: HTML Style: DEFAULT
SEE ALSO
Haworth (2001b) and Burlew(2005, pp. 203–208) both use CALL DEFINE to change the shading
of rows.
8.3.3 Conditional Assignment of Attributes
One of the very real advantages of the CALL DEFINE routine is that it is executable. That means
that we can use it to conditionally assign attributes based on the values in a cell. Although it was
not directly pointed out, this was done in the example in Section 8.3.2. There IF-THEN/ELSE
processing was used to determine which CALL DEFINE statement was to be executed.
This capability can be expanded to look at the data itself with the objective of controlling column
attributes based on the data in those or other columns. In the following example, the background
and foreground colors are conditionally changed based on the cell values.
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
break after region / summarize suppress;
rbreak after
/ summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
if _c3_ < 25e3 then call define('_c3_',
232 Carpenter’s Complete Guide to the SAS REPORT Procedure
'style',
'style={background=red
foreground=white}');
if _c4_ < 24e3 then call define('_c4_',
'style',
'style={background=red
foreground=white}');
if _c5_ < 21e3 then call define('_c5_',
'style',
'style={background=red
foreground=white}');
endcomp;
run;
ods html close;
The resulting HTML table shows that the background and foreground colors have been changed
for selected cells.
The determination
of whether or not
to change the cell
attributes
(foreground and
background color
in this case) is
totally dependent
on the value of the
data in that
particular cell.
Destination: HTML Style: DEFAULT
The process of changing cell attributes based on the cell’s value can quickly become tedious if
there are many conditions or cells. The process of trafficlighting (Section 8.4) simplifies this
process.
8.4 Creating Trafficlighting Effects
The term “trafficlighting” refers to table items whose characteristics, such as background color,
foreground color, font, and size, change according to the value that is being displayed. Most
importantly, the change of attributes is automatic and not implemented manually. Because control
is automatic, you can highlight the values that you really want your reader to see without
manually editing the report table.
Chapter 8: Using PROC REPORT with ODS 233
The implementation is both ingenious and simple. We build custom formats using PROC
FORMAT. These formats map the values of interest into style attributes. This makes changing the
highlighted characteristics as easy as changing a format.
The formats are associated with report locations using various techniques. The trafficlighting
format can be applied to report items, statistics, summary columns, and summary rows.
Techniques include the use of the STYLE= option and the CALL DEFINE statement.
For simple reports, the implementation is very straightforward. However, as the reports increase
in complexity, timing issues relative to the resolution of the formats can become problematic.
Because formats depend on the value that is to be placed in the cell, the attribute determined from
the format is not available until the value has been determined. Because of the way that the
default report row attributes are assigned, we need to be extra careful with the application of
trafficlighting effects when using compute blocks on summary lines and computed variables.
Sections 8.4.6 through 8.4.8 discuss some of these issues.
SEE ALSO
Haworth (2001b), Thornton (2006), and Hadden (2006) each have trafficlighting examples. In a
paper written specifically with PROC REPORT in mind, Carpenter (2006c) includes a discussion
of some of the special issues associated with trafficlighting. Burlew (2005, p.203) uses
trafficlighting techniques to control the row shading. Feng (2006) uses trafficlighting in a report
that is rendered in Microsoft Excel.
8.4.1 Building Trafficlighting Formats
Trafficlighting depends on customized formats. There is nothing special about these formats other
than the fact that they resolve not to a value that is to be displayed, but rather to an attribute that
will be interpreted when the report is displayed.
The following PROC FORMAT creates two formats that can be used to control foreground and
background colors. However, additional formats can be created so that the formatted value (the
item on the right of the equals sign) can be used to alter other style attributes as well.
proc format;
value cfore
low - 21000
21000< - 25000
75000 - high
value cback
low - 21000
21000< - 25000
75000 - high
run;
= 'white'
= 'black'
= 'white';
= 'red'
= 'yellow'
= 'green';
These formats specify that values less than 21,000 are to be displayed with white letters and a red
background, whereas values over 75,000 are to be displayed with a green background. These two
example formats are used throughout the examples in Section 8.4 and are assumed to exist even if
the PROC FORMAT step is not actually shown in each example.
Of course, even though these examples only change the foreground and background color, values
for the other attributes can also be specified. You can control font, font size, and other attribute
values through the use of formats. Other clever usages have even included links and images that
are dependent on the value that is displayed.
234 Carpenter’s Complete Guide to the SAS REPORT Procedure
In these examples, the colors are named. However, whenever colors are specified within SAS,
you can also use RGB hexidecimal codes or gray scales to specify the colors.
MORE INFORMATION
The example in Section 8.5.8 generates a link based on a format that depends on the value
displayed.
8.4.2 Using Formats with the STYLE= Option
The report value that is to be displayed must be associated with a format, and one way that this is
done is through the STYLE= option. In the following example, the analysis variable ACTUAL
has a FORMAT= option n on the DEFINE statement for the display of the values. It also has a
STYLE= option o, which utilizes the trafficlighting formats to control the foreground and
background colors.
proc format;
value cfore
low
21000< >50000
value cback
low
21000< >50000
run;
21000 = 'white'
25000 = 'black'
= 'white';
21000 = 'red'
25000 = 'yellow'
= 'green';
ods listing close;
ods html style=default
path="&path\results"
body='ch8_4_2a.html';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column country region product actual;
define country / group;
define region / group;
define product / group;
define actual / analysis sum
format=dollar8. n
'Sales'
style(column) = {background=cback. o
foreground=cfore.};
run;
ods html close;
Chapter 8: Using PROC REPORT with ODS 235
A portion of the resulting table shows that values less than $25,000 are highlighted for quick
recognition by the sales managers:
Destination: HTML Style: DEFAULT
Users often try to apply the trafficlighting formats to all the data columns by placing the STYLE=
option on the REPORT statement.
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(column) = {background=cback.
foreground=cfore.};
Although this approach can work to some degree, it usually causes some problems as well. This is
because PROC REPORT attempts to apply the formats to all the columns, including the grouping
columns (REGION, COUNTRY, and PRODUCT), and in this case these are all character
variables.
The report in the previous example was fairly straightforward. When we start adding summary
lines, things can become more complex. If we add a BREAK and RBREAK statement to the
PROC REPORT step, a logical extension of what we did in the previous example and Section
8.2.1 would be to add a STYLE(SUMMARY)= option to these two statements.
break after country / summarize suppress
style(summary) = {background=cback.
foreground=cfore.};
rbreak after / summarize
style(summary) = {background=cback.
foreground=cfore.};
Unfortunately, the timing is such that the execution of these statements and the resolution of the
formats will not yield the desired results. When we want to provide trafficlighting for summary
lines or even values under ACROSS variables, we usually need to use the CALL DEFINE
statement. Of course, this statement can be executed only inside of a compute block. However,
because we can assign a compute block for each column and summary line, we can use CALL
DEFINE. This is discussed in the following section.
236 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
The generation of trafficlighting effects in the presence of computed variables and summary lines
is discussed further in Section 8.4.4.
8.4.3 Controlling Trafficlighting with CALL DEFINE
The CALL DEFINE statement can be used in a compute block to create trafficlighting effects.
The trafficlighting in the first example in Section 8.4.2 could also have been accomplished with
CALL DEFINE. In the following code, a compute block containing a CALL DEFINE statement
has replaced the STYLE= option on the DEFINE ACTUAL statement.
ods listing close;
ods html style=default
path="&path\results"
body='ch8_4_3.html';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column country region product actual;
define country / group;
define region / group;
define product / group;
define actual / analysis sum
format=dollar8.
'Sales';
compute actual;
call define(_col_,
'style',
'style = {background=cback.
foreground=cfore.}');
endcomp;
run;
ods html close;
This code, which uses the same formats as were established in Section 8.4.1, creates the same
table as in Section 8.4.2. The compute block exists only to establish a location to place the CALL
DEFINE statement. Certainly other executable compute block statements could also be placed in
this compute block.
SEE ALSO
Fan (2005) uses CALL DEFINE to set style attributes for trafficlighting.
8.4.4 Trafficlighting in the Presence of Computed Variables
and Summary Lines
Often the best way to produce trafficlighting for reports that include computed variables and/or
summary lines is with a combination of CALL DEFINE statements and STYLE= options.
In the following examples, we have complicated the previous examples by changing the
PRODUCT variable to an ACROSS variable, computing a product total, and requesting country
and report summarizations. We would like the trafficlighting formats established in Section 8.4.1
to be applied to all of the numeric values, including summary lines and the computed column.
Chapter 8: Using PROC REPORT with ODS 237
We can do this in several ways, and it is worth discussing the differences between the approaches.
As was mentioned in Section 8.4.2, we cannot just use the STYLE= option on the BREAK,
RBREAK, or DEFINE TOTALSALES statements.
For our first attempt, in the following example we take the approach of assigning the attributes by
columns for all rows (detail and summary). Remember, the color versions of these tables are
available in the “Results” section of the CD that accompanies this book.
proc format;
value cfore
Low
21000<
75000
value cback
Low
21000<
75000
run;
- 21--- = 'white'
- 25000 = 'black'
- high = 'white'
- 21000 = 'red'
- 25000 = 'yellow'
- high = 'green';
ods html style=default
path="&path\results"
body='ch8_4_4a.html';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across; n
define actual / analysis sum
format=dollar8.
'Sales'
style(column) = {background=cback. o
foreground=cfore.};
define totalsales / computed format=dollar10.
'Total Sales'
style(column) = {background=cback. p
foreground=cfore.};
break after region / summarize suppress;
rbreak after
/ summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
endcomp;
run;
ods html close;
n The PRODUCT, which has three levels, is specified as an ACROSS variable (resulting in three
columns).
o The three columns associated with the sales values are assigned these attributes. In the
COLUMNS statement, ACTUAL is nested under the ACROSS variable PRODUCTS.
p The same attributes are applied to the computed ‘Total Sales’ column.
238 Carpenter’s Complete Guide to the SAS REPORT Procedure
Although the detail values have been formatted correctly, the trafficlighting formats have been
incorrectly applied to the summary lines.
Destination: HTML Style: DEFAULT
In fairly simple tables such as this, one way to apply the formatted style attributes to the summary
lines is with the use of the SUMMARY component on the STYLE= option—e.g.,
STYLE(SUMMARY)= q, as in the following:
ods html style=default
path="&path\results"
body='ch8_4_4b.html';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(summary)={background=cback. q
foreground=cfore.};
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales'
style(column) = {background=cback.
foreground=cfore.};
define totalsales / computed format=dollar10.
'Total Sales'
style(column) = {background=cback.
foreground=cfore.};
break after region / summarize suppress;
rbreak after / summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
Chapter 8: Using PROC REPORT with ODS 239
endcomp;
run;
ods html close;
q The style attributes specified here apply to all columns for all summary lines.
Destination: HTML Style: DEFAULT
This is a fairly general “brute force” solution. It might not work for all tables, and it might be too
simplistic for more complex situations. At the very least, control is lost by applying the same
attributes to all of the different summary lines (of course this might be just what you want).
A similar table can be obtained by including the use of CALL DEFINE statements. This
technique, which uses a combination of CALL DEFINE statements and STYLE= options, is far
more extensible.
The following example also produces the previous table; however, the STYLE= option has been
removed from the DEFINE TOTALSALES n, and is replaced with a CALL DEFINE statement
in the compute block o. A good minimum rule of thumb is to use the CALL DEFINE statement
when a compute block is present. In fact, there is no real penalty for creating a compute block just
in order to take advantage of the CALL DEFINE statement.
ods html style=default
path="&path\results"
body='ch8_4_4c.html';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(summary)={background=cback.
foreground=cfore.};
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
240 Carpenter’s Complete Guide to the SAS REPORT Procedure
'Sales'
style(column) = {background=cback.
foreground=cfore.};
define totalsales / computed format=dollar10.
'Total Sales'; n
break after region / summarize suppress;
rbreak after
/ summarize;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
call define(_COL_,'style', 'style={background=cback. o
foreground=cfore.}');
endcomp;
run;
ods html close;
8.4.5 Trafficlighting When Differentiating between Columns
Because the CALL DEFINE statement can be conditionally executed and can reference individual
columns, it provides us a great deal of flexibility that is not available through the STYLE= option.
In the following example, we would like the trafficlighting attributes to vary by values of the
ACROSS variable (PRODUCT). This is accomplished by using direct column references in the
CALL DEFINE statements.
To illustrate this level of control, additional formats have been defined for the sales of chairs n,
desks o, tables p, and for total sales q.
proc format;
value cfore
low
- 21000
21000< - 25000
75000 - high
value cback
low
- 21000
21000< - 25000
75000 - high
value fchair n
low
- 23500
value bchair
low
- 23500
value fdesk o
low
- 25000
value bdesk
low
- 25000
value ftable p
low
- 21000
value btable
low
- 21000
value ftotal q
low
- 72000
value btotal
low
- 72000
run;
= 'white'
= 'black'
= 'white';
= 'red'
= 'yellow'
= 'green';
= 'white';
= 'red';
= 'white';
= 'red';
= 'white';
= 'red';
= 'white';
= 'red';
ods html style=default
path="&path\results"
Chapter 8: Using PROC REPORT with ODS 241
body='ch8_4_5.html';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd
style(summary)={background=cback.
foreground=cfore.}; r
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
break after region / summarize suppress;
rbreak after / summarize;
compute actual;
if _break_ = ' ' then do; s
call define('_C3_t','style','style={background=bchair. n
foreground=fchair.}');
call define('_C4_','style','style={background=bdesk. o
foreground=fdesk.}');
call define('_C5_','style','style={background=btable. p
foreground=ftable.}');
end;
endcomp;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
if _break_ = ' ' then do;
call define(_COL_,'style', 'style={background=btotal. q
foreground=ftotal.}');
end;
endcomp;
run;
ods html close;
Formats are created and used to control the attributes of the sales of chairs n, desks o, tables p,
and total sales q.
r In this example the summary lines are still controlled with a STYLE(SUMMARY)= option.
Trafficlighting for summary lines is discussed more in Section 8.4.6.
s These CALL DEFINE statements are applied only to nonsummary lines (_break_ = ' ').
t The specific column is named explicitly using the absolute column numbers. This permits us to
apply different formats to each type of sale item.
242 Carpenter’s Complete Guide to the SAS REPORT Procedure
n
o
r
q
p
r
Destination: HTML Style: DEFAULT
8.4.6 Differentiating between Columns on Group Summary
Rows
The previous example trafficlights all cells of all summary lines with the same format. When we
need to fine tune the level of format application, we have to be especially careful. The application
of the attribute formats for cells that are determined through multiple compute blocks becomes
complicated very quickly.
In the following example, we want to apply the trafficlighting formats to the four columns of
interest on the summary lines for the grouping variable REGION. To do this, we have created two
new formats to apply to the region summaries (FREGN. n for the foreground and BREGN. o for
the background).
proc format;
value cfore
low
- 21000
21000< - 25000
75000 - high
value cback
low
- 21000
21000< - 25000
75000 - high
value fchair
low
- 23500
value bchair
low
- 23500
value fdesk
low
- 25000
value bdesk
low
- 25000
value ftable
low
- 21000
= 'white'
= 'black'
= 'white';
= 'red'
= 'yellow'
= 'green';
= 'white';
= 'red';
= 'white';
= 'red';
= 'white';
Chapter 8: Using PROC REPORT with ODS 243
value btable
low
- 21000
value ftotal
low
- 72000
value btotal
low
- 72000
value fregn n
low
- 73000
value bregn o
low
- 73000
run;
= 'red';
= 'white';
= 'red';
= 'white';
= 'red';
ods html style=default
path="&path\results"
body='ch8_4_6a.html';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd; p
column region country product,actual totalsales;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
define totalsales / computed format=dollar10.
'Total Sales';
break after region / summarize suppress;
rbreak after / summarize;
compute actual;
if _break_ = ' ' then do;
call define('_C3_','style','style={background=bchair.
foreground=fchair.}');
call define('_C4_','style','style={background=bdesk.
foreground=fdesk.}');
call define('_C5_','style','style={background=btable.
foreground=ftable.}');
end;
else if _break_='REGION' then do; q
call define('_C3_', r 'style','style={background=bregn. s
foreground=fregn.}');
call define('_C4_','style','style={background=bregn.
foreground=fregn.}');
call define('_C5_','style','style={background=bregn.
foreground=fregn.}');
end;
endcomp;
compute totalsales;
totalsales = sum(_c3_, _c4_, _c5_);
if _break_ = ' ' then do;
call define(_COL_,'style', 'style={background=btotal.
foreground=ftotal.}');
end;
else if _break_='REGION' then do; t
244 Carpenter’s Complete Guide to the SAS REPORT Procedure
call define(_col_,'style','style={background=cback.
foreground=cfore.}');
end;
endcomp;
run;
ods html close;
n Foreground format is specified for the REGION totals for sales of individual products.
o Background format is specified for the REGION totals for sales of individual products.
p The STYLE(SUMMARY)= option has been removed from the PROC REPORT statement. If it
had been left in (as it was in the previous example), the CALL DEFINE statements would
override the STYLE= option. However, it has been my experience that CALL DEFINE
does not always override the STYLE (SUMMARY)= option, as the timing for the declaration
of the formats tends to conflict. My rule of thumb is to never allow a cell to be controlled by
attributes derived by more than one format. Attributes specified directly (not through formats)
do not seem to have this problem.
q A different set of CALL DEFINE statements is declared for the summary lines.
r The specific column is designated using the absolute column number.
s The same format is applied to each of the three summary row columns. The formats could have
been different for each of the columns (as they are for the detail rows).
t This style is applied to the summary line for TOTAL SALES.
t
t
s
Destination: HTML Style: DEFAULT
s
Chapter 8: Using PROC REPORT with ODS 245
In this example, all three columns for sales in the summary row (_C3_, _C4_, and _C5_) each
receive the same format. When this happens, the code in the compute block could be simplified
to a single CALL DEFINE statement. The COMPUTE ACTUAL block becomes the following:
compute actual;
if _break_ = ' ' then do;
call define('_C3_','style','style={background=bchair.
foreground=fchair.}');
call define('_C4_','style','style={background=bdesk.
foreground=fdesk.}');
call define('_C5_','style','style={background=btable.
foreground=ftable.}');
end;
else if _break_='REGION' then do;
call define(_row_,'style','style={background=bregn.
foreground=fregn.}');
end;
endcomp;
Although this approach works in this example, I have not found this to always be the case. The
problem, when it is a problem, seems to be caused by the use of _ROW_. Effectively, because we
are using _ROW_, our formatted attributes are applied to columns other than just _C3_, _C4_,
and _C5_, and this could cause a conflict in the TOTALSALES column.
8.4.7 Trafficlighting on the REPORT Summary Row
The specification of trafficlighting formats in the REPORT summary row is similar to that in the
previous example. However, there are additional complications.
First we have create two more formats to handle the overall product totals (FPROD. and
BPROD.). Here are the VALUE statements:
value fprod
low
- 145000 = 'white';
value bprod
low
- 145000 = 'red';
To apply these new formats, we add another ELSE IF statement used to detect the REPORT
summary line (_BREAK_ = ‘_RBREAK_’). The COMPUTE ACTUAL block (from Section
8.4.6) becomes the following:
compute actual;
if _break_ = ' ' then do;
call define('_C3_','style','style={background=bchair.
foreground=fchair.}');
call define('_C4_','style','style={background=bdesk.
foreground=fdesk.}');
call define('_C5_','style','style={background=btable.
foreground=ftable.}');
end;
246 Carpenter’s Complete Guide to the SAS REPORT Procedure
else if _break_='REGION' then do;
call define('_C3_','style','style={background=bregn.
foreground=fregn.}');
call define('_C4_','style','style={background=bregn.
foreground=fregn.}');
call define('_C5_','style','style={background=bregn.
foreground=fregn.}');
end;
else if _break_='_RBREAK_' then do;
call define('_C3_','style','style={background=bprod.
foreground=fprod.}');
call define('_C4_','style','style={background=bprod.
foreground=fprod.}');
call define('_C5_','style','style={background=bprod.
foreground=fprod.}');
end;
endcomp;
The following table is produced:
n
Destination: HTML Style: DEFAULT
Chapter 8: Using PROC REPORT with ODS 247
Clearly there is something wrong for the overall sales of chairs and desks. This error is a typical
result when there is a timing issue with the determination of attributes that are established by
using formats. In this case, the two values that have been “colored over” n did not receive a
attribute from the format (their values are both over 145,000), but it was too late to use the default
foreground and background colors. Here we can solve the problem through the use of the OTHER
range specification, which provides a place for all values in the format definition. Actually, this is
generally a good practice anyway. The format specifications become the following:
value fprod
low
- 145000 = 'white'
other
= 'black';
value bprod
low
- 145000 = 'red'
other
= 'white';
Here is the table that is created using these format definitions:
To highlight what is happening in this example, the background color (WHITE) used here does
not match the default background color for these cells. If you want the unformatted cells to blend
in, the values for the OTHER specification should be the same as the default attributes for that
ODS style.
248 Carpenter’s Complete Guide to the SAS REPORT Procedure
As in the example in Section 8.4.6, we could consolidate the three CALL DEFINE statements into
one, as follows:
compute actual;
if _break_ = ' ' then do;
call define('_C3_','style','style={background=bchair.
foreground=fchair.}');
call define('_C4_','style','style={background=bdesk.
foreground=fdesk.}');
call define('_C5_','style','style={background=btable.
foreground=ftable.}');
end;
else if _break_='REGION' then do;
call define(_row_,'style','style={background=bregn.
foreground=fregn.}');
end;
else if _break_='_RBREAK_' then do;
call define(_row_,'style','style={background=bprod.
foreground=fprod.}');
end;
endcomp;
The resulting table is slightly different. The background color for the overall total for sales (the
cell in the lower right corner) is now receiving trafficlighting formats and as a result has a white
background. This cell is now being assigned attributes because the CALL DEFINE statement has
a location of _ROW_, which includes the total sales column. Here is the resulting table:
Destination: HTML Style: DEFAULT
Chapter 8: Using PROC REPORT with ODS 249
8.4.8 A Few Things to Remember When Using Formats for
Trafficlighting
As has been discussed elsewhere in Section 8.4, a number of issues are associated with the use of
formats in the process of assigning style attribute values. Most of these are a result of timing
issues. The problems are complex enough, and come up often enough, to warrant the summary
that follows. Remember that these are only issues if you are assigning style attributes using
formats.
Trafficlighting formats cannot be applied to statistics that are defined only on the COLUMN
statement. Rather, the statistic should be associated with either an analysis variable or an alias
through the use of a DEFINE statement.
Both CALL DEFINE routines and STYLE= options can be used with trafficlighting formats, and
in some cases they can be used interchangeably. Both can be used in the same step.
As a general rule, trafficlighting effects that are desired for a given cell should be applied through
a single CALL DEFINE routine or STYLE= option. When two different sources address the same
cell, the resulting conflict can produce interesting, but unanticipated, results.
A CALL DEFINE routine in a compute block takes precedence over a corresponding STYLE=
option.
I generally prefer the CALL DEFINE routine to the STYLE= option. The CALL DEFINE routine
usually offers more control, especially when absolute column addressing is involved. However, I
often make an exception if a STYLE= option does the trick, and the use of a CALL DEFINE
would require an otherwise unnecessary compute block.
One caveat associated with the use of the CALL DEFINE routine is memory usage. For very large
reports or for memory-constrained environments, the STYLE= option might be preferred over
CALL DEFINE, as it tends to use less memory. It is anticipated that this will become even less of
an issue in SAS 9.2.
It is not a bad idea to specify your trafficlighting formats to cover the full range. Not doing so
allows the default style attributes to be displayed. However, in some cases, especially for reportwide summary lines (see Section 8.4.7), timing issues prevent the default attributes from being
correctly interpreted.
8.5 Embedding Hyperlinks within Your Table
As we move away from reports that are generated strictly for printing on paper, we can take
advantage of a number of techniques that can be used to link one table to another. In their more
sophisticated application, these techniques allow us to even link individual cells of our report to
another report or table. These links, or hyperlinks as they are more formally known, are used to
point from a specific location in one table to another table.
Generally your linked tables will all be of the same type (HTML, PDF, or RTF), but there is no
reason why this has to be the case. In the following examples, HTML tables link to HTML tables,
PDF tables link to PDF tables, and so on. However, when you create a reference to a file, it rarely
matters which of these file types you are pointing to or from. An HTML table can link to a PDF
file, for example.
250 Carpenter’s Complete Guide to the SAS REPORT Procedure
The process of moving from one table to a linked table of finer detail is known as drilling down,
and this is one of the most common applications of linked tables.
Obviously the process does not apply for documents that are not being viewed electronically.
However, through ODS we now have a number of choices of destinations that allow us to display
our documents in such a way as to take advantage of these techniques.
Section 8.5.1
Titles and footnotes can be used to form links to other documents or
locations within a document.
Section 8.5.2
The techniques in this section apply only to HTML files, but the links could
point to other file types.
Section 8.5.3
Unlike the STYLE= option, CALL DEFINE is an executable statement and
can therefore be controlled with logic in a compute block. This technique
can be used to create links for HTML, PDF, or RTF files.
Sections 8.5.4–
8.5.6
Although a number of destination-specific examples and considerations are
shown for HTML, PDF, and RTF files, there is a great deal of similarity
among techniques used in each of these destinations.
Section 8.5.7
Automating the process of building links through the use of the SAS macro
language can save time and increase accuracy.
Section 8.5.8
Links can also be built through the use of user-defined formats.
Because of the overlap among destinations, if you are new to linked documents or are not very
well versed in ODS, it will probably be wise to read over all of these sections, not just those
associated with the destination of interest.
SEE ALSO
The process of creating linked tables with drill-down capability is discussed by Gilbert (2005) and
Carpenter and Smith (2003a).
8.5.1 Linking Titles and Footnotes Using HTML Anchor Tags
and the LINK= Option
HTML Anchor Tags
Although some knowledge of HTML is helpful, it fortunately is not particularly necessary to
create linked HTML tables. You will, however, need to understand the basic structure of the
HTML anchor tag statement. Here is its general syntax:
display_text
When the HTML statement appears in a SAS title or footnote, the display_text is displayed. If the
display_text is selected by the reader, the browser then links to and displays the file named by the
HREF= option.
In the following somewhat silly example, three reports are generated. The first is the summary of
the two regions and then the detail reports for each of those regions. Each report is directly linked
to the other two through the FOOTNOTE statements, each of which contains HTML anchor tags.
Chapter 8: Using PROC REPORT with ODS 251
* Regional Report ***********************;
ods html style=default
path="&path\results" (url=none) n
body='ch8_5_1a_Region.html';
title1 'Region Summary';
footnote1 "Detail for Western Region";
footnote2 "Detail for Eastern Region";
o
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region product,actual;
define region / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods html close;
* Western Region Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_1a_RegionWEST.html';
title1 'Western Region Summary';
footnote1 "Region Summary";
footnote2 "Detail for Eastern Region";
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column region country product,actual;
define region / group;
define country / group; q
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods html close;
* Eastern Region Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_1a_RegionEAST.html';
title1 'Eastern Region Summary';
footnote1 "Region Summary";
p
252 Carpenter’s Complete Guide to the SAS REPORT Procedure
footnote2 "Detail for Western Region";
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='EAST'))
nowd;
column region country product,actual;
define region / group;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods html close;
n The URL=NONE option allows indirect addressing in the internal HTML code. Generally a
good idea anyway, this option allows you to move your linked images to other locations.
o Each set of footnotes always references the other two tables.
p The WHERE= option includes a subsetting clause for REGION.
q COUNTRY is added to the COLUMN statement as a grouping variable.
Chapter 8: Using PROC REPORT with ODS 253
Links can also be created through the use of the LINE statement. The linked footnotes used in the
previous example are replaced by LINE statements in the following example. Of course you can
still also change the text attributes as was done in Section 8.2.2.
The following is the PROC REPORT step that creates the overall summary:
title1; r
footnote1;
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region product,actual;
define region / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
compute before _page_; s
line @3 'Region Summary';
endcomp;
t
compute after; u
line @3 "Detail for Western Region";
line @3 "Detail for Eastern Region";
endcomp;
run;
ods html close;
r No title or footnotes are defined. Instead both are controlled with LINE statements.
s The LINE statement in this compute block will write at the top of the page.
t This becomes the report title.
u At the end of the report we write the two anchor tags, this time using the LINE statement
instead of the FOOTNOTE statement.
254 Carpenter’s Complete Guide to the SAS REPORT Procedure
Destination: HTML Style: DEFAULT
In some versions of SAS, you may need to have a and a tag surrounding the
anchor tag in the title or footnote so that the parser or processor will not interpret the angle
brackets (< >) as “less than” and “greater than” comparison operators.
Using the LINK= Option
Both the TITLE and FOOTNOTE statements support the LINK= option. This option enables you
to directly specify the link without using the anchor tags shown earlier in this section. Also, unlike
the anchor tags, which are used with the HTML destination, the LINK= option can generally also
be used with the PDF and RTF destinations.
The following PDF example takes the first example of this section and replaces the HTML anchor tags
with the LINK= option. Since PDF and RTF footnotes tend to be at the bottom of the page (rather than
at the bottom of the report), the footnotes have been replaced with titles for this example.
* Regional Report ***********************;
ods pdf style=printer
file="&path\results\ch8_5_1c_Region.pdf";
title1 'Region Summary';
title2 link='ch8_5_1c_RegionWEST.pdf'
"Detail for Western Region";
title3 link='ch8_5_1c_RegionEAST.pdf'
"Detail for Eastern Region";
Usually the LINK= option will work for both the PDF and RTF destinations. However, depending
on the level of PDF file created by your system and the word processor used to open an RTF file,
sometimes LINK= will not be able to create a valid link for those destinations.
MORE INFORMATION
Building a series of tables like these can be time-consuming and tedious. Fortunately, the macro
language excels at building this type of code. Section 8.5.7 generalizes this example using the
macro language. Additional options that can be used on the TITLE and FOOTNOTE statements
are discussed in Section 8.7.
Chapter 8: Using PROC REPORT with ODS 255
SEE ALSO
DeAngelis (2005) creates a variable that holds a URL that is used in a LINE statement.
8.5.2 HTML Anchor Tags as Data Values
Anchor tags can also be placed in data fields as well as column and row labels. The tags can be
built into a data value in a DATA step or in a compute block. Because the latter is more fun, this
is approach taken in the next example.
ods listing close;
* Regional Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_2_Region.html';
title1 'Region Summary';
footnote1;
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region regtag product,actual; n
define region / group noprint; o
define regtag / computed format=$4. 'Region'; p
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute regtag / char length=60; q
if region='WEST' then r
regtag = "West";
else if region='EAST' then
regtag = "East";
endcomp;
rbreak after / summarize;
run;
ods html close;
n The computed variable REGTAG is added to the COLUMN statement.
o The variable REGION is not printed.
p The computed variable that holds the anchor tag is defined.
q Even though only four characters are displayed, be sure to use a LENGTH= specification
sufficient to hold the whole anchor tag designation.
r The anchor tag for each region is assigned to the computed variable. By using more flexible
but slightly more complex code, we could reduce these two IF/ELSE IF statements into one
assignment statement (without the comparison). This refined statement will work for any
number of regions, and avoids the use of an IF-THEN/ELSE statement.
regtag = ""||trim(region)||"";
256 Carpenter’s Complete Guide to the SAS REPORT Procedure
r
Destination: HTML Style: DEFAULT
The code that generates the detailed reports for each of the two regions is similar to that shown
above. Instead of grouping on REGION, however, we are using COUNTRY. The link for the
summary tables for the individual regions points back to the regional summary t.
* Western Region Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_2_RegionWEST.html';
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column country ctag product,actual;
define country / group noprint;
define ctag
/ computed format=$7. 'Country'; s
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute ctag / char length=60;
if _break_='_RBREAK_' then t
ctag = "Total";
else ctag=country;
endcomp;
rbreak after / summarize;
run;
ods html close;
s The computed variable, CTAG, is defined to hold the anchor tag for country.
t On the line summarizing the region, an anchor tag points back to the overall summary.
Chapter 8: Using PROC REPORT with ODS 257
t
Destination: HTML Style: DEFAULT
MORE INFORMATION
This process can be automated somewhat through the use of the CALL DEFINE routine, which is
used in Section 8.5.3 to build HTML links. The use of HTML tags in report headers is discussed
in Section 9.3.3.
SEE ALSO
Squire (2003) and Don Li (2006) both discuss the creation of embedded links in HTML
documents. Yu and Shen (2006) place the links on numeric cells through the use of user-defined
formats.
8.5.3 Establishing Links Using CALL DEFINE
The examples in Section 8.5.2 create computed variables to hold the HTML anchor tags. Rather
than creating special computed variables, you can specify the file references directly by using the
CALL DEFINE statement.
The same links are created in the following examples as were created in Section 8.5.2. However,
here there are no computed variables. Instead, the CALL DEFINE statement is used with the URL
attribute to assign the URL to the column values. Although the examples in this section are for the
HTML destination, the URL also works for PDF and, depending on the word processor, the RTF
destination as well.
* Regional Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_3a_Region.html';
title1 'Region Summary';
footnote1;
258 Carpenter’s Complete Guide to the SAS REPORT Procedure
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region product,actual;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute region; n
rtag = "ch8_5_3a_Region"||trim(region)||".html";
call define(_col_,'url',rtag); p
endcomp;
rbreak after / summarize;
run;
ods html close;
o
n A compute block is established for the variable to which we want to assign the link.
o The temporary variable RTAG is used to store the location. Notice that the value to be
displayed is not included, only the location. The value of the variable REGION is the display
value. This makes our coding easier than in the similar example in Section 8.5.2. We can also
make the assignment of the location directly without first creating a temporary variable s.
p The location stored in the temporary variable RTAG is assigned as a URL attribute value for
this column.
In this example, we have decided that the summary for an individual region (here the “Eastern
Region Summary” is shown) will only link back to the primary table (“Region Summary”). This
means that the detail summaries for the individual regions will have a link only for one value in
the column.
* Western Region Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_3a_RegionWEST.html';
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
Chapter 8: Using PROC REPORT with ODS 259
column
define
define
define
nowd;
country
country
product
actual
product,actual;
/ group;
/ across;
/ analysis sum
format=dollar8.
'Sales';
compute country;
if _break_='_RBREAK_' then do; q
country = 'Region'; r
call define(_col_,'url',"ch8_5_3a_Region.html");
end;
endcomp;
rbreak after / summarize;
run;
ods html close;
s
q We only want to create the link for the summary line.
r The link will be available only if there is something for it to attach to in the cell. Because
country is otherwise missing (blank) for this summary row, we have added text to the cell.
s Rather than create a temporary variable (as was done at o), the value has been placed directly
into the CALL DEFINE statement.
In the previous example, three HTML files were created, each with links that pointed to other
files. It is also possible to create links that point to other locations within a document. The
following example builds on the previous example. However, rather than creating three files, it
creates a single file with three internal links.
When a link points to an internal location, an extension is added to the file name o using a pound
sign (#). In this case we add the region (EAST or WEST).
* Regional Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_3b.html'; n
title1 'Region Summary';
footnote1;
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region product,actual;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute region;
rtag = "ch8_5_3b.html#"||trim(region); o
call define(_col_,'url',rtag);
endcomp;
rbreak after / summarize;
run;
260 Carpenter’s Complete Guide to the SAS REPORT Procedure
n A single HTML file is used to hold all three reports.
o The internal link is named by appending the link identifier to the file name. The identifier
follows the pound sign (#). This identifier will be used in the ANCHOR= option p to tie the
individual reports together.
The two reports for the individual regions are generated without closing the HTML destination.
The ODS HTML statement p that precedes the REPORT step does not open a new report (there
is no BODY= option); it only exists to add the ANCHOR= option.
* Western Region Report
ods html anchor='WEST';
***********************;
p
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column country product,actual;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute country;
if _break_='_RBREAK_' then do;
country = 'Region';
call define(_col_,'url',"ch8_5_3b.html"); q
end;
endcomp;
rbreak after / summarize;
run;
p The ANCHOR= option provides the identifier that follows the pound sign (#) at o.
q Since there is no anchor specification (no #) for this link, this link points back to the top of
the report.
MORE INFORMATION
The example in Section 8.5.5 uses internal links in a PDF document.
SEE ALSO
Yeh (2004) uses CALL DEFINE to specify a URL in a COMPUTE BLOCK. Don Li (2006) uses
internal anchors to link files.
8.5.4 Forming Links Using STYLE=
Considering the considerable overlap between the capabilities of the CALL DEFINE routine and
the STYLE= option, it should not be surprising that you can form URL links by using the
STYLE= option as well. Because this option is not executable as is the CALL DEFINE routine, it
is more suitable when the link is either a constant or at least does not include data dependencies.
Chapter 8: Using PROC REPORT with ODS 261
In the following example, as in the example in Section 8.5.3, we want to link from the regionspecific summary back to the overall summary. The STYLE= option is used to form the link. The
code for the Western Region becomes the following:
ods html style=default
path="&path\results" (url=none)
body='ch8_5_4_RegionWEST.html';
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column region country product,actual;
define region / group
style(header)={url='ch8_5_4_Region.html'};n
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods html close;
n The header for REGION is made to be a link through the use of the URL attribute.
n
Destination: HTML Style: DEFAULT
Establishing links with the STYLE= option is also appropriate for PDF and RTF file types.
MORE INFORMATION
Issues associated with the direct use of HTML tags in report header text are discussed in
Section 9.3.3.
262 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.5.5 Creating Links in a PDF Document
The syntax for creating linked documents when using the PDF destination is very similar to each
of the previous methods used with HTML. The difference is in the appearance of the link on the
table, and how the files are addressed in the code (with an extension of PDF rather than HTML).
The example in Section 8.5.4 has been rewritten here to create a series of linked PDF documents.
For the PDF destination, the style has been specified as PRINTER. This is the default style for
PDF, but I like to explicitly specify the STYLE= option, even when it is the default. The code that
creates the table for the Western Region is shown:
ods pdf style=printer
file="&path\results\ch8_5_5a_RegionWEST.pdf";
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column region country product,actual;
define region / group style(header)={url='ch8_5_5a_Region.pdf'};
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods pdf close;
The resulting link is located on the column header. The report table for the Western Region
(ch8_5_5a_RegionWEST.pdf) is shown here:
Destination: PDF Style: PRINTER
In the previous example, three different documents are linked. As in Section 8.5.3, it would also
have been possible to create a single document with links pointing to other locations within that
one document. In the following example, a single PDF file is created with the same three
interconnected tables as in the previous example. However, the links all point to other places
within the same PDF file, rather than to other PDF files.
Chapter 8: Using PROC REPORT with ODS 263
* Regional Report ***********************;
ods pdf style=printer
file="&path\results\ch8_5_5b.pdf"; n
title1 'Sales Summary';
ods proclabel='Sales Summary'; o
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd
contents='Overall' p
;
column region product,actual;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute region;
rtag = "#"||trim(region); q
call define(_col_,'url',rtag);
endcomp;
rbreak after / summarize;
run;
* Western Region Report ***********************;
ods pdf anchor="WEST" r
startpage=now; s
ods proclabel="Western";
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd; t
column country product,actual;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
* Eastern Region Report ***********************;
ods pdf anchor="EAST"
startpage=now;
ods proclabel="Eastern";
title1 'Eastern Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='EAST'))
contents='' u
nowd;
column country product,actual;
define country / group ;
264 Carpenter’s Complete Guide to the SAS REPORT Procedure
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods pdf close;
n A single PDF file is written for all three PROC REPORT steps.
o The ODS PROCLABEL= option can be used to provide a name for the bookmark tab.
p The CONTENTS= option provides an additional location with the associated text on the
bookmark panel.
q The CALL DEFINE routine is used to create the association with the text and the specified
link. Notice that because the link is to be internal to the document, it is prefixed with a pound
sign (#). For this report, the two links are #WEST and #EAST.
r The ANCHOR= option is used to specify the link to the output from the upcoming procedure.
Notice that the link does not include a # (it is assumed).
s The STARTPAGE= option forces a new page in the PDF document.
t The bookmark area contains a default value when the CONTENTS= option is not included as it
was at p.
u The CONTENTS= option overrides the default contents value t that is displayed in the
bookmark section. When you want to suppress the value altogether, you should try using
CONTENTS=' ' (although there have been some problems reported with this option for
SAS 9.1).
o
p
t
u
Typically, bookmarks are created for each step between ODS PDF and ODS PDF CLOSE.
However, the bookmarks are linked only by the ANCHOR= option. The display of the bookmarks
can be controlled through the use of the BOOKMARKLIST= option.
* Regional Report ***********************;
ods pdf style=printer
file="&path\results\ch8_5_5b.pdf"
bookmarklist=hide;
The BOOKMARKLIST= option can take on the following values:
NONE
The bookmarks are not created.
HIDE
The bookmarks are created but not displayed (until requested).
SHOW
The bookmarks are displayed as in the previous example (this is the default).
Chapter 8: Using PROC REPORT with ODS 265
The table of bookmarks can also be turned off by using the NOTOC option on the ODS PDF
statement.
CAVEATS
As was mentioned earlier, the ODS CONTENTS= option does not always perform as expected in
SAS 9.1. This becomes more evident as the tables become more complex, and especially when
BREAK statements are included. Extensive changes are anticipated for SAS 9.2 that should
correct these problems.
It is hoped that PROC REPORT and PROC DOCUMENT will work together in SAS 9.2. If so,
tracking bookmarks should become much easier.
SEE ALSO
FAQ #4473 discusses links between pages within a PDF document, and FAQ #4148 discusses
PDF links in general. The CONTENTS= and other PDF bookmarking options are described by
Delaney (2003b). The STARTPAGE= option is discussed by Burlew (2005, p. 250).
Karunasundera (2006) discusses PDF bookmarks in more detail and includes a discussion of
corrections to their limitations.
8.5.6 Creating Links in an RTF Document
The generation of links in RTF is similar to the process used in both Sections 8.5.4 and 8.5.5.
* Western Region Report ***********************;
ods rtf style= rtf n
file="&path\results\ch8_5_6_RegionWEST.rtf";
o
title1 'Western Region Summary';
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE' and region='WEST'))
nowd;
column region country product,actual;
define region / group style(header)={url='ch8_5_6_Region.rtf'p};
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods rtf close;
n The RTF destination and the RTF file extension o create the RTF file that contains the table
with the embedded links. The RTF style has been especially designed for use with the RTF
destination.
p This extension could also be PDF or HTML if you wanted to link to a non-RTF file.
A portion of this report as it is viewed in Microsoft Word is shown here.
266 Carpenter’s Complete Guide to the SAS REPORT Procedure
When viewed in a word processor,
such as Microsoft Word, the
header “Region” is linked to
overall summary
ch8_5_6_Region.rtf. Although hard
to see in black and white, by
default the label is shown in an
alternate color.
Destination: RTF Style: RTF
When you want to follow a link from an RTF document, be sure to use the CTRL key with a
single click, rather than a double click.
SEE ALSO
FAQ #4019 has a couple of examples that create HTML hyperlinks in RTF documents. Burlew
(2005, pp. 225–232) creates hyperlinks in an RTF report using the URL= and URLLINK=
attribute options. Osowski and Fritchey (2006) use RTF control words to establish internal and
external links.
8.5.7 Automation Using the Macro Language
Building a series of linked tables by hand can be tedious. After two or perhaps three tables, I have
reached my tolerance for repeated code. Fortunately, the SAS macro language has a number of
extremely powerful techniques that can be used to automate the process of generating the
necessary code.
In the examples in Sections 8.5.2 through 8.5.6, one primary table is used as the index to point to
a series of secondary tables. This particular example uses only two secondary tables, one for each
region. However, if the number of regions were either unknown or perhaps dependent on the data,
we would need to write more flexible code.
The idea is to create code that removes data dependencies or hardcoded data elements. The
following code takes the example from Section 8.5.6 and generalizes it to work for any number of
regions. When you generalize code in this way, you need to watch for things like the following:
(where=(prodtype='OFFICE' and region='WEST'))
In this WHERE clause, both the value of PRODTYPE and REGION are hardcoded, and to
generalize for all regions we need to eliminate any hardcoded items that change within the
program.
The first step in the process is to determine the number of regions and their individual values. One
easy way to do this is to use a PROC SQL step to create a series of macro variables.
Chapter 8: Using PROC REPORT with ODS 267
%macro linked(prod=OFFICE);
%local i;
n
* Determine the count and list of regions;
proc sql noprint;
select distinct region o
into :reg1- :reg99 p
from sashelp.prdsale(where=(prodtype="&prod"));
%let regcnt = &sqlobs; q
quit;
n Because we are using macro %DO loops, we need to define a macro.
o We are interested in each distinct value of the variable REGION.
p Each individual value of REGION is saved into a macro variable of the form ®1,
®2, ®3,… . This code allows up to 99 distinct regions. There is no real penalty for
picking a number that is too big.
q The PROC SQL step counts the number of distinct values of REGION and stores them
temporarily in the macro variable &SQLOBS. This number is saved in the macro variable
®CNT.
The PROC REPORT step that creates the index table does not need to change, as it will
automatically adjust for each value of REGION r. In a sense it is already generalized.
* Regional Report ***********************;
ods rtf style=rtf
file="&path\results\ch8_5_7_Region.rtf";
title1 'Region Summary';
footnote1;
proc report data=sashelp.prdsale
(where=(prodtype="&prod"))
nowd;
column region product,actual;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute region;
rtag = "ch8_5_7_Region"||trim(region)||".rtf";
call define(_col_,'url',rtag);
endcomp;
rbreak after / summarize;
run;
ods rtf close;
r
r Because we are using the name of the region (which is the value of the variable REGION) as
the nonconstant part of the name of the file, this portion of the code is already data-independent
and does not require any further generalization.
Rather than creating a separate PROC REPORT step for each region, we generalize the step and
put it inside of a macro %DO loop s, which is executed once for each region (the number of
268 Carpenter’s Complete Guide to the SAS REPORT Procedure
regions is stored in ®CNT). Whenever we want to code for a particular value of REGION,
we use the indirect macro variable reference &®&I t.
%do i = 1 %to ®cnt; s
* Individual Region Report ***********************;
ods rtf style=rtf
file="&path\results\ch8_5_7_Region&®&it...rtf";
title1 "Region Summary for &®&i"t;
proc report data=sashelp.prdsale
(where=(prodtype="&prod" and region="&®&i"t))
nowd;
column region country product,actual;
define region / group
style(header)={url='ch8_5_7_Region.rtf'};
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods rtf close;
%end; s
%mend linked; u
%linked(prod=OFFICE) v
s The %DO loop cycles through ®CNT iterations. For each iteration the macro variable &I
is incremented by 1. %DO loop definitions are terminated with an %END statement.
t The value of the i region is stored in the macro variable &®&I. When &I is 2, this
becomes ®2, which for our example becomes WEST.
th
u The macro definition is terminated with a %MEND statement.
v The macro %LINKED is called.
SEE ALSO
A more complete discussion of the process of automatically generating links can be found in
Carpenter and Smith (2003a) and Carpenter (2004a, Section 10.4). Additional discussion of the
macro language itself can be found in Carpenter (2004a).
8.5.8 Using Formats to Build a Link
The links can also be established through the use of user-defined formats. Using this technique
enables you to store the link in a format rather than in the code itself. This approach has the
advantage of not having hardcoded links embedded within the code. To change a link, all we have
to do is change the format.
Here the example from Section 8.5.2 is rewritten using a format to hold the links that point to the
secondary tables.
Chapter 8: Using PROC REPORT with ODS 269
proc format;
value $regtag n
'WEST' = "West"
'EAST' = "East";
run;
* Regional Report ***********************;
ods html style=default
path="&path\results" (url=none)
body='ch8_5_8_Region.html';
title1 'Region Summary';
footnote1;
proc report data=sashelp.prdsale
(where=(prodtype='OFFICE'))
nowd;
column region product,actual;
define region / group format=$regtag40. o 'Region';
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
rbreak after / summarize;
run;
ods html close;
In the similar example in Section 8.5.2, a computed column is created to hold the link. This
becomes unnecessary when you use this technique.
n The format $REGTAG defines the HTML anchor tags that are used to form the groups in
the PROC REPORT step.
o The format is used directly against the grouping variable. Only the display_text portion of the
formatted value appears in the table.
o
Destination: HTML Style: DEFAULT
SEE ALSO
Burlew (2005, p. 229) creates hyperlinks in a user-defined format for an RTF report.
270 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.6 Using the Escape Character for In-Line
Formatting
The timing of events can often be important when you are dealing with the Output Delivery
System. Because SAS is an interpreted language, we are used to having the code that we write
translated into actions. But as we take the results of PROC REPORT and render them using ODS,
sometimes we want to pass instructions to the ODS process itself. At other times, usually when
using RTF, we want some portion of our code to be interpreted not by ODS at all, but by the
application, such as Microsoft Word, that receives the document.
Because a sequence of events always takes place, we must have some way of marking code that is
to be interpreted differently, or that is to be interpreted later by a different process. In SAS this
marking is done with an escape character, which is designated by the ODS ESCAPECHAR=
option. You can use almost any character as an escape character, but it is generally recommended
that you use a character that does not already have meaning in your code.
In the examples that follow in this book, the escape character is specified as the tilde (~).
It is important to note that the discussion in this section does not apply to all ODS destinations
and versions of SAS equally. In SAS 9.1.3 the escape character has been fully implemented and
should work for a majority of the described techniques and formatting functions in the PDF
destination, for most in the RTF destination, and only generally in the HTML destination. Check
the documentation for destination specifics, or easier yet, try the destinations using the
supplemental SAS programs that are referenced in this book.
The escape character is used with special escape character strings and special escape character
functions to pass information to ODS and at times beyond ODS. Functionality within the Output
Delivery System is evolving rapidly but currently these take several general forms.
Formatting functions
Used to control pagination, superscripts, and subscripts
General form:
~{function }
Discussed in Sections 8.6.1 - 8.6.3
S={ } and {style .... }
Used to assign style attributes
General form:
~S={attribute characteristics}
~{style style elements and attributes}
Discussed in Section 8.6.4
Escape character sequence codes
Used to manipulate line breaks, wrapping, and indentations
General form:
~code
Discussed in Section 8.6.5
Chapter 8: Using PROC REPORT with ODS 271
Raw text insertion sequence codes
Used to insert destination-specific codes
General form:
~R/destination"rawtext"
~R"rawtext"
Discussed in Section 8.6.6
Significant changes in the use of escape character sequences are anticipated starting in SAS 9.2;
however, the current syntax (SAS 9.1.3) is expected to continue to work.
MORE INFORMATION
In addition to the SAS programs discussed in the examples, the test program s8_6TestAll.sas can
be used to further test the various in-line sequences, codes, and functions described in this section.
You will find s8_6 TestAll.SAS in the sample code on the bonus CD that accompanies this book.
SEE ALSO
Lund (2005), Parsons (2006), Sevick (2006), and Gianneschi (2006) each demonstrate many of
the techniques that utilize the escape character in a series of examples. Huntley (2006) discusses
some of these techniques and shows some new ones that are anticipated for SAS 9.2.
The original documentation for in-line formatting sequences can be found at
http://support.sas.com/rnd/base/topics/expv8/inline82.html
8.6.1 Controlling Superscripts and Subscripts
The in-line formatting functions SUPER and SUB can be used to generate superscripts and
subscripts. The functions take the form of ~{super text} and ~{sub text}. For the supported
destinations, these functions are not at all restricted to PROC REPORT, and (excluding
SAS/GRAPH) can be used wherever text is specified. These text locations include titles, headers,
labels, and even formatted values.
The ODS ESCAPECHAR= option is used to designate the character that you would like to use
as the escape character n. When used, the escape character is followed by curly brackets that
contain the in-line formatting function and the text to which it is to apply o.
ods pdf style=printer
file="&path\results\ch8_6_1a.pdf";
ods escapechar = '~'; n
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height~{super 1}'
define weight / analysis mean
format=6.2
'Weight~{super 2}'
rbreak after / summarize;
o;
o;
272 Carpenter’s Complete Guide to the SAS REPORT Procedure
compute after;
line @3 '~{super 1}
line @3 '~{super 2}
endcomp;
run;
ods pdf close;
p Mean height in inches.';
p Mean weight in pounds.';
n The tilde (~) is designated as the escape character.
o The 1 and 2 will both appear as superscripts.
p A footnote containing the superscripts is generated using LINE statements.
Here is the resulting PDF table:
Destination: PDF Style: PRINTER
Superscripts and subscripts can also be generated using these same in-line functions in RTF and
HTML destinations.
These in-line formatting functions can also be specified in formats. In the following example, the
superscripts are placed on formatted values through the use of a user-defined format.
ods pdf file="&path\results\ch8_6_1b.pdf";
ods escapechar = '~';
proc format;
value $mgen
'M' = 'Male~{super 2}' q
'F' = 'Female~{super 1}';
run;
title1 "Ages 11 - 16";
proc report data=sashelp.class nofs;
columns sex height weight;
define sex
/ group format=$mgen.;
define height / analysis mean
format=5.2
'Height';
define weight / analysis mean
format=6.2
'Weight';
rbreak after / summarize;
r
Chapter 8: Using PROC REPORT with ODS 273
compute after;
line @3 '~{super 1} Girls Swim Team';
line @3 '~{super 2} Boys Soccer Team';
endcomp;
run;
ods pdf close;
q The in-line superscript function is placed in the VALUE statement.
r The format is applied to the grouping variable.
s The formatted value contains a superscript.
s
Destination: PDF Style: PRINTER
8.6.2 Displaying Page Numbers
With the exception of the RTF destination, SAS determines page numbers directly. However,
sometimes we want a bit more control over how the page number is to be displayed. This is
especially true when we want to display the total page count along with the current page. Here is
one such commonly requested pagination scheme:
Page x of y
Generation of this kind of page counting in RTF is especially problematic for SAS, because the
page count determination is not made until after SAS has relinquished control. Once the
document is rendered in the final application, such as Microsoft Word, the page count can be
determined. This means that the pagination must take place when the document is finalized—
long after SAS has completed its part of the process. SAS solves this problem by using in-line
formatting functions to pass pagination instructions, which can be executed when the pages can be
determined.
Three in-line formatting functions can be used with page numbering. These are {THISPAGE},
{LASTPAGE}, and {PAGEOF}. Although these formatting functions do not behave the same or
even necessarily work for both the RTF and PDF destinations, they can still be very useful.
Because these in-line functions are used to pass destination-specific instructions to the rendering
software, there are some issues with regard to these functions and the parts of the report in which
they are being used.
274 Carpenter’s Complete Guide to the SAS REPORT Procedure
In-Line Paging
Function
Can Be Used in
Report Body
Areas
Can Generally Be
Used in Page Header
Areas
{PAGEOF}
---
RTF only
{THISPAGE}
PDF only
RTF and PDF
{LASTPAGE}
---
RTF and PDF
Notice that HTML does not support these paging functions.
Using {PAGEOF}
The in-line formatting function {PAGEOF} does not work for the PDF destination; however, it
can be very useful in the RTF destination. In the following example, a new page is generated for
each value of the BY variable. The page counts are noted using a LINE statement r.
ods rtf style=rtf
file="&path\results\ch8_6_2a.rtf"
bodytitle;
ods escapechar='~';
n
proc sort data=sashelp.prdsale
out=prdsale;
by prodtype;
run;
option nobyline; o
title1 '#byval1'; p
proc report data=prdsale
nowd;
by prodtype; q
column region product,actual;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute after _page_;
line @3 'Page ~{pageof}'; r
endcomp;
rbreak after / summarize;
run;
ods rtf close;
n The tilde is designated as the escape character for use with the in-line formatting functions.
o The NOBYLINE option turns off the BY line generated by PROC REPORT (since it is going
to be displayed in the title through the use of #BYVAL1 p).
p The value of the first BY variable (PRODTYPE) is placed into the title.
q The BY PRODTYPE forces a new page for each value of the variable PRODTYPE.
r The in-line function {PAGEOF} is used to form the x of y portion of the page number.
Chapter 8: Using PROC REPORT with ODS 275
Here are the two pages of the report:
Destination: RTF Style: RTF
One of the really nice features of PROC REPORT when you use the BY statement is that the
pages do not necessarily have to have the same columns. In these two pages, the products within
product type are completely distinct.
When you use the RTF destination, the page numbers usually do not show up until the document
is either printed, viewed through a Print Preview window, or included in another document. This
means that when you view the document in the SAS viewer, the page numbers will probably not
all be visible. Remember that the page number is actually not simple text, but an embedded
command that expresses the page number when the document is finalized. Delays in the
expression of the page numbers can occur, depending on the size and complexity of the document.
Using {THISPAGE} and {LASTPAGE}
The {THISPAGE} and {LASTPAGE} formatting functions are similar to {PAGEOF} in their
usage, but because they are separate functions, they give us a bit more flexibility. These two
functions can, in most cases, be used with reports that are to be written to both the PDF and RTF
destinations. The following example repeats the previous example using these two functions. In
addition, it adds them to the title for demonstration purposes and writes to both the PDF and the
RTF destinations.
ods pdf style=printer
file="&path\results\ch8_6_2b.pdf";
ods rtf style=rtf
file="&path\results\ch8_6_2b.rtf"
bodytitle;
ods escapechar='~';
proc sort data=sashelp.prdsale
out=prdsale;
by prodtype region;
run;
276 Carpenter’s Complete Guide to the SAS REPORT Procedure
options nobyline;
title1 '#byval1';
title2 'In the Title: Page ~{thispage} out of ~{lastpage} ';
footnote1;
proc report data=prdsale
nowd;
by prodtype;
column prodtype region product,actual;
define prodtype / group page;
define region / group ;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
compute after _page_;
line @3 'Page ~{thispage} out of ~{lastpage}';
endcomp;
rbreak after / summarize;
run;
ods _all_ close;
Here is the first page of the RTF file:
Destination: RTF Style: RTF
Chapter 8: Using PROC REPORT with ODS 277
Here is the first page in the PDF file:
Destination: PDF Style: PRINTER
Notice that although both the functions work correctly in the RTF table, in the PDF table the
paging generated with the LINE statement is not only compressed, but the value for
{LASTPAGE} is not available.
MORE INFORMATION
The #BYVAL option for the TITLE statement is described in Section 6.6.3.
SEE ALSO
Chung and Dunn (2005) and DeAngelis (2005) both discuss how to create a page x of y counter
by counting the pages themselves. Hamilton (2003) and Smoak (2004) both use RTF commands
to generate the pagination in SAS 8.2. FAQ #4010 shows examples of the generation of page x of
y. Pagination is controlled with help from PROC TEMPLATE in FAQ #4473. FAQ #4450 has an
example that uses PAGEOF, THISPAGE, and LASTPAGE.
8.6.3 Generating a Dagger
A symbol commonly called a dagger is often used instead of a numeric superscript (or subscript)
to call attention to a footnote. The dagger symbol can be generated directly through the use of the
in-line {DAGGER} function. The syntax is similar to that used with {PAGEOF}.
278 Carpenter’s Complete Guide to the SAS REPORT Procedure
ods pdf style=printer
file="&path\results\ch8_6_3.pdf";
ods escapechar = '~';
title1 "Ages 11 - 16 ~{dagger}";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height';
define weight / analysis mean
format=6.2
'Weight';
rbreak after / summarize;
compute after;
line @3 '~{dagger} Data extracted from the ABC study';
endcomp;
run;
ods pdf close;
Destination: PDF Style: PRINTER
The dagger symbol can also be generated equally well in RTF and HTML, although in HTML the
symbol looks more like a cross than a dagger.
Chapter 8: Using PROC REPORT with ODS 279
8.6.4 Using the Escape Character with S={ } and {STYLE} to
Change Style Attributes
In Sections 8.2 and 8.4.2, the STYLE= option was discussed as it pertains to various aspects of
PROC REPORT, and it was shown that this option can modify or change a wide variety of style
attributes. Many of these same attributes can be changed using the ~S={ } in-line formatting
strings and the nested ~{STYLE} formatting sequence.
~S={ } In-Line Formatting Syntax
Here is the general syntax to set attribute values:
~S={style_attribute=attribute_value}
The tilde (~) is the designated escape character, and the S= must be in uppercase.
Empty curly brackets are used to turn off or reset attributes that were specified using the previous
in-line sequence:
~S={}
There are a couple of advantages of this approach over the STYLE= option. First, it can be used
where the STYLE= option does not apply, such as within titles and footnotes. Second, the in-line
formatting string can be stored as part of a character variable and as such can be used in variable
values and formats.
Using ~S={ } in Titles and Footnotes
In the following example, we use in-line formatting to change the attributes of the text in the
TITLE and LINE statements.
ods pdf style=printer
file="&path\results\ch8_6_4a.pdf";
ods escapechar = '~';
title1 "~S={font_face=Arial}Ages "
"~S={font_style=roman}11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height';
define weight / analysis mean
format=6.2
'Weight';
rbreak after / summarize;
n
compute after;
line @3 '~S={font_weight=bold}Height(in.)'
'~S={font_weight=light}Weight(lbs.)';
endcomp;
run;
ods pdf close;
o
280 Carpenter’s Complete Guide to the SAS REPORT Procedure
n By default the title line is in italics; however, here italics have been removed from the
numbers in the title.
o The bolding is removed from the second half of the text generated by the LINE statement. The
in-line string ~S={} also removes previous attributes (reestablishes the defaults). The LINE
statement could also have been specified as follows:
compute after;
line @3 '~S={font_weight=bold}Height(in.)'
'~S={}Weight(lbs.)';
endcomp;
Destination: PDF Style: PRINTER
Using ~S={ } in Formats
Because you can use the in-line formatting ~S={} sequence in most instances that you can specify
text, it stands to reason that you could use it in a format as well. The following example builds a
format that maps the values of SEX ( ‘F’ and ‘M’) into ‘Female’ and ‘Male’, while at the same
time bolding the first letter.
ods pdf style=printer
file="&path\results\ch8_6_4b.pdf";
ods escapechar = '~';
proc format;
value $genttl
'f','F'='~S={font_weight=bold}F~S={font_weight=light}emale'
'm','M'='~S={font_weight=bold}M~S={font_weight=light}ale';
run;
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group format=$genttl.;
define height / analysis mean
. . . Portions of the code are not shown . . .
Chapter 8: Using PROC REPORT with ODS 281
The generated table shows:
Destination: PDF Style: PRINTER
Nested In-Line Style Attributes Using ~{STYLE}
Starting in SAS 9.2, the ~S={} technique will be augmented by the more flexible syntax known as
nested in-line formatting. The ~S={} syntax will continue to work as described earlier in this
section; however, the newer syntax will give the user better control.
Here is the general syntax:
~{style <[attribute(s)]>formatted text}
The syntax always begins with ~{STYLE.
style_element
specifies the portion of the report to which the style attributes are to be
applied e.g., headerfixed, systemtitle.
attribute(s)
specifies the attributes, enclosed in square brackets. Multiple attributes can
be specified.
formatted_text
specifies the text to which the style elements and attributes will be applied.
This syntax is most similar to ~S={} when the style element is not specified. For example,
~{style [color=red] text}
is equivalent to
~S={color=red} text.
The following TITLE statement uses the HEADERFIXED style element.
title "~{style headerfixed title with style element headerfixed}";
In the previous statement, the element following the ~{style is the optional style_element. In this
case there is no attribute specified in square brackets, so the text to be formatted follows the
style_element. The following TITLE statements include the use of style_elements.
282 Carpenter’s Complete Guide to the SAS REPORT Procedure
title2 "~{style [color = greenish blue] title in greenish blue color}";
title3 "~{style headerstrong[color = dark red fontstyle=italic]
title in dark red as headerstrong element}";
title4 "test of ~{super ~{style [color=red] red
~{style [color=green] green} and
~{style [color=blue] blue } formatting }}
etc.";
MORE INFORMATION
You can use the sample program s8_6TestAll.sas (primarily written by Cynthia Zender) to test inline formatting sequences; this program is included among the bonus programs on the CD that
accompanies this book.
SEE ALSO
Huntley (2006) discusses these and other text modification techniques. Further examples of the
use of ~S={} can be found in Lund (2005) and Morgan (2006).
8.6.5 Line Breaks and Wrapping
When long text strings do not fit in the space that has been allocated for the text, we often need to
have the text wrap. Although the FLOW option has some utility in the LISTING destination, we
might need to take even more control. One way to achieve this control is through the use of in-line
formatting text sequences. While these sequences should all work in PDF, they do not work
equally well in each of the three primary destinations (RTF, PDF, and HTML). You might need to
experiment a bit with your destination and version of SAS.
The following set of in-line commands tells the destination where to break a line of text and,
optionally, where to establish the indentation location when wrapping.
m
specifies the location for indentation of subsequent wrapped lines (without this
marker subsequent lines are left-justified).
-2n
forces a line break that takes the ~m location into consideration.
xn
forces x line breaks that do not take the ~m indentation location into consideration.
w
specifies the suggested location for an optional line break.
_
(underscore) creates a nonbreaking space.
You can also insert up to four types of error conditions into the table.
xz
inserts error codes. The z is lowercase, and the value of x determines the type of
error code, as follows:
x=1
x=2
x=3
x=4
ERROR:
WARNING:
NOTE:
FATAL:
Line Breaks and Indentations
Lines of text can be split by using the ~xn sequence. The value of x, except for the special case of
x = -2 that is described in the next paragraph, determines how many line feeds to include.
Chapter 8: Using PROC REPORT with ODS 283
If you want to mark a location for indentations for subsequent lines, you can use the ~m code.
However, if you want to force a line break and you want the next line to be indented at the
location specified by the ~m, you need to use the special split sequence ~-2n as the line split
character.
These breaks and indentations are demonstrated in the following example, in which the split
sequences are placed within a format definition n as well as within a LINE statement o, “just
because we can.”
ods html style=default
file="&path\results\ch8_6_5a.html";
ods pdf style=printer
file="&path\results\ch8_6_5a.pdf";
ods rtf style=rtf
file="&path\results\ch8_6_5a.rtf"
bodytitle;
ods escapechar = '~';
proc format;
value $genttl
'f','F'='Fe~mmale~-2nStudents'
'm','M'='Ma~mle~-2nStudents';
run;
n
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group format=$genttl.;
define height / analysis mean
format=5.2
'Height';
define weight / analysis mean
format=6.2
'Weight';
rbreak after / summarize;
compute after;
line @1 'Eng~molish Measures~-2npHeight(in.)~nqWeight(lbs.)';
endcomp;
run;
ods _all_ close;
n The indentation (margin) marker (~m) and the line feed are specified in a formatted value.
Notice that although the line feed is specified as -2n, only one line feed is introduced. The -2n
is just a code, and the 2 does not indicate the number of line feeds to introduce. A positive
number, such as 2n, inserts that number of line feeds.
o A margin marker is set to be aligned with the third letter in the text string.
p The line is split using ~-2n. This split sequence recognizes the ~m as an indentation marker.
q The line split (~n) successfully splits the line, but the margin marker is not recognized and the
line is left-justified. Only line feeds induced by the code -2n recognize the margin marker
(~m).
284 Carpenter’s Complete Guide to the SAS REPORT Procedure
The indentation marker ~m o is only
recognized when the line is split with a ~-2n.
Hence “Weight” is left-justified, whereas
“Height” is indented.
Destination: PDF Style: PRINTER
Notice that the text for “Weight(lbs.)” is left-justified. The margin marker (~m) is only recognized
in the PDF destination (and then only when the text is either wrapped naturally or split using the
~-2n sequence). The other line splits are recognized in RTF and HTML.
Placing Nonbreaking Spaces in Your Text
If you have text that is either being allowed to flow or is otherwise wrapping, it is possible to
replace a blank with a nonbreaking space. The ~_ (underscore) designates a nonbreaking space.
Multiple adjacent nonbreaking spaces can be specified by placing a series of marked
underscores—e.g., ~_~_~_.
The following example places nonbreaking spaces before ‘Asia’ n, between ‘United’ and ‘States’
o, and in the header text for REGION p. Leading nonbreaking spaces can be used to change the
order of the rows.
ods html style=default
file="&path\results\ch8_6_5b.html";
ods pdf style=printer
file="&path\results\ch8_6_5b.pdf";
ods rtf style=rtf
file="&path\results\ch8_6_5b.rtf"
bodytitle;
proc format;
value $NewReg
'Asia' = '~_~_~_~_~_~_~_~_~_~_~_~_Asia' n
'United States' = 'United~_~_~_~_~_States'o;
run;
Chapter 8: Using PROC REPORT with ODS 285
ods escapechar='~';
title 'Total Sales';
proc report data=sashelp.shoes nowd;
column region sales;
define region / group 'Region~_~_~_~_Name'
format=$NewReg.;
define sales / sum 'Sales';
run;
ods _all_ close;
In the HTML table “Asia” is indented and it is sorted LAST.
Destination: HTML Style: DEFAULT
p
286 Carpenter’s Complete Guide to the SAS REPORT Procedure
In the PDF table, the leading nonbreaking spaces for “Asia” are not used to form an indentation;
however, they are still used by PROC REPORT to determine the ordering of the rows.
Destination: PDF Style: PRINTER
In the RTF table, as in the PDF table, the leading nonbreaking spaces are not used except when
ordering the rows of the table.
Destination: RTF Style: RTF
Chapter 8: Using PROC REPORT with ODS 287
When you are working with a RTF table in most word processors, such as Microsoft Word, it is
possible to adjust the widths of the columns dynamically. During the adjustment, an attempt is
made to split text that does not fit in the allocated width at spaces in the text. Nonbreaking spaces,
however, are treated like other characters and are not used to break the text for wrapping. The
following report redisplays the previous RTF table after the first column has been given a reduced
width. Notice how “South America” has split on the space, while “United States” has not.
Destination: RTF Style: RTF (with the first column resized)
Inserting Error and Other Warning Text
The ~xz in-line formatting sequences can be used to insert warning and error text. The type of text
is determined by the value of x, as follows:
x=1 ERROR:
x=2 WARNING:
x=3 NOTE:
x=4 FATAL:
In the following example, the macro variable &RC has been set as a return code from the process
that builds the data (this process is not shown as we are just pretending anyway). In this example,
&RC=2. The macro variable’s value is used to create the value of x, and the resulting in-line
sequence (~2z) is then used to create the warning in the LINE statement.
288 Carpenter’s Complete Guide to the SAS REPORT Procedure
ods html style=default
file="&path\results\ch8_6_5c.html";
ods pdf style=printer
file="&path\results\ch8_6_5c.pdf";
ods rtf style=rtf
file="&path\results\ch8_6_5c.rtf"
bodytitle;
ods escapechar = '~';
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height';
define weight / analysis mean
format=6.2
'Weight';
rbreak after / summarize;
compute after;
line @1 "~&rc.z Return Code Status";
endcomp;
run;
ods _all_ close;
Before execution of the LINE statement, the &RC resolves to 2 (presumably &RC was set in
some previous step). The LINE statement then becomes the following:
line @1 "~2z Return Code Status";
The sequence ~2z becomes the word “Warning:”
Instead of “Warning:” escape sequences
could also have been used to produce
“Note:”, “Error:”, and “Fatal:”.
Destination: PDF Style: PRINTER
Chapter 8: Using PROC REPORT with ODS 289
In both the PDF and RTF destinations, the LINE statement produces the desired text. However,
for the HTML destination, the ~2z is ignored.
MORE INFORMATION
Controlling the wrapping of text using the TAGATTR= attribute is discussed in Section 9.3.2, and
using the FLOW option for the LISTING destination in Section 4.5.2.
SEE ALSO
Lund (2005) includes examples that split lines of text.
8.6.6 Passing Raw Destination-Specific Codes
When writing to the HTML and RTF destinations, you can use in-line code sequences to pass raw
destination-specific codes directly into the resultant file. This can be useful when you understand
the specifics of the internal workings of the destination, and you have a need that cannot
otherwise be addressed using other ODS options or techniques.
Here is the syntax used to place raw code directly into the destination file:
~R/destination"command and text"
The destination can be HTML or RTF, and the command and text is destination-specific. The raw
text is only written to the appropriate destination. If the destination is not specified as in
~R"command and text"
the text is written to each appropriate destination in the PRINTER family (currently HTML, RTF,
PCL, PRINTER, or PS).
Sample text strings that use destination-specific codes include the following, which are taken
from the sample program s8_6TestAll.sas.
Insert RTF tabs:
'Begin~R/RTF"\tab\tab" Two Tabs Later'
Highlight selected RTF text:
'Before Test ~R/RTF"\highlight3" After Test'
Turn on RTF underlining and split a line:
'~R/RTF"\ul" 1st Line~R/RTF"\line" 2nd Line'
Change text colors in HTML:
'The color is ~R/HTML""'||
'BLUE (HTML ONLY)~R/HTML""'
290 Carpenter’s Complete Guide to the SAS REPORT Procedure
The following example highlights these and some other RTF control words:
ods escapechar = '~';
title1 '~R/RTF"\ul " Title is italic and contains underlined text';
n
data rtfcontrol;
text = 'This is default text.';
run;
ods rtf file = "&path\results\8_6_6a.rtf"
style=rtf;
PROC REPORT data = rtfcontrol nowd
style(header column)=[font_face='arial' font_weight=bold];
column text my_text;
define text
/display 'Default';
define my_text/computed width=1 'RTF Control Words' flow;
compute my_text / char length=200;
my_text = "The text uses RTF control words " ||
'~R"\i italic text \i0 "' || " regular text " ||
'~R"\ul underlined text \ul0 "' ||
'~R"\strike strike text \strike0 "' || ".";
endcomp;
o
run;
ods rtf close;
n Underlining in the TITLE is accomplished with the ~R/RTF "\ul" command word. In the
TITLE the command “\ul” must be enclosed in quotation marks. In this case, the title is already
going to be displayed in italics, so only the underline is specified. Because /RTF is included,
the underlining only applies to the RTF destination.
o For the RTF destination, italics are turned on with the "\i" RTF control word. The italics are
also turned off with the "\i0". Notice that RTF is not specified in the control commands used
in this assignment statement. Unlike the command in the title n, these commands are applied
to all appropriate destinations (of course, in this case there is only one).
Destination: RTF Style: RTF
Chapter 8: Using PROC REPORT with ODS 291
Quotation marks can be an issue when using these control words. Often the control words
themselves need to be enclosed in quotation marks, as well as the text to which they are to be
applied. The text sequences in this section follow this basic pattern:
"text" || '~R"\rtf-control-word raw-text \rtf-control-word0 "'
Either double or single quotation marks are used around text strings without control
words.
The concatenation operator ( || ) is used to concatenate all strings.
When nested quotes are needed, double quotation marks are used around the RTF control
word string and affected text, and then single quotation marks are used to surround the
~R sequence.
Although it is not always needed, you should close each RTF control word string with a
zero ( 0 ).
Because macro variables do not resolve inside of single quotation marks, this embedded quoting
can become a problem when you are including macro variables. Let us assume that the two macro
variables &CODE and &TXT have been defined as follows:
%let code = ul;
%let txt = underlined text;
The RTF codes specified in the following assignment statement will not work, because the macro
variables are inside of single quotation marks and will not be resolved.
compute my_text / char length=200;
my_text = "RTF control word as a macro variable "
'~R"\&code &txt \&code.0 "' || ".";
endcomp;
||
Instead, we need to mask the single quotation marks from the macro parser by using a macro
quoting function such as %BQUOTE.
compute my_text / char length=200;
my_text = "RTF control word as a macro variable " ||
%bquote(')~R"\&code &txt \&code.0 "%bquote(') || ".";
endcomp;
Now the macro variables &CODE and &TXT can be resolved (to ul and underlined text
respectively), and it is as if we had written the compute block as follows:
compute my_text / char length=200;
my_text = "RTF control word as a macro variable "
'~R"\ul underlined text \ul0 "' || ".";
endcomp;
||
SEE ALSO
The second example in this section was suggested by Sunil Gupta, author of Quick Results with
the Output Delivery System (Gupta, 2003). More information on the RTF specification can be
found at
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnrtfspec/html/rtfspec.asp
292 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.7 Using TITLE and FOOTNOTE Statement
Options
For the most part, the destinations that support font and color control also accept attribute control
options in the TITLE and FOOTNOTE statements. These options were originally included in
these two statements for use with SAS/GRAPH and are otherwise ignored unless you are writing
to one of the supported ODS destinations.
These TITLE/FOOTNOTE statement options include the following:
Color=
designates color.
BColor=
specifies the background color.
Height=
specifies the height of the text (usually specified in points).
Justify=
specifies the text justification (left, center, right).
Font=
designates the font (can include hardware and software fonts).
Most of these options can be abbreviated using the uppercase letters in the option name shown
here.
There are also a few font modification options. These include:
BOLD
creates bolded text.
ITALIC
italicizes the text.
UNDERLINE
underlines the text.
Colors can include any of the RGB or gray scale colors appropriate for the output destination.
When used, these options precede the text to which they are to apply. The following example
highlights a few of these options.
. . . Portions of the code are not shown . . .
ods html style=default
file="&path\results\ch8_7.html";
title1 f='times new roman' h=20pt c=blue bc=yellow 'Student Summary';
title2 f='Arial' h=20pt c=red j=l bold 'English Units';
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group format=$genttl.;
. . . Portions of the code are not shown . . .
Chapter 8: Using PROC REPORT with ODS 293
This example shows only a few of the TITLE
and FOOTNOTE statement options that can
be used to modify the text characteristics.
Destination: HTML Style: DEFAULT
The UNDERLINE option is not expected to be available for the PDF destination until SAS 9.2.
MORE INFORMATION
The LINK= TITLE/FOOTNOTE option is discussed in Section 8.5.1, and the Font= option is
used in the example in Section 10.5.1.
SEE ALSO
Additional information on these TITLE and FOOTNOTE options can be found in the
documentation for SAS/GRAPH.
8.8 Creating Tip or “Flyover” Text for HTML and
PDF
Tip text or “flyover” text pops up into view when the cursor hovers or points over a location in the
file. You can create this text for both HTML and PDF destinations. Tip text can be very
advantageous if you have limited space on your table, but want to have additional information
available to the reader at the reader’s discretion.
The attribute that needs to be set is FLYOVER, and the character string assigned to FLYOVER
becomes the tip text.
8.8.1 Using CALL DEFINE
In the following example, the data is summarized for clinic number and the clinic name is not
shown. Instead the clinic name is added as tip text.
294 Carpenter’s Complete Guide to the SAS REPORT Procedure
. . . Portions of the code are not shown . . .
title1 "Clinic Summaries";
proc report data=rptdata.clinics nowd;
columns clinname clinnum ht ht=htmean wt; n
define
define
define
define
define
clinname/
clinnum /
ht
/
htmean /
wt
/
rbreak after
group noprint; o
group 'Clinic Number';
analysis n 'N';
analysis mean 'Height';
analysis mean 'Weight';
/ summarize;
compute clinnum; p
attrib = 'style={flyover="'||trim(clinname)||'"}';
call define(_col_,'style',attrib); r
endcomp;
run;
ods _all_ close;
q
n The clinic name is on the column statement to the left of the clinic number.
o The clinic name is used as a nonprinting group variable. Because it is to the left of CLINNUM
in the column statement, its value is available in the assignment statement that builds the
temporary variable ATTRIB in the CLINNUM compute block p.
p A compute block is established for the variable whose values are to receive the tip text. This
enables us to use the CALL DEFINE statement.
q The STYLE= option specifies the clinic name as the tip text. For clinic number 049060, the
value of ATTRIB would be as follows:
attrib = 'style={flyover="Atlanta General Hospital"}';
Notice the use of the double and single quotation marks. Here we are constructing a temporary
character variable (ATTRIB) that is composed of character constants (enclosed in single quotation
marks) and a report item (CLINNAME) whose value is to be included inside of quotation marks
(double quotation marks in this case).
r CALL DEFINE is used to set the FLYOVER attribute. Notice that the temporary variable’s
name, ATTRIB, is not in quotation marks.
A portion of the HTML report is shown here. When the report is viewed in the browser, placing
the cursor over the clinic number reveals the clinic name.
Destination: HTML Style: DEFAULT
Chapter 8: Using PROC REPORT with ODS 295
The PDF report shows where tip text is available by including an icon in the cell with tip text.
Again, hovering the cursor over the cell reveals the tip text. In the most recent PDF readers, you
can double-click on the tip text icon to have the tip text expanded.
Destination: PDF Style: PRINTER
For the HTML destination you can change the cursor by using the HTMLSTYLE= attribute. This
attribute can alter, among other things, the shape of the cursor. To change the HTML cursor from
the default (a character similar to an uppercase I) to a hand, you can use the following compute
block:
compute clinnum;
attrib = 'style={flyover="'||trim(clinname)||
'" htmlstyle="cursor:hand"}';
call define(_col_,'style',attrib);
endcomp;
MORE INFORMATION
The construction of character variables that contain style attributes is further complicated when
macro variables are involved; see Section 11.1.2.
SEE ALSO
Downing (2004) discusses these techniques and shows an example for the HTML destination
using macro variables and CALL DEFINE.
8.8.2 Placing Tip Text Using STYLE=
The STYLE= option is used to set attributes for cells, rows, and columns. In this example, the
STYLE= option is used on the DEFINE statements. Because the component is not specified, the
DEFINE statement default components (HEADER and COLUMN) are assumed. Consequently
the header and each cell in the specified column will receive the same tip text.
. . . Portions of the code are not shown . . .
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height'
style={flyover='Measured in Inches'};
296 Carpenter’s Complete Guide to the SAS REPORT Procedure
define weight / analysis mean
format=6.2
'Weight'
style={flyover='Measured in Pounds'};
rbreak after / summarize;
run;
. . .Portions of the code are not shown . . .
We can see the tip text indicators in the PDF file.
Destination: PDF Style: PRINTER
As a general rule, when using the STYLE= option, I prefer to specify the component. This gives
me more flexibility and control. In the previous example, the tip text is a constant for the entire
column and really only needs to be associated with the header, not all the cells within the column.
We can see that this happens by specifying the HEADER component.
The previous example has been refined in the following code. The HEADER component causes
the style attributes (in this case the flyover text) to be applied only to the header.
. . . Portions of the code are not shown . . .
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'Height'
style(header)={flyover='Measured in Inches'};
define weight / analysis mean
format=6.2
'Weight'
style(header)={flyover='Measured in Pounds'};
rbreak after / summarize;
run;
ods _all_ close;
In the PDF report the tip text indicators are now only present on the headers.
Chapter 8: Using PROC REPORT with ODS 297
Destination: PDF Style: PRINTER
8.8.3 Placing Tip Text Using ~S={ }
In the first example in Section 8.8.2, the tip text was associated with each cell in the column. By
using the ~S={ } in-line formatting sequence, we can associate the tip text with any set of
characters, including only the header text.
. . . Portions of the code are not shown . . .
ods escapechar=~;
title1 "Ages 11 - 16";
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2
'~S={flyover="Measured in Inches"
htmlstyle="cursor:hand"}Height';
define weight / analysis mean
format=6.2
'~S={flyover="Measured in Pounds"
htmlstyle="cursor:hand"}Weight';
rbreak after / summarize;
run;
ods _all_ close;
Notice that the ~S={ } sequence appears within the text string that is to become the header for the
individual columns. Other attributes, such as color, font, and character size, could also have been
specified at the same time.
In SAS 9.2 you could also use the new ~{STYLE} nested in-line formatting sequence to insert tip
text.
MORE INFORMATION
The ~S={ } and ~{STYLE} in-line formatting sequences are discussed in more detail in Section
8.6.4. Tip text is created using a CALL DEFINE routine in Section 8.8.1 and using the STYLE=
option in Section 8.8.2.
298 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.9 Specifying Multiple Columns for RTF and PDF
When writing to the RTF and PDF destinations, you can use the COLUMNS= option on the ODS
statement to specify multiple columns on the output page. This option can be especially useful if
you have long narrow reports. In the following example, we are listing the names of the patients
within each clinic. The COLUMNS= option (n,o) is used to specify the number of columns that
are desired.
ods pdf style=printer
columns=3 n
file="&path\results\ch8_9a.pdf";
ods rtf style=rtf
columns=3 o
file="&path\results\ch8_9a.rtf"
bodytitle;
title1 "Name Lists by Clinic";
proc report data=rptdata.clinics nowd;
columns clinname clinnum ("Name" lname fname);
define
define
define
define
clinname/
clinnum /
lname
/
fname
/
display noprint;
order 'Clinic Number';
order 'Last';
order 'First';
compute clinnum;
attrib = 'style={flyover="'||trim(clinname)||'"}';
call define(_col_,'style',attrib); p
endcomp;
run;
ods _all_ close;
Three columns are specified for the report that goes to the PDF destination n, as well as for the
RTF destination o.
A portion of the RTF table is shown here; notice the placement of the title.
Destination: RTF Style: RTF
Chapter 8: Using PROC REPORT with ODS 299
The following is a portion of the PDF table. Notice that the tip text specified in the CALL
DEFINE statement p is only available in the PDF table.
Destination: PDF Style: PRINTER
You do have to be a bit careful when using this option. In the previous example, three columns fit
on the page rather nicely. However, if we were to increase our request to four columns, which will
not fit, the table degenerates and is no longer usable. The following table is an image of a portion
of the same RTF output as from the previous example, except with COLUMNS=4.
Destination: RTF Style: RTF
MORE INFORMATION
The PANELS= option can be used to produce a similar effect in the LISTING destination (see
Section 4.4.4).
SEE ALSO
Burlew (2005, pp. 238–240) creates a multi-panel RTF document using the COLUMNS= option.
300 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.10 Adding Text through the TEXT= Option
The TITLE and FOOTNOTE statements can add text to tables generated by SAS procedures, and
when using PROC REPORT, you also have the LINE statement to add text to the table. You can
also add text directly by using the TEXT= option on the ODS statement.
ods pdf file="&path\results\ch8_10.pdf"
startpage=no
text='Example 8.10';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
. . . Portions of the code are not shown . . .
In this example, the text is inserted into the table after any title lines. However, the TEXT= option
will, by default, cause the generation of a page. Since this is usually not what we want, the
STARTPAGE= option can be used to control paging. If STARTPAGE= NO is not specified, the
text appears on a page previous to the table.
Destination: PDF Style: PRINTER
It is anticipated that in SAS 9.2, the ODS TEXT statement will be available to add text to the
report. It is not yet clear how this statement will interact with the TEXT= option. Whether the text
is inserted before or after the title is destination-dependent.
SEE ALSO
Delaney (2003b) includes an example of the use of the TEXT= option.
Chapter 8: Using PROC REPORT with ODS 301
8.11 RIGHTMARGIN: Aligning Numbers When Using
CELLWIDTH
It is not unusual for the PROC REPORT table to seem crowded. This is especially a problem
when the numbers are large relative to the header text. There are a number of ways that we can
approach this problem. However, for non-LISTING destinations such as PDF, HTML, and RTF,
one of the more robust techniques is to increase the CELLWIDTH.
The following PROC REPORT step creates a simple PDF table of shoe sales information.
ods pdf style=printer
file="&path\results\ch8_11a.pdf"
notoc
startpage=never;
ods pdf text='~nDefault';
proc report data=sashelp.shoes nowd;
column region stores sales inventory returns;
define region
/ 'Region' group;
define stores
/ 'Number of Stores';
define sales
/ 'Total Sales';
define inventory / 'Total Inventory';
define returns
/ 'Total Returns';
run;
ods pdf close;
In the resulting table, the columns containing total sales, total inventory, and total returns seem
crowded.
Destination: PDF Style: PRINTER
302 Carpenter’s Complete Guide to the SAS REPORT Procedure
To alleviate the crowding, the CELLWIDTH attribute modifier can be used to specify a width for
the cells within a column. The previous table has been modified by increasing the cell width
(specified in millimeters).
ods pdf text='~nUsing CELLWIDTH, Numbers are too far to the Right';
proc report data=sashelp.shoes nowd;
column region stores sales inventory returns;
define region
/ 'Region'
group
style(column)={cellwidth=25mm};
define stores
/ 'Number of Stores'
style(column)={cellwidth=20mm};
define sales
/ 'Total Sales'
style(column)={cellwidth=35mm};
define inventory / 'Total Inventory'
style(column)={cellwidth=35mm};
define returns
/ 'Total Returns'
style(column)={cellwidth=35mm};
run;
Destination: PDF Style: PRINTER
The table is less crowded; however, because the columns of numbers are right-justified, they now
appear to be too far to the right. Centering the numbers will not help, because we want the
columns to be aligned. Instead, we can use the RIGHTMARGIN attribute modifier to increase the
space between the units digit and the right border of the cell.
Chapter 8: Using PROC REPORT with ODS 303
ods pdf text='~nUsing CELLWIDTH and RIGHTMARGIN, Looks Much Better';
proc report data=sashelp.shoes nowd;
column region stores sales inventory returns;
define region
/ 'Region'
group
style(column)={cellwidth=25mm};
define stores
/ 'Number of Stores'
style(column)={cellwidth=20mm
rightmargin=7mm};
define sales
/ 'Total Sales'
style(column)={cellwidth=35mm
rightmargin=7mm};
define inventory / 'Total Inventory'
style(column)={cellwidth=35mm
rightmargin=7mm};
define returns
/ 'Total Returns'
style(column)={cellwidth=35mm
rightmargin=7mm};
run;
The columns of numbers are now centered under their respective headers.
Destination: PDF Style: PRINTER
The selection of the appropriate size for CELLWIDTH and RIGHTMARGIN is somewhat
arbitrary. Nominally, at least for RTF and PDF, these values relate to the paper size. However, in
practice it is more likely that you will need to do some trial and error tests to come up with the
sizes that are right for you. You can use either inches (in) or millimeters (mm). Because they are
smaller, millimeters are often easier to work with.
304 Carpenter’s Complete Guide to the SAS REPORT Procedure
8.12 Chapter Exercises
Unless a specific ODS destination has been specified for a given exercise, answer each of the
questions using an ODS destination appropriate for your work. Ideally, you should experiment
with PDF, RTF, and a markup destination such as HTML.
The following exercises build on a basic program first created in Exercise 1 in Chapter 3. Here are
the pertinent portions of that program:
* E8_0Basic.sas
*
* Chapter 8 Exercise Basic program
*
* Total profit for each year within Product line.;
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Profit Summaries';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit;
define year
/ group;
define product_line
/ group
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress skip;
rbreak after / summarize;
run;
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports clothing
and equipment store. Using the BREAK and RBREAK statements, summarize across product
lines and years. You can build on the results of Exercise 1 in Chapter 3 or start with program
E8_0Basic.sas .
Using the STYLE= option, change attributes on the following:
column headers
row header
data values
something on a summary row
Experiment with different components on different PROC REPORT step statements.
If the same component / attribute combination appears with different attribute characteristics
on both the DEFINE statement and the PROC REPORT statement, which receives
precedence?
2. Repeat Exercise 1 in this section using the CALL DEFINE statement. Is the order that the
attributes are applied the same? Which portions of the report are not controllable through the
CALL DEFINE statement?
Chapter 8: Using PROC REPORT with ODS 305
3. In the basic table from Exercise 1 in this section, add trafficlighting to alert the reader to
profits that are less than $1 million. Solve using both the STYLE= option and the CALL
DEFINE statement.
4. In the basic table, create links for the ‘Sports’ product line that drill down to a detail table that
shows PRODUCT_CATEGORY within PRODUCT_LINE.
Extra Credit:
Make the links and detail tables year-specific.
5. Using the E8_0Basic.sas program, use three or more title statement options to change text
attributes.
306 Carpenter’s Complete Guide to the SAS REPORT Procedure
P a r t
3
Extending PROC REPORT
Chapter 9
Reporting Specifics for ODS Destinations
Chapter 10
Solving Other Common Report Problems
Chapter 11
Details of the PROC REPORT Process
309
325
367
308 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
9
Reporting Specifics for ODS Destinations
9.1 RTF 310
9.1.1 Using the BODYTITLE Option 311
9.1.2 Adding RTF Control Words 312
9.1.3 Post-processing of RTF Files 313
9.2 PDF 314
9.2.1 Adding PDF File Descriptors 314
9.2.2 Setting the Default Margins 315
9.3 HTML and Other Markup Destinations 316
9.3.1 Exporting a Report to Microsoft Excel 316
9.3.2 Setting Tagset Attributes 322
9.3.3 HTML Tags and Repeat Characters 323
With the ever expanding list of both ODS destinations and capabilities within a destination, the
topic of this chapter is itself a subject worthy of its own book. Unfortunately, a full treatment of
the topic of the Output Delivery System is not possible within the context of PROC REPORT.
Consequently, this chapter will only deal with a few of the ODS options and destinations that
directly affect the user of this procedure. Even in this narrower range, the topics are extensive and
only a limited number will be included.
Because the underlying technologies for information delivery are evolving very quickly, the
Output Delivery System is constantly undergoing modifications and improvements at SAS. It is
reasonable, therefore, to expect that new capabilities and options will have been added during and
after the publication of this book.
310 Carpenter’s Complete Guide to the SAS REPORT Procedure
A great deal of the content of this chapter deals with report appearance characteristics. Many of
these appearance attributes can also be controlled through the use of ODS styles, which is well
outside the scope of this book. The reader is encouraged to learn more about PROC TEMPLATE
and styles in general by examining the extensive offerings on the subject. The books by Haworth
(2001a) and Gupta (2003) on ODS, as well as a series of papers on the topic serve as excellent
starting points for this learning process.
MORE INFORMATION
Chapter 8, “Using PROC REPORT with ODS,” deals with ODS-specific issues, including the use
of in-line formatting sequences.
SEE ALSO
Haworth (2001b) provides an overview of ODS relative to PROC REPORT. DelGobbo (2005)
discusses most of the destinations covered in this chapter, and he includes corrections to a number
of styles associated with reporting. The first of a series of papers on building and modifying styles
is presented by Haworth (2002).
Two books on ODS include the comprehensive Output Delivery System: The Basics by Lauren
Haworth (2001a) and the shorter Quick Results with the Output Delivery System by Sunil Gupta
(2003).
Various techniques that allow the production of reports that are usable by persons with disabilities
are discussed by Harper and Pappas (2005).
9.1 RTF
One of the more commonly used destinations for PROC REPORT is the Rich Text Format (RTF)
destination. Unfortunately, because of its very nature, this destination presents a number of unique
challenges.
When ODS writes to the RTF destination, control of the final rendering of the table does not
actually take place until the RTF file is incorporated into the final document. This means that SAS
cannot control or even fully anticipate basic things such as margins, page counts, or cell widths.
Much of the literature referenced in this section deals with these issues.
An RTF file contains special embedded formatting instructions. These instructions are then
interpreted by whichever program (Microsoft Word, WordPerfect, or others) ultimately renders
the file. Fortunately, ODS builds the RTF file for us automatically, and consequently we are not
required to know its internal structure. The current versions of SAS optimize the file for Microsoft
Word 2002.
SEE ALSO
Hull (2001) and Cochran (2005) introduce the topic of the RTF destination. Although part of an
introductory overview, most of the examples by Shah (2003) relate to RTF issues. Hamilton
(2003) and Sevick (2006) both demonstrate a number of great RTF techniques. Haworth (2004)
creates an ODS style for use with reports generated for the RTF destination. Hester (2006)
discusses the RTF tagset and how it can be modified and enhanced. Collins and Hopkins (2005)
Chapter 9: Reporting Specifics for ODS Destinations 311
contrast RTF with XML. Feder (2004) and Shannon (2002) both discuss RTF characteristics, style
changes, and the use of the STYLE= option. Yeh (2004) shows an ODS RTF bookmark
statement. Zhang and Li (2006) create a vertically concatenated RTF table.
9.1.1 Using the BODYTITLE Option
By default, when a table is generated using the RTF destination, the titles and footnotes are placed
in the document's header and footer sections respectively. This allows the titles and footnotes to
be placed on each page when the table spans multiple pages.
Regardless of the size of the table, the footer appears at the bottom of the page, and this can be a
problem when the table is either small (only part of a page) or the table is to be incorporated into
another document.
You can change this behavior by adding the BODYTITLE option to the ODS RTF statement.
This option places the titles and footnotes as a part of the table itself.
ods rtf file="&path\results\ch9_1_1b.rtf"
bodytitle;
title1 "Ages 11 - 16";
footnote1 'Height in Inches';
footnote2 'Weight in Pounds';
proc report data=sashelp.class nowd;
columns sex height weight;
define sex
/ group;
define height / analysis mean
format=5.2;
define weight / analysis mean
format=6.2;
rbreak after / summarize;
run;
ods rtf close;
Here is the resulting table:
Without first using the BODYTITLE option on the
ODS statement, the titles and footnotes would have
been placed in the header and footer sections of the
RTF table.
Destination: RTF Style: RTF (default)
312 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
The BODYTITLE option is used extensively in the RTF examples in Chapter 8.
SEE ALSO
Squire and Tai (2005) show how to control various aspects of the report, including footnote
placement, that depend on line counting and page size when the BODYTITLE option is used.
9.1.2 Adding RTF Control Words
The Rich Text Format file was originally designed to be imported into Microsoft Word. The RTF
file form has evolved somewhat in recent years, and each of the releases of SAS have adjusted to
create optimized code for the current versions of Word. Although SAS 9.x creates RTF code that
is optimized for Word 2002, other applications, including other versions of Word, will still, for
the most part, be able to utilize the file.
Because an RTF file is rendered by the application that receives it, at least a basic understanding
of the RTF file command language and structure can be helpful. As the ODS RTF destination
matures, SAS has continued to move functionality into ODS options, making this knowledge less
and less important. For now, however, RTF knowledge can be helpful.
RTF control words, which are preceded by a backslash (\), can be used to pass instructions
directly to RTF.
* Create a style that turns off the protection for
* special characters (like the backslash);
proc template;
define style styles.test; n
parent=styles.rtf;
style systemtitle from systemtitle / o
protectspecialchars=off;
end;
run;
ods rtf style=styles.test
file="&path\results\ch9_1_2.rtf"
bodytitle;
title1 '\b \highlight3 Product Summary'; Z
footnote1;
. . . Portions of the code are not shown . . .
n A new style (TEST) is created, based on the RTF style.
o The PROTECTSPECIALCHARS attribute is turned off for the titles only.
Z RTF commands are used on the title to bold (\B) and change background color
(\HIGHLIGHT3). These commands are then passed directly to RTF for interpretation.
Chapter 9: Reporting Specifics for ODS Destinations 313
Destination: RTF Style: User-defined
The technique shown here requires the use of PROC TEMPLATE to alter an ODS style, and,
because we use RTF control words directly, it only works for the RTF destination. The techniques
discussed in Section 8.6.6 allow RTF codes to be passed without the modification of the ODS
style. In that same section, specific RTF control words are also passed with the use of in-line
commands.
SEE ALSO
Gianneschi (2006) and Parsons (2006) both provide several examples and tables of RTF control
words. Haworth (2005a) discusses the use of RTF escape sequence characters and PROC
TEMPLATE to write tables into Microsoft Word. Smoak (2004) provides examples of a number
of simple techniques that can be used with RTF tables that are specifically to be moved into
Word.
Osowski and Fritchey (2006) discuss the use of RTF control words in general terms.
FAQ #4007 discusses which versions of Word are supported for RTF files generated by different
versions of SAS. More on RTF control words can be found at
http://support.sas.com/rnd/base/topics/templateFAQ/Template_rtf.html#control
9.1.3 Post-processing of RTF Files
Once ODS RTF has created the RTF file, the file itself can be included in Word or whatever final
application will receive it. However, because the file is technically just a text file, it can be further
manipulated prior to being imported. Any text editor can be used; however, within SAS, the
DATA step itself can be used to find and replace unwanted codes. These techniques require an
understanding of what RTF codes need to be removed or modified, but are otherwise fairly
straightforward.
Although they are of general programming interest, these techniques are only marginally
associated with PROC REPORT and consequently are not discussed further here. The SEE ALSO
314 Carpenter’s Complete Guide to the SAS REPORT Procedure
section includes references to papers on the topic. Fortunately, as the capabilities within ODS
RTF continue to improve, techniques such as this will become less necessary.
SEE ALSO
Zhang (2005) and Zhang and Li (2006) use a DATA step to remove unwanted RTF codes
resulting in combined tables.
Qi and Zhang (2003) discuss a macro that enhances an RTF table. Although they do not give
solution details, they do suggest a number of issues that are encountered when writing to RTF.
Parker (2005) discusses the concepts, but not the details, of creating a macro that post-processes
an RTF file after it is produced by ODS. Using some of the same concepts, Hagendoorn, Squire,
and Tai (2006) use SAS DATA steps to read RTF reports as part of a QC process.
9.2 PDF
Most of the items discussed in this section have more to do with the PDF destination than they do
with PROC REPORT. However, since many users of PROC REPORT generate reports using the
ODS PDF destination, some of these capabilities are briefly introduced in this section.
SEE ALSO
The creation of a customized PDF style is discussed by Haworth (2005b).
A number of techniques and options geared towards the PDF destination are presented by
Thornton (2006).
Although not directly written about PROC REPORT, the two papers by Kevin Delaney (2003a
and 2003b), both serve as nice introductions to a number of options available on the ODS PDF
statement. A number of PDF destination options are also demonstrated by Parsons (2005).
Lund (2005) has a number of advanced reporting examples for the PDF destination, and Hamilton
(2004) includes examples of PDF in-line formatting characters.
Okerson (2004) uses ODS LAYOUT with PDF to create a hospital scorecard.
9.2.1 Adding PDF File Descriptors
When a PDF file is generated, its metadata can include information about the file itself. This
information can be added interactively through Adobe Acrobat, or, when the file is created with
ODS, through a series of options on the ODS PDF statement. These options include the following:
TITLE=
AUTHOR=
SUBJECT=
KEYWORDS=
specifies the document title.
specifies the author of the document or table.
specifies the topic of the table.
specifies searchable keywords for this table.
Chapter 9: Reporting Specifics for ODS Destinations 315
These options have been included in the following ODS PDF statement:
ods pdf file="&path\results\ch9_2_1.pdf"
title='Example 9.2.1'
author='Art Carpenter'
subject='PDF Property Items'
keywords='PDF s9_2_1.sas properties';
title1 'Sales Summary';
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
. . . Portions of the code are not shown . . .
The PDF document’s properties can be viewed by selecting File X Document Properties
(CTRL+D). Interestingly enough, the properties show that the document was created by SAS and
the date/time of creation is the start of the SAS session rather than the actual date/time of creation.
SEE ALSO
Parsons (2005) includes an example that uses these options.
9.2.2 Setting the Default Margins
In the PDF destination, the margins are generally allowed to default to those of the receiving
document. However, SAS system options can be used to override the defaults. In the following
example, the left margin and the top margin have been set to 5 inches and 3 inches respectively
for both the PDF and RTF destinations.
316 Carpenter’s Complete Guide to the SAS REPORT Procedure
ods listing close;
option leftmargin="5 in" topmargin="3 in";
ods pdf file="&path\results\ch9_2_2.pdf";
ods rtf file="&path\results\ch9_2_2.rtf"
bodytitle;
. . . Portions of the code are not shown . . .
Inspection of the PDF file shows that these margin definitions were respected, while an
examination of the RTF output shows that the margin settings were not utilized. As a general rule,
these margins are more successfully applied by the PDF destination as well.
MORE INFORMATION
If you are adjusting the margins so that the table appears at a specific location on the page, the
ODS LAYOUT statement may be an easier, more flexible, way to place the table (see Section
10.5.1).
9.3 HTML and Other Markup Destinations
HTML is one of the original class of ODS destinations that are known collectively as “markup”
destinations. This group of destinations has some unique capabilities, including the ability to use
tagsets and cascading style sheets (CSS). Although the bulk of these topics are clearly outside of
the scope of this book, many of the capabilities of these destinations can be very helpful to
programmers who need to have a report rendered by one of the markup languages, such as
HTML.
This section touches on a very few of those capabilities.
SEE ALSO
A nice introduction to the MARKUP destination is presented by Gebhart (2006). Lafler (2005)
introduces the use of the HTML destination, and Gebhart (2005) gives a strong overview of
markup destinations with specific examples. Pass and McNeil (2003) give a series of fairly
advanced HTML examples.
Pagé (2004) uses HTML, REPORT and ODS to directly e-mail a report.
9.3.1 Exporting a Report to Microsoft Excel
Although we like to think that we can do just about anything in PROC REPORT, at times it is
necessary to export a report table to a Microsoft Excel spreadsheet. Fortunately, SAS provides us
with several alternative approaches to the process of creating a spreadsheet.
This section describes three techniques that convert a report to an Excel spreadsheet. These
techniques do not produce the same report, so it is best to consider the strengths and weaknesses
of each technique. None of these techniques create a “pure” Excel file. In each case, after you
open the file in Excel you need to select Save As and change to an Excel worksheet to get a native
or “pure” Excel file.
Chapter 9: Reporting Specifics for ODS Destinations 317
Using the HTML Destination with the XLS Extension
In SAS 8, HTML was the only ODS destination that could be used to transfer a report to Excel. In
®
SAS 9 it can still be used; however, it is no longer either the only or even necessarily the best
®
technique. If you have legacy code that uses the HTML destination and you are using SAS 9, you
might want to convert it to one of the other more versatile techniques described in this section.
When using the HTML destination to transfer a report to Excel, you simply use the HTML
destination and create the resultant HTML file with an extension of .XLS. Although the file will
not be a pure Excel file, it will be opened by Excel as long as XLS is a registered extension of
Excel.
The following table was first created in Section 7.2.4. Here we use ODS HTML to write the
resulting table to a file with the XLS extension.
proc format;
value $regname
'1','2','3' = 'No. East'
'4'
= 'So. East'
'5' - '8'
= 'Mid West'
'9', '10'
= 'Western';
value $gender
'M' = 'Male'
'F' = 'Female';
run;
ods html style=default n
path="&path\results"
body="ch9_3_1a.xls";
o
title1 'ODS Destination Specifics';
title2 "Writing a Table to Excel";
proc report data=rptdata.clinics
nowd
split='*';
column region ('Mean Weight*in Pounds' sex,wt ratio);
define region / group width=10 'Region'
format=$regname. order=formatted;
define sex
/ across 'Gender'
format=$gender. order=data;
define wt
/ analysis mean format=6. ' ';
define ratio
/ computed format=6.3 'Ratio*F/M ';
rbreak after
/ dol skip summarize;
compute ratio;
ratio = _c3_ / _c2_;
endcomp;
run;
ods _all_ close;
n The DEFAULT style has been selected. When writing to the spreadsheet, background colors
are changed.
o The file is created with an XLS extension. When opened in Excel, the file appears as follows:
318 Carpenter’s Complete Guide to the SAS REPORT Procedure
Destination: HTML Style: DEFAULT (with an XLS file extension)
Notice that the first column (A) has been widened to accommodate the titles. This can be
controlled somewhat through the use of the COLSPAN= option on an HTML tag such as the table
header tag (/TH) or the table data tag (/TD), both of which can be embedded in the TITLE
statement, as shown in these examples:
title1 'ODS Destination Specifics ';
title2 "Writing a Table to Excel ";
Here is the resulting Excel table:
Destination: HTML3 Style: DEFAULT (with an XLS file extension)
Chapter 9: Reporting Specifics for ODS Destinations 319
For the previous two spreadsheets, notice that although the blue foreground color of the
DEFAULT style has been transferred to the spreadsheet, none of the background colors
associated with the DEFAULT style have been transferred. This is an artifact of how the HTML
destination has been implemented in SAS 9.1. In this version of SAS, the destination definition
includes the use of cascading style sheets (CSS). Although this is advantageous for ODS users
who want to manipulate the attributes of the destination, Excel does not know how to correctly
interpret all of these attributes, and some characteristics are lost.
In SAS 8.2, the HTML destination writes to the HTML 3.2 standard (SAS 9.1 writes to the
HTML 4.0 standard). Because the earlier HTML destination was constructed differently, it
handled the attributes differently. In SAS 9.1 we can still access the older version (the SAS 8.2
version) of the HTML destination by using the destination name HTML3.
ods html3 style=default
path="&path\results"
body="ch9_3_1c.xls";
title1 'ODS Destination Specifics';
title2 "Writing a Table to Excel Using HTML3";
. . . Portions of the code are not shown . . .
Destination: HTML3 Style: DEFAULT (with an XLS file extension)
320 Carpenter’s Complete Guide to the SAS REPORT Procedure
If you are using a version of Excel prior to Microsoft Office 2000, then HTML3 should probably
be your preferred destination for moving reports into Excel. For later versions of Excel, and if you
are also running SAS 9.1 and later, you should consider the MSOFFICE2K and EXCELXP
tagsets, which are discussed in the following subsection. Essentially, for SAS 9.1 and later, the
HTML destination is designed to produce output for the browser, and is not properly tuned for
writing to spreadsheets.
SEE ALSO
Feng (2006) uses this technique to generate an Excel table with trafficlighting.
Using the MSOFFICE2K Tagset
The MSOFFICE2K tagset has been designed specifically for transferring information and tables
into the Microsoft Office 2000 environment. To take advantage of this tagset, we replace the
name of the ODS destination in the previous example with this tagset name. Here are the modified
statements:
ods msoffice2k style=default
path="&path\results"
body="ch9_3_1d.xls";
title1 'ODS Destination Specifics';
title2 "Using the MSOFFICE2k Tagset";
. . . Portions of the code are not shown . . . .
Notice that the HTML tags in the titles have been removed, and the Excel table changes as
follows:
Destination: MSOFFICE2K Style: DEFAULT (with an XLS file extension)
Chapter 9: Reporting Specifics for ODS Destinations 321
The MSOFFICE2K tagset supports importing SAS/GRAPH images into Excel, and, like the
HTML destination, produces an HTML file that can be read by Excel.
Using the EXCELXP Tagset
At the same time as this book was being written, a great deal of effort was also going into the
further development of the EXCELXP tagset. This tagset is being tailored for users with
Microsoft Office 2002 and higher who specifically need to write to Excel tables. This tagset
emphasizes the data rather than the text, and it does a much better job of placing numbers into
numeric columns. It also allows you to generate multiple worksheets per workbook. It does all of
this by automatically generating an XML file that is then passed to Excel.
ods markup tagset=excelxp style=default
path="&path\results"
body="ch9_3_1e.xls";
title1 'ODS Destination Specifics';
title2 "Using the ExcelXP Tagset";
. . . Portions of the code are not shown . . .
Destination: markup (with EXCELXP tagset) Style: DEFAULT (with an XLS file extension)
Notice that, at least in SAS 9.1, there is no alias for this tagset. Consequently we need to either
use the MARKUP destination and the TAGSET= option shown previously, or the following dot
notation:
ods tagsets.excelxp style=default
path="&path\results"
body="ch9_3_1f.xls";
The current version of this tagset has the following capabilities and limitations:
It supports multiple worksheets within an Excel table. When multiple REPORT tables
are specified within the ODS block, each is written to its own worksheet.
It does not work for SAS/GRAPH images (future versions may).
It automatically generates XML to transfer the table information to Excel.
322 Carpenter’s Complete Guide to the SAS REPORT Procedure
Other supported ODS statement options for this destination include the following:
DOC =
ORIENTATION =
EMBEDDED_TITLES =
FROZEN_HEADERS
'help' or 'quick' (no default value)
Lists options in the log.
'portrait' or 'landscape'
Specifies either portrait or landscape orientation.
'no' or 'yes'
Specifies whether or not titles should be included.
'no' or 'yes'
Header text can be frozen when scrolling long tables.
To get a current list of options for this tagset, add the following option:
ODS tagsets.excelxp options(doc="help")...;
Inspection of the SAS log will show the options and their default values.
SEE ALSO
A very detailed and readable paper by DelGobbo (2006) highlights a number of options and
techniques that can be used with the EXCELXP tagset.
Parker (2003) gives several detailed examples of a number of techniques that ODS can use to
modify the appearance and functionality of reports that are written to Excel tables.
Godard and Williamson (2004) discuss the advantages and disadvantages of a number of different
techniques for writing information from SAS to Excel. Feder (2005) discusses the use of PROC
REPORT to create XLS files and includes examples of passing formulas to Excel as well as the
use of DDE libraries. Brown (2005) presents a macro that uses tagsets to create multi-sheet Excel
spreadsheets (although the demonstrated techniques have become somewhat superceded by
EXCELXP). Frey (2005) discusses several techniques, including the use of ODS, for writing to
Excel spreadsheets.
Updated MARKUP destination tagsets can be downloaded from
http://support.sas.com/rnd/base/topics/odsmarkup/#download
9.3.2 Setting Tagset Attributes
One of the advantages of using markup language destinations is the ability to create and modify
tagsets. This gives us another layer of control. You can, of course, create your own tagsets by
using PROC TEMPLATE, or you can alter individual attributes by using the TAGATTR= style
attribute modifier on the STYLE= option.
In the following rather silly example, the default behavior of wrapping long text strings is altered
for one of the two columns. As a general rule, it is an advantage to have long text strings
automatically wrap. However, by changing the wrapping attribute we can, when it is necessary,
force the text to not wrap.
data one;
longtext1="this is a very long text string that will wrap if you do
not use the Nowrap attribute.";
longtext2=longtext1;
run;
Chapter 9: Reporting Specifics for ODS Destinations 323
title1 'ODS Destination Specifics';
title2 'Using NOWRAP on LONGTEXT1';
ods msoffice2k file="&path\results\ch9_3_2.html";
proc report data=one nowd;
column longtext1 longtext2;
define longtext1 / display
style(column)={tagattr="nowrap"};
define longtext2 / display;
run;
ods msoffice2k close;
Destination: HTML Style: DEFAULT
MORE INFORMATION
Controlling the wrapping of text using in-line escape character sequences is discussed in
Section 8.6.5, and for the LISTING destination using the FLOW option in Section 4.5.2.
9.3.3 HTML Tags and Repeat Characters
In monospace destinations, such as LISTING, repeat characters can be added to text that is to be
used on spanning headers (see Section 4.3.1). In the LISTING destination, the following
COLUMN statement:
column region sex ('_ wt(lb) _' wt,(n mean));
produces an output report with the following spanning header.
______ wt(lb) ______
In other non-monospaced destinations, such as HTML, these repeat characters are ignored (as
repeat characters). The HTML header appears as follows:
It is anticipated that starting in SAS 9.2 these repeat characters will be stripped off altogether.
However, an exception will be made for the repeat character pair < >. These characters could also
be used to specify HTML tags. Consequently PROC REPORT does not strip the leading < and
trailing > in markup destinations. This can cause some “interesting” titles when writing to both
324 Carpenter’s Complete Guide to the SAS REPORT Procedure
the LISTING and HTML destinations. In the following COLUMN statement, header text is
assigned an attribute of italics using an HTML tag.
column region sex (" Weight " wt,(n mean));
Here are portions of the resulting LISTING and HTML destinations:
<<< Weight >>>
As an aside, notice that the tag is closed with rather than the more traditional . In markup
destination examples throughout this book, either form will work. However, in PROC REPORT
the slash is the default character used to form splits. Using the COLUMN statement
column region sex (" Weight " wt,(n mean));
will negate repeated text in the LISTING destination as well as prevent the tag assignment in the
HTML destination.
If you need to include the slash when closing the tag, you can change the split character in the
PROC REPORT statement. The following will now work as you might anticipate.
proc report data=rptdata.clinics nowd split='|';
column region sex (" Weight " wt,(n mean));
. . . Portions of the code are not shown . . .
Here are the resulting reports for the LISTING and HTML destinations (in part):
<< Weight >>>
C h a p t e r
10
Solving Other Common Report Problems
10.1 Creating Vertically Concatenated Tables 326
10.1.1 A Simple Table 326
10.1.2 Ordering the Generated Classifications 332
10.1.3 Text and Number Alignment in Derived Columns 335
10.1.4 Doing More in the PROC REPORT Step 338
10.2 Automating the PROC REPORT Process 341
10.2.1 Things to Think about When Automating 342
10.2.2 Macro Variable Resolution Issues 343
10.3 Coordinating Graphics with PROC REPORT 345
10.3.1 Using CALL DEFINE to Import Graphics 345
10.3.2 Using GPRINT and GREPLAY 352
10.3.3 Using the Annotate Facility to Generate Lines 355
10.4 Workarounds for Monospace-Only Options 357
10.5 Generating Separate Reports on the Same Page 360
10.5.1 ODS LAYOUT 360
10.5.2 HTML Reports 362
10.5.3 RTF and PDF Reports: Using STARTPAGE=NEVER 363
10.5.4 Aligning Columns across Reports 365
It does not take very long as a PROC REPORT programmer before “special” reports are
requested. Often these require some “special” techniques that are not immediately obvious. This
chapter presents and discusses some of these techniques. Many others have been published in the
326 Carpenter’s Complete Guide to the SAS REPORT Procedure
SAS user group literature. Although many of these other techniques are just too “special” to
include in this book, they may still have interest to advanced and specialty users of PROC
REPORT.
10.1 Creating Vertically Concatenated Tables
PROC REPORT does an excellent job of concatenating columns horizontally. All we have to do
is list the variable names in the COLUMN statement. We even have the ACROSS define usage to
make transposing easier. It is not so easy when we want to concatenate variables or groups
vertically. The primary problem with creating vertically concatenated tables with PROC REPORT
is that it is basically the wrong tool. PROC TABULATE creates these types of tables fairly easily.
However, because PROC TABULATE lacks the compute block, it does not have the flexibility to
transform variables or to create composite values.
Consider the following table, which displays counts and percentages for patient gender and
various statistics for height and weight for two regions.
Common REPORT Problems
Vertically Concatenated Tables
Gender
All
Female
Male
Eastern
38 (48%)
15 (19%)
23 (29%)
Western
42 (53%)
17 (21%)
25 (31%)
Overall
80 (100%)
32 (40%)
48 (60%)
Height
MEAN (STD)
MEDIAN
N
67.9 (3.91)
67.00
38.00
67.0 (3.04)
68.00
42.00
67.5 (3.49)
67.00
80.00
Weight
MEAN (STD)
MEDIAN
N
146 (34.96)
155.00
38.00
176 (30.51)
177.00
42.00
162 (35.87)
161.00
80.00
One of the issues with constructing a report like this is that we have to do a lot of work outside of
PROC REPORT. Fortunately, with the power of SAS at our finger tips, we can easily do the
necessary data manipulations. Still, this is a book on PROC REPORT, not the other procedures,
so naturally it is less exciting when we cannot do it all in PROC REPORT.
SEE ALSO
DeAngelis (2005) discusses a number of issues that come up when building this type of table.
Burlew (2005, pp. 152–162) uses the DATA step to build a similar table.
10.1.1 A Simple Table
Although the code to generate the table shown in the previous section is not too complex (about
100 lines), it does require a number of distinct steps.
Chapter 10: Solving Other Common Report Problems 327
Depending on what actually has to be done, the order of the steps leading up to the PROC
REPORT step varies. However, these types of reports typically have the primary steps that are
shown here:
n proc format;
o proc summary data=rptdata.clinics;
Z proc transpose data=clinsum out=tran1;
q data prep1(keep=class subclass regname value);
set tran1(Keep=_type_ region sex _name_ col1);
r proc sort data=prep1;
s proc transpose data=prep1 out=tran2;
t proc report data=tran2 nowd;
n Formats that will be used in the report are created.
o The data is summarized. Normally we would let PROC REPORT do the row and column
summaries; however, here we need to do them first so that they will be available in the DATA
step.
Z The summarized data is transposed so that it can be more easily handled in the DATA step.
q The classification variables, which form the concatenated values, are formed. The
composite values, such as COUNT (%) and MEAN (STD) are also created in this step.
r Sorting occurs before the transpose.
s The regions are transposed to form columns.
t Finally, the PROC REPORT step occurs.
In more detail, these steps form the processing steps of the program.
n User-defined formats might not be required, or they might already be available in established
libraries. The process of creating these formats is discussed by Carpenter (2004b). For this
example, the following step creates two formats for later use.
proc format;
value $EWReg
'1','2','3','4' = 'Eastern'
'5'-'9','10'
= 'Western'
other
= 'Overall';
value $gender
'F'
= 'Female'
'M'
= 'Male'
other = 'All';
run;
o The data is summarized using PROC SUMMARY. It is here that the statistics of interest are
specified. The fomat $EWREG. has been applied to REGION to collapse regions into these
area designations.
328 Carpenter’s Complete Guide to the SAS REPORT Procedure
proc summary data=rptdata.clinics;
class region /order=internal;
class sex;
var ht wt;
output out=clinsum
n= htn wtn
mean=htmean wtmean
median=htmedian wtmedian
std=htstd wtstd;
format region $ewreg.;
run;
The resulting data set (WORK.CLINSUM) has one row for each distinct summarization and a
column for each of the requested statistics. Here are the first three lines of the table:
Obs region sex _TYPE_ _FREQ_ htn wtn htmean
1 Overall
2 Overall F
3 Overall M
0
1
1
80
32
48
80
32
48
80
32
48
wtmean htmedian wtmedian htstd
67.4500 161.775
65.0000 145.875
69.0833 172.375
67
64
69
161
155
177
wtstd
3.49285 35.8721
2.63965 32.4154
3.01650 34.3948
p The transpose could be accomplished in the DATA step, but the resulting code is a bit more
complex. The following PROC TRANSPOSE step minimizes the number of columns that we
need to deal with in the DATA step. The BY statement takes advantage of the sort order of the
data that is generated by PROC SUMMARY.
proc transpose data=clinsum
out=tran1;
by _type_ region sex;
var _freq_ ht: wt:;
run;
The table WORK.TRAN1 (a portion of which is shown here) is now much more narrow, and is
easier to work with in the DATA step.
Obs
1
2
3
4
5
6
7
8
9
10
11
12
_TYPE_
0
0
0
0
0
0
0
0
0
1
1
1
region
Overall
Overall
Overall
Overall
Overall
Overall
Overall
Overall
Overall
Overall
Overall
Overall
sex
F
F
F
_NAME_
_FREQ_
htn
htmean
htmedian
htstd
wtn
wtmean
wtmedian
wtstd
_FREQ_
htn
htmean
COL1
80.000
80.000
67.450
67.000
3.493
80.000
161.775
161.000
35.872
32.000
32.000
65.00
Chapter 10: Solving Other Common Report Problems 329
q This table has all that we need to construct the various parts of the final table. The special
variable _TYPE_ is constructed by SUMMARY so that we can differentiate between the
various levels of summarizations. Summaries for REGION and SEX have _TYPE_=3,
whereas a summary across both REGION and SEX has _TYPE_=0. Here is the resulting
DATA step:
data prep1(keep=class subclass regname value);
set tran1(Keep=_type_ region sex _name_ col1);
length class subclass value $12 regname $7;
* Retain MEAN and N (denominator);
retain holdmean holdn .;
* Save the region name;
regname = put(region,$ewreg.);
* Build the stats for SEX;
if _name_='_FREQ_' then do;
* Only use observation counts (_FREQ_);
class = 'Gender';
subclass = put(sex,$gender.);
* Retain the denominator for future use;
if _type_ = 0 then holdn = col1;
* Value is count (%);
value = trim(left(put(col1,4.)))
||' ('
||trim(left(put(col1/holdn,percent.)))
||')';
output;
end;
* Build the stats for HEIGHT and WEIGHT;
* Ignore observations that contain _FREQ_;
else if _type_ in(0,2) then do;
if _name_ =: 'ht' then class='Height';
else class='Weight';
* Retain the MEAN for future use;
if index(_name_,'mean') then holdmean=col1;
else do;
if index(_name_,'std') then do;
subclass='MEAN (STD)';
* Value is a combination of MEAN and STD;
value = trim(left(put(holdmean,4.1)))
||' ('
||trim(left(put(col1,5.2)))
||')';
end;
else do;
* Other statistics;
subclass = upcase(substr(_name_,3));
value = put(col1,6.2);
end;
output;
end;
end;
run;
330 Carpenter’s Complete Guide to the SAS REPORT Procedure
The WORK.PREP1 data table is now very close to what we want for the final table. The character
variable VALUE contains the final display value. Here are the first few observations of
WORK.PREP1:
Obs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
class
subclass
value
regname
Gender
Gender
Gender
Gender
Gender
Gender
Gender
Gender
Gender
Height
Height
Height
Height
Height
All
All
All
Female
Female
Female
Male
Male
Male
MEAN (STD)
MEAN (STD)
MEAN (STD)
MEDIAN
MEDIAN
38 (48%)
80 (100%)
42 (53%)
15 (19%)
32 (40%)
17 (21%)
23 (29%)
48 (60%)
25 (31%)
67.9 (3.91)
67.5 (3.49)
67.0 (3.04)
67.00
67.00
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Overall
If we could use REGNAME as an ACROSS variable in PROC REPORT, we could now let PROC
REPORT take over for us. Unfortunately, you have to nest a statistic or numeric value under an
ACROSS variable, so a PROC REPORT step such as the following would not work (the report
item VALUE is character and has a define usage of DISPLAY):
proc report ......;
column class subclass regname,value;
define class / group;
define subclass / group;
define regname / across;
define value / display;
rsInstead we need to transpose the regions ourselves rather than in the PROC REPORT step.
Here is the resulting transpose step:
proc sort data=prep1;
by class subclass regname;
run;
proc transpose data=prep1 out=tran2;
by class subclass;
id regname;
var value;
run;
Chapter 10: Solving Other Common Report Problems 331
The ID statement results in a column for each value of REGNAME. Here is a portion of
WORK.TRAN2:
Obs class
1
2
3
4
5
6
7
8
9
Gender
Gender
Gender
Height
Height
Height
Weight
Weight
Weight
subclass
_NAME_
Eastern
Overall
Western
All
Female
Male
MEAN (STD)
MEDIAN
N
MEAN (STD)
MEDIAN
N
value
value
value
value
value
value
value
value
value
38 (48%)
15 (19%)
23 (29%)
67.9 (3.91)
67.00
38.00
146 (34.96)
155.00
38.00
80 (100%)
32 (40%)
48 (60%)
67.5 (3.49)
67.00
80.00
162 (35.87)
161.00
80.00
42 (53%)
17 (21%)
25 (31%)
67.0 (3.04)
68.00
42.00
176 (30.51)
177.00
42.00
t We can now use PROC REPORT to add the finishing touches. Because all the heavy lifting
has already been done, the PROC REPORT step itself is simple.
title1 'Common REPORT Problems';
title2 'Vertically Concatenated Tables';
proc report data=tran2 nowd;
column class subclass eastern western overall;
define class / group ' ';
define subclass / group ' ';
define eastern / display;
define western / display;
define overall / display;
compute after class;
line ' ';
endcomp;
run;
Here is the final table:
Common REPORT Problems
Vertically Concatenated Tables
Gender
All
Female
Male
Eastern
38 (48%)
15 (19%)
23 (29%)
Western
42 (53%)
17 (21%)
25 (31%)
Overall
80 (100%)
32 (40%)
48 (60%)
Height
MEAN (STD)
MEDIAN
N
67.9 (3.91)
67.00
38.00
67.0 (3.04)
68.00
42.00
67.5 (3.49)
67.00
80.00
Weight
MEAN (STD)
MEDIAN
N
146 (34.96)
155.00
38.00
176 (30.51)
177.00
42.00
162 (35.87)
161.00
80.00
332 Carpenter’s Complete Guide to the SAS REPORT Procedure
10.1.2 Ordering the Generated Classifications
In the previous table, the order of the values of the grouping variable CLASS is alphabetical, with
GENDER first. If we had wanted the classification value to be SEX, it would have sorted between
HEIGHT and WEIGHT, and this would almost certainly have been unacceptable. The order of the
second column, SUBCLASS, is a bit different from the order in which these values are usually
presented. In both cases, we need to have a better way to control the order of these classification
levels.
In the following version of this table, neither of the classification columns is in alphabetical order.
Common REPORT Problems
Vertically Concatenated Tables
Controlling the Order of the Groups
Sex
Female
Male
All
Height N
Median
Mean (Std)
Eastern
15 (19%)
23 (29%)
38 (48%)
Western
17 (21%)
25 (31%)
42 (53%)
Overall
32 (40%)
48 (60%)
80 (100%)
38.00
67.00
67.9 (3.91)
42.00
68.00
67.0 (3.04)
80.00
67.00
67.5 (3.49)
Weight N
38.00
Median
155.00
Mean (Std) 146 (34.96)
42.00
177.00
176 (30.51)
80.00
161.00
162 (35.87)
One way to achieve control of the order is through the use of formats. In this version of the report,
the values of the two grouping variables (CLASS and SUBCLASS) are integers (numeric starting
at 1) rather than characters. Formats are then used to display the appropriate values.
Most of the code from Section 10.1.1 remains the same with the following changes. First, two
additional formats are built.
proc format;
value $EWReg
'1','2','3','4' = 'Eastern'
'5'-'9','10'
= 'Western'
other
= 'Overall';
value $gender
'F'
= 'Female'
'M'
= 'Male'
other = 'All';
value class n
1 = 'Sex'
2 = 'Height'
3 = 'Weight';
value subclass o
1 = 'Female'
2 = 'Male'
3 = 'All'
Chapter 10: Solving Other Common Report Problems 333
4 = 'N'
5 = 'Median'
6 = 'Mean (Std)';
run;
The variable CLASS now takes on the values of 1 to 3 n and SUBCLASS the values of 1 to 6 o.
These assignments are made in the DATA step.
data prep1(keep=class subclass regname value);
set tran1(Keep=_type_ region sex _name_ col1);
length class subclass 4 p value $12 regname $7;
* Retain MEAN and N (denominator);
retain holdmean holdn .;
* Save the region name;
regname = put(region,$ewreg.);
* Build the stats for SEX;
if _name_='_FREQ_' then do;
* Only use observation counts (_FREQ_);
class = 1; q
subclass = 1*(sex="F") + 2*(sex='M') + 3*(sex=' ');
* Retain the denominator for future use;
if _type_ = 0 then holdn = col1;
* Value is count (%);
value = trim(left(put(col1,4.)))
||' ('
||trim(left(put(col1/holdn,percent.)))
||')';
output;
end;
r
* Build the stats for HEIGHT and WEIGHT;
* Ignore observations that contain _FREQ_;
else if _type_ in(0,2) then do;
if _name_ =: 'ht' then class=2; s
else class=3;
* Retain the MEAN for future use;
if index(_name_,'mean') then holdmean=col1;
else do;
if index(_name_,'std') then do;
subclass=6; t
* Value is a combination of MEAN and STD;
value = trim(left(put(holdmean,4.1)))
||' ('
||trim(left(put(col1,5.2)))
||')';
end;
else do;
* Other statistics;
subclass = 4*(upcase(substr(_name_,3))='N') u
+5*(upcase(substr(_name_,3))='MEDIAN');
334 Carpenter’s Complete Guide to the SAS REPORT Procedure
value = put(col1,6.2);
end;
output;
end;
end;
run;
p CLASS and SUBCLASS are designated as numeric variables.
q SEX has a CLASS value of 1.
r For the variable SEX, SUBCLASS takes on the values of 1, 2, or 3. Later these values are
displayed as ‘Female’, ‘Male’, and ‘All’ respectively through the use of the SUBCLASS.
format.
s Values are assigned to CLASS for the statistics associated with Height and Weight.
t The MEAN / STD combination has a SUBCLASS of 6.
u SUBCLASS receives the value of either 4 or 5 for N and MEDIAN.
After sorting, the first few observations of WORK.PREP1 are as follows:
Obs
1
2
3
4
5
6
7
8
9
10
11
12
13
class
1
1
1
1
1
1
1
1
1
2
2
2
2
subclass
1
1
1
2
2
2
3
3
3
4
4
4
5
value
15 (19%)
32 (40%)
17 (21%)
23 (29%)
48 (60%)
25 (31%)
38 (48%)
80 (100%)
42 (53%)
38.00
80.00
42.00
67.00
regname
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Overall
Western
Eastern
Notice that the rows are now in a different order than they were at this point in Section 10.1.1.
The PROC REPORT step now can take on slightly more responsibility. Because the CLASS and
SUBCLASS variables need to be formatted, the sort can be based on the unformatted value. This
allows the use of the ORDER=DATA option, while displaying the value with a format.
title1 'Common REPORT Problems';
title2 'Vertically Concatenated Tables';
title3 'Controlling the Order of the Groups';
proc report data=tran2 nowd;
column class subclass eastern western overall;
define class / group ' '
format=class. order=data;
define subclass / group ' '
format=subclass. order=data;
define eastern / display;
define western / display;
define overall / display;
Chapter 10: Solving Other Common Report Problems 335
compute after class;
line ' ';
endcomp;
run;
Here is the final table:
Common REPORT Problems
Vertically Concatenated Tables
Controlling the Order of the Groups
Eastern
15 (19%)
23 (29%)
38 (48%)
Western
17 (21%)
25 (31%)
42 (53%)
Overall
32 (40%)
48 (60%)
80 (100%)
Height N
Median
Mean (Std)
38.00
67.00
67.9 (3.91)
42.00
68.00
67.0 (3.04)
80.00
67.00
67.5 (3.49)
Weight N
Median
Mean (Std)
38.00
155.00
146 (34.96)
42.00
177.00
176 (30.51)
80.00
161.00
162 (35.87)
Sex
Female
Male
All
This table does what we want and does not look too bad. However, the numbers in the columns do
not line up very well. This problem is addressed in Section 10.1.3.
10.1.3 Text and Number Alignment in Derived Columns
In the previous table, the location of the decimal point (or where it would appear if present) is not
consistent from row to row. This can make the table more difficult to read. Fortunately, this
problem is easily solved by only slight alterations to the code.
Here is our desired table:
Common REPORT Problems
Vertically Concatenated Tables
Number Alignment
Eastern
15 (18.8%)
23 (28.8%)
38 (47.5%)
Western
17 (21.3%)
25 (31.3%)
42 (52.5%)
Overall
32 (40.0%)
48 (60.0%)
80 (100.0%)
Height N
Median
Mean (Std)
38
67.00
67.95 (3.91)
42
68.00
67.00 (3.04)
80
67.00
67.45 (3.49)
Weight N
Median
Mean (std)
38
155.00
145.89 (34.96)
42
177.00
176.14 (30.51)
80
161.00
161.78 (35.87)
Sex
Female
Male
All
336 Carpenter’s Complete Guide to the SAS REPORT Procedure
In the DATA step, several small changes have been made that alter the overall formatting of the
variable VALUE. In the example in Section 10.1.2, we were not consistent about how the leading
numbers were formatted and whether or not they were left-justified. In the following code, the
overall width of the format and the number of decimal points (if any) have been counted with the
objective of lining up the numbers. The same could have been done for the numbers in the
parentheses, but this was felt to be a bit of overkill.
data prep1(keep=class subclass regname value);
set tran1(Keep=_type_ region sex _name_ col1);
length class subclass 4 value $15 n regname $7;
* Retain MEAN and N (denominator);
retain holdmean holdn .;
* Save the region name;
regname = put(region,$ewreg.);
* Build the stats for SEX;
if _name_='_FREQ_' then do;
* Only use observation counts (_FREQ_);
class = 1;
subclass = 1*(sex="F") + 2*(sex='M') + 3*(sex=' ');
* Retain the denominator for future use;
if _type_ = 0 then holdn = col1;
* Value is count (%);
value = trim(put(col1,3.0)) o
||' ('
||trim(left(put(col1/holdn,percent8.1)))
||')';
output;
end;
* Build the stats for HEIGHT and WEIGHT;
* Ignore observations that contain _FREQ_;
else if _type_ in(0,2) then do;
if _name_ =: 'ht' then class=2;
else class=3;
* Retain the MEAN for future use;
if index(_name_,'mean') then holdmean=col1;
else do;
if index(_name_,'std') then do;
subclass=6;
* Value is a combination of MEAN and STD;
value = trim(put(holdmean,6.2)) q
||' ('
||trim(left(put(col1,5.2)))
||')';
end;
else do;
* Other statistics;
if upcase(substr(_name_,3))='N' then do;
subclass=4;
value = put(col1,3.); r
end;
else if upcase(substr(_name_,3))='MEDIAN' then do;
subclass=5;
value = put(col1,6.2); s
end;
p
Chapter 10: Solving Other Common Report Problems 337
end;
output;
end;
end;
run;
Because the largest number is in the 100s, we need to allow three spaces to the left of the column
that might or might not be needed for decimal places. Notice that in all of the formats that are
applied to the numbers outside of the parentheses, we consistently leave these three spaces.
n The size of the VALUE variable has been increased to accommodate the increased format
widths.
o The number of patients is displayed as an integer using the 3. format. Of course, this only
leaves space in our table for at most 999 patients.
p The percentage has been given more space and now includes a decimal point. This makes
things look nicer, but has nothing to do with the alignment issue.
q The mean is written with two decimal places, and the result of the PUT function is no longer
left-justified.
r N has no decimal places.
s The median is displayed with two decimal places. If you wanted only one decimal place,
the format would have been 5.1, which still preserves the three leading spaces.
When the display value contains a decimal point, you can use it to align the numbers directly. The
JUST=DEC attribute option can be used to align values in a column. In the following example,
the character variable X takes on four values (the fourth does not have a decimal point).
title1 'Common REPORT Problems';
title2 'Vertically Concatenated Tables';
title3 'Decimal Point Alignment';
proc report data=test nowd;
column x x=new;
define x / display
style={cellwidth=1in just=dec}
'Using Just=' ;
define new / display ;
run;
338 Carpenter’s Complete Guide to the SAS REPORT Procedure
The numbers in the left column are
aligned on the decimal point, even
though the values come from a
character variable.
Destination: PDF Style: PRINTER
The JUST=DEC does not work for HTML, but can be very useful in PDF and RTF.
SEE ALSO
In a discussion of some fairly sophisticated RTF techniques, Hamilton (2003) uses the STYLE=
option along with various RTF commands to produce aligned numbers.
10.1.4 Doing More in the PROC REPORT Step
In the previous examples, most of the work has been done in the steps leading up to the PROC
REPORT step. The DATA step is used as the primary tool for the calculation of the value to be
displayed. Because the PROC REPORT step can have a compute block, we should be able to
move at least some of these calculations out of the DATA step.
The following example takes advantage of the compute block, and it also uses formats for the
ordering of the regions. In this example, the region variable is now an ACROSS variable. This
eliminates the second PROC TRANSPOSE and a PROC SORT.
Rather than transposing the values of REGION, formats are used to control the values so that
REGION can be used as an ACROSS variable. Two new formats are created. The first n creates
a region code that can be used to put the regions into the correct order. The second o translates
the region code into the values that are to be displayed in the report.
proc format;
value $Reggrp n
'1','2','3','4' = 'a'
'5'-'9','10'
= 'b'
other
= 'c';
value $RegName o
'a' = 'Eastern'
'b' = 'Western'
'c' = 'Overall';
Chapter 10: Solving Other Common Report Problems 339
The PROC SUMMARY and PROC TRANSPOSE steps remain the same, except the $REGGRP.
n format is used to form the three regional groups (‘a’, ‘b’, and ‘c’).
In the DATA step, which has been simplified, several differences should be noted. Because we
are going to pass more of the work to the PROC REPORT step, we should expect the compute
blocks to become more interesting. Rather than constructing the composite character variable
VALUE in the DATA step, we pass the needed information (VAL1 and VAL2 p) to the compute
block, where the composite value is constructed as a computed variable.
data prep1(keep=class subclass reggrp val1 val2 p);
set tran1(Keep=_type_ region sex _name_ col1);
length class subclass 4 val1 val2 8 reggrp $1;
* Retain MEAN and N (denominator);
retain holdmean holdn .;
* Assign a region group code;
reggrp = put(region,$reggrp.);
q
* Build the stats for SEX;
if _name_='_FREQ_' then do;
* Only use observation counts (_FREQ_);
class = 1;
subclass = 1*(sex="F") + 2*(sex='M') + 3*(sex=' ');
* Retain the denominator for future use;
if _type_ = 0 then holdn = col1;
* Value is count (%);
val1 = col1; val2=holdn; r
output;
end;
* Build the stats for HEIGHT and WEIGHT;
* Ignore observations that contain _FREQ_;
else if _type_ in(0,2) then do;
if _name_ =: 'ht' then class=2;
else class=3;
* Retain the MEAN for future use;
if index(_name_,'mean') then holdmean=col1;
else do;
if index(_name_,'std') then do;
subclass=6;
* Value is a combination of MEAN and STD;
val1 =col1; val2 = holdmean; s
end;
else do;
* Other statistics;
if upcase(substr(_name_,3))='N' then subclass=4;
else if upcase(substr(_name_,3))='MEDIAN' then subclass=5;
val1=col1; val2=.; t
end;
output;
end;
end;
run;
340 Carpenter’s Complete Guide to the SAS REPORT Procedure
p REGGRP is used to hold the region codes. It will contain the formatted value of REGION
based on the $REGGRP. n format. In the previous examples, the character variable VALUE
was used to store the composite values of count and percent. In this approach, both numeric
values are passed into REPORT.
q In WORK.TRAN1, the variable REGION contains unformatted values of the original variable.
REGGRP contains the grouping codes that can be used for ordering the regions.
r The denominator is stored in VAL2.
s For the MEAN(STD), the two values are stored in VAL1 and VAL2.
t For other statistics, only VAL1 is used.
The second TRANSPOSE step is no longer needed, because the PROC REPORT step now uses a
define usage of ACROSS for REGGRP. Here is the resulting PROC REPORT step:
title1 'Common REPORT Problems';
title2 'Vertically Concatenated Tables';
title3 'Using ACROSS';
proc report data=prep1 nowd;
column class subclass reggrp,(val1 val2 value u);
define class / group ' '
format=class. order=data;
define subclass / group ' '
format=subclass. order=internal;
define reggrp / across 'Region'
format=$regname. order=internal; `
define val1
/ analysis noprint;
define val2
/ analysis noprint;
define value / computed ' ';
compute value / char length=15; w
if subclass in(1,2,3) then do;
_c5_ = trim(put(_c3_,3.0))
||' ('
||trim(left(put(_c3_/_c4_,percent8.1)))
||')';
_c8_ = trim(put(_c6_,3.0))
||' ('
||trim(left(put(_c6_/_c7_,percent8.1)))
||')';
_c11_ = trim(put(_c9_,3.0))
||' ('
||trim(left(put(_c9_/_c10_,percent8.1)))
||')';
end;
else do;
if subclass=4 then do;
* N;
_c5_ = put(_c3_,3.);
_c8_ = put(_c6_,3.);
_c11_ = put(_c9_,3.);
end;
else if subclass=5 then do;
* Median;
Chapter 10: Solving Other Common Report Problems 341
_c5_ = put(_c3_,6.2);
_c8_ = put(_c6_,6.2);
_c11_ = put(_c9_,6.2);
end;
else if subclass=6 then do;
* Mean (std);
* Value is a combination of MEAN and STD;
_c5_ = trim(put(_c4_,6.2))
||' ('
||trim(left(put(_c3_,5.2)))
||')';
_c8_ = trim(put(_c7_,6.2))
||' ('
||trim(left(put(_c6_,5.2)))
||')';
_c11_ = trim(put(_c10_,6.2))
||' ('
||trim(left(put(_c9_,5.2)))
||')';
end;
end;
endcomp;
compute after class;
line ' ';
endcomp;
run;
u The variables VAL1, VAL2, and VALUE are all nested under the value of REGGRP. This
arrangement forms nine columns with the internal names of _C3_, _C4_, ..., _C11_.
` The region names are assigned using the $REGNAME. o format. Because coded values (not
the formatted values) hold the desired order, the ORDER=INTERNAL option is specified.
w Because REGGRP is an ACROSS variable, the variable VALUE is never actually created.
Instead, the composite value is placed in the columns _C5_, _C8_, and _C11_. When forming
the value for _C5_, the columns _C3_ and _C4_ hold the values of VAL1 and VAL2
respectively. Except for the names of the columns, the form of the assignment statement is the
same in the compute block as it was in the DATA step in Section 10.1.3.
10.2 Automating the PROC REPORT Process
As we develop more and more sophisticated reports and groups of reports, the process of
managing and executing our programs becomes complicated. Manual steps are subject to user
error, and besides, we usually have other things we need to do. Automation of the process is
potentially a solution.
Within SAS, automation generally involves the SAS macro language, and within the context of
this book, we need to be able to use the macro language with PROC REPORT. The macro
language is well covered in Burlew (1995) and Carpenter (2004a), so only limited portions of the
language are covered in this section.
342 Carpenter’s Complete Guide to the SAS REPORT Procedure
MORE INFORMATION
The automation of the process that establishes links between documents is discussed in
Section 8.5.7.
SEE ALSO
Fairfield-Carter, et al. (2005) discuss the use of Jscript and VBScript to build an application that
writes the PROC REPORT step. Taylor (2005) discusses the use of VBA macros to import RTF
documents into MS WORD.
Michel (2005) uses the FLOW= option in a CALL EXECUTE example.
Foose (2002) presents a macro that writes a generalized PROC REPORT step.
Carpenter and Smith (2003a) use macro loops to create linked graphs and tables in an automated
process.
10.2.1 Things to Think about When Automating
Planning is essential when you start to automate a reporting process. We are often tempted to start
writing code and then figure out the details later. Doing some up-front planning can save a lot of
programming heartache later—especially as you try to solve problems using the macro language.
How you prepare, what steps you take, and what order you take them in will of course depend on
your own particular situation. There are, however, some common practices that you should
consider.
Define the End Products
Usually these will be the reports themselves. You need to know not only the shape and content of
the report, but also things such as the ODS destinations, file names, file locations, and file
replacement policies.
In an existing system, you can determine quite a bit of this information by surveying the current
system, reports, and file structures. Try to avoid the trap of doing everything the same as before.
When automating, you might have the opportunity to clean up a reporting process that has
“evolved”.
Determine Commonalities
As a code generator, the macro language is especially good at creating repeated code segments.
Within a reporting process, you often need to do things again and again. Repeated code might be a
series of statements within a step or might be a series of steps.
If you can identify themes within your code, it is much easier to isolate them and convert them to
macros or macro statements. When generalizing, you code the differences as macro parameters
and macro variables. Look for the following common elements:
Chapter 10: Solving Other Common Report Problems 343
Data preparation
Similar DATA and PROC steps might be needed to prepare the data
for a PROC REPORT step.
PROC REPORT step
Every PROC REPORT step has some common elements,
especially the REPORT, COLUMN, and DEFINE statements.
Report computations
Compute blocks are often repeated across PROC REPORT steps.
ODS styles
If you are standardizing styles across reports, try to move the
definitions, style options and attributes into macro statements, or
better yet into the style definitions themselves.
Use macro parameters and macro variables to allow for variations between reports. This practice
is especially useful for file names.
Establish Naming Conventions and File Structure
An automated system will be building the names of files, data sets, macros, macro variables, and
data variables. These names might be based on instructions passed into the macro by the user, or
more likely they will be based on the data itself. The coding is vastly simplified if there is a
consistency in the naming. For instance, if we want to create two data tables of patients, one for
each gender, naming them PAT_M and PAT_F would be more logical than MALEPAT and
FEMALES. This would be especially true if the data contained the variable SEX that took on the
values of F and M.
SEE ALSO
The process of automating a series of programs is described by Carpenter and Smith (2000 and
2001). Although PROC REPORT is not directly mentioned, McQuown (2004) discusses some of
the issues that come up when preparing to automate a reporting process. Rhodes (2005) discusses
report automation in general terms.
10.2.2 Macro Variable Resolution Issues
Within SAS, it is standard knowledge that macro variables are not resolved when they are placed
inside of single quotation marks. Generally, when macro variables are to be resolved within a
quoted string, the quotation marks must be double. In the following two TITLE statements, the
macro variable &BCOLOR remains unresolved in the first title, whereas it resolves to YELLOW
in the second.
%let bcolor = yellow;
title1 'background color is &bcolor';
title2 "background color is &bcolor";
Although this rule is generally true within a REPORT compute block, it is not strictly true, and
the exceptions are worth understanding. Consider the following compute block:
%let bcolor = blue;
proc report data=sashelp.class nowd;
. . . PROC REPORT statements are not shown . . .
344 Carpenter’s Complete Guide to the SAS REPORT Procedure
compute weight;
attrib = "style={background=&bcolor}";
call define(_row_,'style',attrib);
endcomp;
compute after;
* Show that &bcolor is resolved;
line attrib $62.;
endcomp;
run;
As we would anticipate, the macro variable &BCOLOR, which is enclosed in double quotation
marks, resolves to BLUE, and the background color is successfully reset. Interestingly, the
following, which encloses the macro variable in single quotation marks, will also work.
%let bcolor = red;
proc report data=sashelp.class nowd;
. . . PROC REPORT statements are not shown . . . .
compute weight;
attrib = 'style={background=&bcolor}';
call define(_row_,'style',attrib);
endcomp;
run;
It works, and why it works can be important to us, especially in more complex examples. The
variable ATTRIB contains the unresolved macro variable reference. However, because it is in
single quotation marks, the parser has yet to recognize it as such. When CALL DEFINE is
executed, it must first resolve the variable ATTRIB (this happens at execution, not compilation).
This causes the scanner to parse the resolved value of ATTRIB. This rescanning reveals the
macro variable to the parser, where it is now recognized and resolved (into RED). This second
pass is important in the context of the compute block, because this does not happen in the DATA
step.
In the previous examples, the attribute is fairly simple. However, for some attributes, the attribute
information itself needs to be quoted. This was the case for the examples in Section 8.8.2. The
following compute block creates constant tip text for the CLINNUM column. Notice that the tip
text itself is within double quotation marks and the whole string is within single quotation marks.
compute clinnum;
attrib = 'style={flyover="Bethesda"}';
call define(_col_,'style',attrib);
endcomp;
The previous assignment creates tip text that would be constant for all values in the CLINNUM
column. We could use the CLINNAME variable to provide different tip text for each row, and
doing so slightly complicates the assignment statement. Essentially, we are replacing “Bethesda”
with the value of CLINNAME.
compute clinnum;
attrib = 'style={flyover="'||trim(clinname)||'"}';
call define(_col_,'style',attrib);
endcomp;
Chapter 10: Solving Other Common Report Problems 345
The quoting in the previous compute block becomes a bit complicated. Another approach that
simplifies the quoting uses a macro variable. The following step takes advantage of the way that
macro variables are parsed when they are enclosed within single quotation marks.
compute clinnum;
call symput('clin',clinname);
attrib = 'style={flyover="&clin"}';
call define(_col_,'style',attrib);
endcomp;
If &CLIN had not been enclosed in single quotation marks (here the whole STYLE= option is in
single quotation marks), this approach would not work. The single quotation marks delay the
resolution of the macro variable until after the macro variable has been created by CALL
SYMPUT. This timing is a bit tricky, especially for DATA step programmers who could never
get away with this. Remember, a macro variable is resolved when it is “seen” by the macro parser.
In this case, it is not “seen” until the value of ATTRIB is evaluated.
SEE ALSO
Downing (2004) shows an example of the use of a macro variable in a compute block.
10.3 Coordinating Graphics with PROC REPORT
On first thought, it seems that the only necessary coordination between PROC REPORT and
SAS/GRAPH would be when a graph and report appear on the same page. However, when we
look at the topic with a bit more depth, it becomes apparent that there is much more that can be
done. Certainly, in Section 8.2.2 we used the PREIMAGE style attribute to add a logo to the
graph. Of course, instead of a logo, that object (which was a GIF file) could have been any graph
created by SAS/GRAPH.
It turns out, of course, that this is only one small application of the broader topic.
SEE ALSO
Mitchell (2006) discusses the inclusion of Likert scale plots in a report.
10.3.1 Using CALL DEFINE to Import Graphics
There is more than one way to import graphs into a table generated by PROC REPORT. Often
these graphs are generated by SAS/GRAPH. However, much of the discussion in this section
applies equally to graphs generated in other ways.
There are two primary methods of identifying the graph to be imported. Unfortunately, although
the code is simple, the process is not as straightforward as it seems it should be.
STYLE=
GRSEG
identifies the file containing the graphic.
identifies a GRSEG catalog entry that has been generated by SAS/GRAPH.
This attribute only works with markup destinations and is not
appropriate for the PDF and RTF destinations.
346 Carpenter’s Complete Guide to the SAS REPORT Procedure
The easiest and most flexible method involves the use of the STYLE= option and the PREIMAGE
or POSTIMAGE attribute (PREIMAGE was introduced in Section 8.2.2). Specified in the CALL
DEFINE statement, the general syntax is as follows:
call define('cell_location',
'style',
"style={postimage='image_file_name'}");
Assuming you want to place a GIF file in a cell for the IMAGE column, the compute block might
be the following:
compute image / char length=10;
image=' ';
call define('image',
'style',
"style={postimage='c:\temp\western.gif'}");
endcomp;
Another method that works under more limited conditions is the use of the GRSEG attribute.
Again, the CALL DEFINE statement is used, but this time we point directly to a graphics
(GRSEG) entry in a SAS catalog rather than to a file. Here is the general syntax
call define('cell_location',
'grseg',
'catalog_entry_name');
A potential compute block using the GRSEG attribute for the IMAGE2 column might be the
following:
compute image2 / char length=80;
image2=' ';
call define('image2',
'grseg',
"work.gseg.western.grseg");
endcomp;
For both of these methods, the sizing of the graphs is tricky at best. This difficulty is an artifact of
how the graphs are rendered by the ODS destination. Unfortunately, the CELLHEIGHT and
CELLWIDTH attributes do not help for resizing cells that contain graphics. The ultimate size of
the graphic as it appears on the table usually needs to be controlled when the graph is generated.
Fortunately, if you are using SAS/GRAPH, there are several ways to do this, and these are
discussed in the following paragraphs. However, you will almost certainly need to do some trial
and error work to get things to appear as you would like.
The size of the graph is not the only issue that you have to deal with. Because the graphs are
imported into PROC REPORT, they must first be created in some other form. In addition to
GRSEG entries in SAS graphics catalogs, there are a number of file types that can be used to hold
graphic images. In SAS/GRAPH, these different file types are built using specialized software
known as devices. The term devices dates to when all the graphs were rendered on physical
devices such as plotters. Now files are often built with virtual rather than physical devices, and
these are designated using the DEVICE (DEV) option. Some of the more popular of the file types
used to store graphs include CGM, EMF, GIF, PNG, and JPEG. Just to complicate matters a bit
more, these file forms are not equally well received by the various ODS destinations.
Chapter 10: Solving Other Common Report Problems 347
The following table shows the interactions between ODS destination (SAS 9.1.3), file type, and
method for importing the graphic.
Attribute Type = STYLE
File
type
EMF
GIF
JPEG
PNG
HTML
OK; scale sensitive; style
background color is
preserved; image GIF files
are created.
OK; scale sensitive.
OK; scale sensitive; image
GIF files are created.
OK scale sensitive; image
GIF files are created.
ODS Destinations
PDF
RTF
Images load, but tend to be
ugly.
Images are not imported.
Large images overlap and
are outside of the cell
boundaries;
smaller images can fit OK.
Images are imported
correctly.
Images are imported
correctly.
Images are not imported.
Images are imported
correctly.
Images are imported
correctly.
Attribute type = GRSEG
File
type
EMF
GIF
JPEG
PNG
HTML
OK; scale sensitive.
OK; scale sensitive.
OK; scale sensitive; image
GIF files are created.
OK; scale sensitive; image
GIF files are created.
ODS Destinations
PDF
RTF
Images are distorted and outside of the cell boundaries.
It is anticipated that these relationships will change under SAS 9.2. PNG will become the default
file type for HTML in SAS 9.2.
Using Device Drivers to Control Size
Device drivers are used by SAS/GRAPH to construct the graphic file in a way that is appropriate
to the final destination. When a file is built, the destination is designated through the extension or
file type, e.g., GIF or JPEG. Each device driver (and SAS/GRAPH is shipped with hundreds of
drivers) contains sizing information as part of the driver definition. Although it is possible to build
customized drivers (see Carpenter (1995, p.124) and Carpenter and Smith (2003b)), the topic is
outside the scope of this book. Fortunately, options are available that can override the size
specifications that are contained in the driver’s definition.
When you want to create and import a GIF file, several drivers are available that have already
been defined to produce images of various sizes ( Carpenter and Smith 2003b). The smallest
might be small enough for some tables.
348 Carpenter’s Complete Guide to the SAS REPORT Procedure
The following example is used with variations throughout this section. The table is rather simple;
however, for illustration purposes two columns of graphics have been added (in this case a simple
pie chart). The graphic in the first column (IMAGE) is added using the STYLE= option, and the
second (IMAGE2) is added using the GRSEG option.
ods listing; n
* Adding graphics;
proc format; o
value $regfile
'1','2','3' = 'NoEast'
'4'
= 'SoEast'
'5' - '8'
= 'MidWest'
'9', '10'
= 'Western';
run;
pattern1
pattern2
pattern3
pattern4
v=psolid
v=psolid
v=psolid
v=psolid
c=gray22
c=gray66
c=grayaa
c=grayee
r=1;
r=1;
r=1;
r=1;
Z
title1;
* Clear the graphics catalog;
proc datasets library=work mt=cat nolist;
delete gseg;
quit;
* Dummy records force proper pattern statement usage;
data dummy(keep=region proced wt ht edu);
do region='5 ','1','4','9';
do proced=' ','1','2','3';
wt=.; ht=.; edu=.;
output;
end;
end;
run;
* Dummy records are appended to force each region to
* have all procedures. This causes patterns to match
* across regions;
data clinics(keep=reggrp proced wt ht edu cnt);
set rptdata.clinics(keep=region proced wt ht edu)
dummy(in=indum);
reggrp = put(region,$regfile.); q
* CNT=1 for valid records;
cnt = 1-indum;
run;
%macro bldimage;
proc sql noprint; r
select distinct(reggrp)
into :reg1-:reg99
from clinics;
%let regcnt = &sqlobs;
quit;
Chapter 10: Solving Other Common Report Problems 349
%do i = 1 %to ®cnt;
filename propie "&path\results\&®&i...gif"; s
goptions dev=gif160 t
gsfname=propie;
proc gchart data=clinics(where=(reggrp="&®&i"));
pie proced / noheading missing u
type=sum
sumvar=cnt
slice=none
name="&®&i"; `
run;
quit;
%end;
%mend bldimage;
%bldimage
ods listing close;
ods html path="&path\results"
body="ch10_3_1b.html"
style=default;
ods pdf file="&path\results\ch10_3_1b.pdf"
style=minimal;
ods rtf file="&path\results\ch10_3_1b.rtf"
style=rtf;
title1 'Interfacing with REPORT';
title2 'Adding Graphics';
title3 'Reduced Size GIF Files';
proc report data=clinics nowd split='*';
column reggrp image image2 edu ht wt;
define reggrp / group width=10 'Region' order=formatted;
define image / computed 'style';
define image2 / computed 'grseg';
define edu / analysis mean 'Years of*Education'
format=9.2 ;
define ht
/ analysis mean format=6.2 'Height';
define wt
/ analysis mean format=6.2 'Weight';
compute image / char length=10; w
image=' ';
imageloc = "style={postimage='&path\results\"
||trim(reggrp)|| ".gif'}";
* Specify the image using the STYLE=;
call define('image',
'style',
imageloc);
endcomp;
compute image2 / char length=10;
image2=' ';
imageloc = "work.gseg."||trim(reggrp)||'.grseg';
350 Carpenter’s Complete Guide to the SAS REPORT Procedure
* Specify the image using the GRSEG attribute;
call define('image2',
'grseg',
imageloc);
endcomp;
run;
ods _all_ close;
n The ODS LISTING destination must be open to create graphs.
o The format $REGFILE. is defined to group values of regions.
p PATTERN statements define four gray scale colors for the four levels of the procedure variable
PROCED.
q The user-defined format $REGFILE. is used to consolidate and name regions.
r A PROC SQL step is used to count the number of distinct regions (®CNT), and to store
each region name in a macro variable of the form ®1, ®2, ... .
s The file to be created is named in a FILENAME statement. The macro variable &®&I..
resolves to the region name. The fileref, PROPIE, is tied to the graph with the GSFNAME
option.
t The device driver GIF160 creates the smallest image of any of the predefined GIF drivers.
u A pie chart is created for procedure type. Missing values (no procedure) are included in the
diagram.
` Each GRSEG entry receives the unique name of that region.
w The images are loaded into a cell through the use of the CALL DEFINE statement. Both
methods are demonstrated here.
The HTML file created by using the GIF160 device driver results in a table that includes the PIE
diagrams associated with the types of medical procedures. Notice that both methods work equally
well.
Destination: HTML Style: Default
Chapter 10: Solving Other Common Report Problems 351
The PDF file imports successfully only the images using the STYLE= option.
Destination: PDF Style: Minimal
Using GOPTIONS to Control Size
If we want the images to be even smaller, we need to either create a customized device driver,
which is not particularly difficult, or override the size specifications by using GOPTIONS. The
latter is easier to discuss in the context of this book.
The graphics options XPIXELS and YPIXELS are two of several options that can be used to
control the size of the graphics image. In the following example, the size of the graph has been
reduced from the previous example, which used the device GIF160, by a fourth.
* GIF160 is the smallest gif driver pre-defined in SAS/GRAPH
* It has XPIXELS=160 and YPIXELS=120;
goptions dev=gif
xpixels=40 ypixels=30
gsfname=propie;
The HTML file shows:
Destination: HTML Style: Default
352 Carpenter’s Complete Guide to the SAS REPORT Procedure
In the examples in this section, all of the text on the reduced graphics is eliminated by using the
SLICE=NONE option on the PIE statement. Another technique that has limited effectiveness on
most devices is to use the GOPTION statement to set the default text height to very small (say
HSIZE=.001). Another technique for eliminating text is to make the text the same color as the
background. Most graphics procedures have options that can be used to eliminate text altogether.
Text elimination is fairly easy for procedures that use the AXIS statement.
SEE ALSO
Poppe (2005) uses the CALL DEFINE routine to bring GRSEG entries into the report.
10.3.2 Using GPRINT and GREPLAY
In Section 10.3.1, graphs have been inserted into the report. You might also want to combine the
report and graph onto a single page. Although the ODS LAYOUT statement holds a great deal of
promise for this type of task, it is currently only experimentally available (SAS 9.1.3). Often this
task can also be accomplished manually at the word-processing stage.
An alternative methodology that can be used under limited circumstances is to render the table as
a graphics object and then use SAS/GRAPH to redisplay the graph and the table together. This
technique utilizes the GPRINT procedure, which can capture a text file and convert it to a graphic
object (GRSEG entry). Notice that the table itself must be a text file, which means that you are
limited to either the LISTING destination or to a text file created using the PRINTER destination.
Once the table has been brought into SAS/GRAPH, it can be redisplayed by using the GREPLAY
procedure. This procedure can display multiple graphics at one time and is therefore handy for
combining text with graphics.
ods listing;
goptions reset=all;
filename textrpt "&path/results/ch10_3_2.txt";
proc printto print=textrpt new;
run;
n
title1;
proc report data=sashelp.class nowd;
column sex age height weight;
define sex / group 'Gender' format=$6.;
define age / group "Age" width=3;
define height / mean 'Height' format=6.1;
define weight / mean 'Weight' format=6.2;
break after sex / skip summarize suppress;
run;
proc printto;
run;
* Clear the GSEG catalog from the work library;
proc datasets library=work mt=cat nolist; o
delete gseg;
quit;
Chapter 10: Solving Other Common Report Problems 353
goptions ftext=simplexu htext=1.8;
* Import the table into SAS/GRAPH; p
proc gprint fileref=textrpt name='classrpt';
run;
goptions ftext=swiss htext=2;
* create an overall TITLE; q
proc gslide name='Slide';
title1 'Interfacing with REPORT';
title2 h=1.5 'Rendering a Report with SAS/GRAPH';
run;
symbol1 c=red value=dot i=stdm2j l=33;
symbol2 c=blue value=square i=stdm2j l=1;
data class;
set sashelp.class;
* Separate (dither) gender in the plots; r
age = age + (sex='F')*.05 + (sex='M')*-.05;
run;
* Create a scatter plot;
proc gplot data=class; s
plot height*age=sex/name='height';
title1 'Height';
run;
plot weight*age=sex/name='weight';
title1 'Weight';
run;
symbol1 c=red value=dot i=none;
symbol2 c=blue value=square i=none;
plot weight*height=sex/name='w_h';
title1 'Weight vs Height';
run;
quit;
* Build a template and replay the graphs and table;
proc greplay igout=gseg nofs tc=gtempl8;
tdef three_t t
1/ ulx=0 uly=100 urx=100 ury=100
llx=0 lly= 0 lrx=100 lry= 0
2/ ulx=0 uly=90 urx=50 ury=90
llx=0 lly=45 lrx=50 lry=45
3/ ulx=50 uly=90 urx=100 ury=90
llx=50 lly=45 lrx=100 lry=45
4/ ulx=0 uly=40 urx=50 ury=40
llx=0 lly= 0 lrx=50 lry= 0
5/ ulx=50 uly=40 urx=100 ury=40
llx=50 lly= 0 lrx=100 lry= 0;
template three_t;
354 Carpenter’s Complete Guide to the SAS REPORT Procedure
treplay 1:Slide
2:height
3:w_h
4:weight
5:classrpt;
run;
quit;
u
n PROC PRINTTO is used to route the report to a text file.
o The GSEG catalog is cleared as a preparation for the new graphs that are to be created.
p PROC GPRINT converts the text table to a graphics image, which is then stored in the GSEG
catalog.
q An overall title for the combined plots is created.
r Ages are slightly dithered so that they will not overlap in the graphs.
s The three scatter plots are generated in a GPLOT step.
t Each template panel is defined with the coordinates of each of the four corners. The coordinate
pairs are in percentages from the lower left corner.
u Specific graphs are assigned to specific template panels.
The resulting graphic contains all three plots and the overall titles, as well as the PROC REPORT
table.
MORE INFORMATION
Section 10.5 discusses alternative methods of placing multiple reports on a page, including the use
of the ODS LAYOUT statement in Section 10.5.1.
Chapter 10: Solving Other Common Report Problems 355
SEE ALSO
Leprince and Li (2003) discuss methods for improving this technique. They also show how to
automate the process using the macro language. Carpenter and Shipp (1995) discuss the use of
templates, GREPLAY, and the redisplay of graphics. The automated generation of GREPLAY
templates is discussed in detail by Perry Watts (2002).
Pete Lund (2005) shows an example that displays both reports and graphics through the use of the
ODS LAYOUT statement.
10.3.3 Using the Annotate Facility to Generate Lines
In the example in Section 8.2.4, blank lines and columns were filled using cell attributes assigned
with the STYLE= option. You can also use the annotate facility in SAS/GRAPH to add content to
your report. In the following example, the annotate facility has been used to generate lines that
can be used to separate portions of the report.
ods pdf style=printer
file="&path\results\ch10_3_3.pdf"
startpage=never; n
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
. . . Portions of this step are not shown . . .
* Add the annotate macros to the autocall library;
%annomac o
* Build the annotate data set;
data Annolines; p
length function color $8; q
retain xsys ysys '5'; r
* Syntax for the LINE macro;
%*LINE (x1, y1, x2, y2, color, line_type, size);
%line(.5,69.6,50.5,69.6,blue,1,30) s
%line(.5,80.5,50.5,80.5,red,1,30)
%line(42.2,67,42.2,94,green,1,30)
run;
* Use the annotate data set;
proc gslide anno=annolines; t
run;
quit;
ods _all_ close;
356 Carpenter’s Complete Guide to the SAS REPORT Procedure
n Because we are overlaying the results of two different procedures, we do not want the results to
be presented on separate pages. STARTPAGE=NEVER prevents the generation of a new page
at the procedure step boundary.
o A number of predefined annotate macros (%LINE s is one of these) can be used to generate
annotate instructions. These macros are not normally available until they are loaded into the
autocall library by the %ANNOMAC macro.
p The annotate data set is named ANNOLINES. Each observation in this data set contains
an instruction that will be used by PROC GSLIDE t.
q The variables FUNCTION and COLOR are generated by the %LINE macro. These two
variables should always be explicitly specified as $8.
r XSYS and YSYS specify the coordinate system that is used by the annotate facility. With this
specification, all (x,y) coordinate pairs measure percentages from the lower left corner of
the graphic area (exclusive of titles and footnotes).
s The %LINE annotate macro is used to generate the annotate observations. You could also
build these observations directly through the use of a series of assignment statements. The
%LINE macro is defined with the following series of positional parameters:
X1,Y1
X2,Y2
COLOR
LINE_TYPE
SIZE
coordinate of the start of the line
coordinate of the end of the line
color of the line (do not quote the value)
type of line, designated by a number (1=solid, 2=dashed, etc.)
thickness of the line
t PROC GSLIDE is used to render the annotation instructions held in the annotate data set.
Because the STARTPAGE= option has been set to NEVER n, the results of this procedure
overlay the table produced by PROC REPORT.
Destination: PDF Style: PRINTER
Chapter 10: Solving Other Common Report Problems 357
Because the annotate elements (in this case lines) must be placed manually through trial and error,
this technique has a limited utility in an environment where report tables are constantly
undergoing change or where the numbers of rows or columns are not fixed.
Sometimes the annotate lines have a better visual impact if the report does not already have a lot
of lines. The previous example has been repeated below using the JOURNAL style.
Destination: PDF Style: JOURNAL
Although this example adds lines to the report, the annotate facility could also be used to add a
multitude of symbols (including arrows), additional text, and even other graphics. You can even
add boxes or irregularly shaped polygons to highlight groups of cells.
MORE INFORMATION
Section 8.2.4 uses the STYLE= option to add lines to the report.
SEE ALSO
Lund (2005) has an example of a table that has lines added using the annotate facility. Carpenter
(1999 and 2006a) introduces the annotate facility.
10.4 Workarounds for Monospace-Only Options
As discussed in Chapter 4, “Only in the LISTING Destination,” a number of PROC REPORT
options work only in the LISTING destination. For many of these options, this is fine, as they are
no longer needed for other destinations. However, occasionally you need to be able to mimic the
functionality of these options for non-LISTING destinations.
358 Carpenter’s Complete Guide to the SAS REPORT Procedure
A number of examples that are spread throughout this book discuss a number of these work
arounds. A FAQ page that addresses these issues is currently available at
http://support.sas.com/rnd/base/topics/templateFAQ/repoption.html
Column Width
For the LISTING destination, the width of a column is by default determined by the length and
type of the variable. Further control can be gained through the specification of a format (Section
2.4.2). However, more robust methods for controlling column width are available. For the
LISTING destination, the COLWIDTH option (Section 4.4.3) and the WIDTH= option (Section
4.5.1) can also be used to control column width independently from the formatted width.
These options have little or no effect on column width for other destinations. For non-monospace
destinations, the CELLWIDTH attribute modifier can be used (Section 8.2.4).
Text Wrapping
In the LISTING destination, text that does not fit in the designated width of the cell is truncated.
However, the FLOW= option (Section 4.5.2) can be used to wrap the text within the cell. For
other destinations, the cell width is generally expanded to accommodate the text. When you want
to force text to wrap, you can use the escape character sequences shown in Section 8.6.5.
For markup language destinations, you can force text not to wrap by setting the tagset attribute
modifier TAGATTR to NOWRAP on the STYLE= option (Section 9.3.2).
style(column)={tagattr="nowrap"}
Column Spacing
The SPACING= option (Section 4.5.3) can be used to increase or decrease (Section 6.4.1
concatenates columns) the space between columns in the LISTING destination. A combination of
LINE statements and the STYLE= option (Section 8.2.4) can be used to accomplish similar
results in other destinations. The physical concatenation of values can be accomplished in the
compute block, and the results can be displayed in any destination (Section 7.6.2).
Page Control
Page numbering is automatic in the LISTING destination and becomes more complex in other
destinations. The very definition of a page is quite destination-specific (Section 6.8). Horizontal
page breaks can be controlled with the _PAGE_ option (Section 5.4) through the use of the
PAGESIZE/LINESIZE options (Section 4.4.6), vertical page breaks (Section 6.1.7), and the BY
statement (Section 6.6).
Compute blocks are used to detect and count page breaks in Section 7.7.3. Escape character
sequences are used in Section 8.6.1 to produce page numbers in the PDF and RTF destinations.
Column Justification
Justification options can appear as system options, PROC statement options (Section 4.4.2),
DEFINE statement options (Section 6.1.2), and COMPUTE statement options (Section 8.2.2).
The effect of these options depends on the destination and, probably in the future, the release of
SAS.
For non-LISTING destinations, the JUST= attribute modifier, which can be used on the STYLE=
option and on the CALL DEFINE statement (Section 8.2 and 8.3), is preferred.
Chapter 10: Solving Other Common Report Problems 359
Generation of Panels
The LISTING destination can create panels through the use of the PANELS= option (Section
4.4.4). The ODS COLUMNS= option can be used to create multicolumn reports for other
destinations (Section 8.9). ODS LAYOUT (Section 10.5.1) can be used to combine multiple
reports on a page.
HEADLINE and Other Horizontal Lines
The HEADLINE option (Section 4.1) places a solid line between the header text and the body of
the report. This option is generally not needed for non-LISTING destinations; however, it can be
simulated in a number of ways when it is needed. Other LISTING destination options that create
horizontal lines include OL, UL, DOL, and DUL on the BREAK and RBREAK statements.
All of these options can easily be replaced using the LINE statement. This statement can be used
to create a series of repeated characters (Section 7.8.2). It can also create a blank line that can be
made to appear solid through the use of style attributes (Section 8.2.4).
Repeated characters used as spanning headers are not recognized outside of the LISTING
destination. Actually, in some circumstances they can even cause problems (Section 9.3.3).
HEADSKIP and Other Vertical Spaces
In the LISTING destination, a blank space is placed between the header text and the body of the
report with the HEADSKIP option, and between summary groups in the body of the report with
the SKIP option (Section 3.2.1). These spaces can also be supplied through the use of the LINE
statement, which is actually more flexible (Sections 2.6.1 and 8.2.4).
Instead of inserting a space, you can increase the size of a given cell to simulate a space. In the
following example, the height of the summary row is increased (to 60 pixels). This adds blank
space following the summary line.
ods html path="&path\results"
body="ch10_4.html";
title1 'Inserting blank spaces';
title2 'Adding Summary Row Height';
proc report data=sashelp.class nowd;
column sex age height weight;
define sex / group;
define age / group;
define height / mean;
define weight / mean;
break after sex / summarize suppress;
compute after sex;
call define(_row_,'style','style={htmlstyle="height:60px"}');
endcomp;
run;
ods html close;
360 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here is a portion of the HTML table:
Destination: HTML Style: DEFAULT
10.5 Generating Separate Reports on the Same
Page
When you need to put multiple reports on a page, you can take several approaches. The manual
approach is to generate the reports separately and then manually cut or copy them and paste them
onto the page the way you want them. Because this approach is not nearly enough fun to discuss
in this book, it is fortunate that we do have some alternative approaches.
There have already been a couple of examples in this chapter that combine the results of multiple
procedures on the same report.
Section 10.3.1
Section 10.3.2
Section 10.3.3
Graphics objects are imported into a table created by a REPORT step.
GREPLAY is used to add one or more reports to a graphic.
The output from a PROC REPORT step is augmented with lines built by the
annotate facility in SAS/GRAPH.
10.5.1 ODS LAYOUT
The ODS LAYOUT statement allows us to effectively design how a final page is configured. It
does this by using the ODS REGION statement to define one or more regions on the page. These
regions are then filled by the output of procedures such as PROC REPORT.
Chapter 10: Solving Other Common Report Problems 361
ODS LAYOUT is turned on and off by using the ODS LAYOUT START and ODS LAYOUT
STOP statements. Before each step that generates output for the page, there is an ODS REGION
statement, which has as a minimum the following options:
X=
Y=
WIDTH=
HEIGHT=
horizontal position to start the region
vertical position to start the region
width of the region
height of the region
Each of these options can be measured in the standard scales such as inches (IN), millimeters
(MM), or points (PTS). The regions have to be at least large enough to hold your table. If the
region size is too small, the table is truncated.
The following program writes a PROC GSLIDE graphic and two PROC REPORT tables to a
single PDF page.
ods pdf style=printer
file="&path\results\ch10_5_1.pdf";
ods layout start;
n
ODS region x=.5in y=.1in width=7in height=10in; o
title1 'Common REPORT Problems';
title2 f=swissb 'Multiple Reports on a Page';
title3 f='Arial bold' 'Using ODS Layout';
title4 f=swiss '
Office
Furniture';
proc gslide; p
run;
ods region x=1in y=1.5in width=3in height=4in; q
title1;
proc report data=sashelp.prdsale(where=(prodtype='OFFICE'))
nowd;
column country product,actual;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
run;
ods region x=5in y=1.5in width=3in height=4in; r
proc report data=sashelp.prdsale(where=(prodtype='FURNITURE'))
nowd;
column country product,actual;
define country / group;
define product / across;
define actual / analysis sum
format=dollar8.
'Sales';
run;
ods layout end; s
ods _all_ close;
362 Carpenter’s Complete Guide to the SAS REPORT Procedure
n ODS LAYOUT is turned on.
o The first region is defined. It contains the across report titles generated by PROC GSLIDE.
Z The GSLIDE procedure is very useful for displaying just titles. This step places the three titles
into the first region.
q The second region (Office sales report) is defined. This region actually overlaps the first
region. This could cause a problem if the titles in the first region were larger.
r The third region (Furniture sales report) is defined. Using the same value for Y as in q causes
both regions to be aligned horizontally.
s ODS LAYOUT must be closed when all the assignments have been made.
Destination: PDF Style: PRINTER
Currently (SAS 9.1.3) ODS LAYOUT has not yet matured enough to be fully useful to the PROC
REPORT programmer. However, it is anticipated that substantial improvements, including a wide
range of options, will be available in SAS 9.2.
MORE INFORMATION
The use of graphics options in the TITLE statement is discussed in Section 8.7.
10.5.2 HTML Reports
The definition of a page in HTML is very different from the definition in other destinations.
(HTML pages are virtual and have no real relationship to sheets of paper). However, placing
multiple reports within an HTML file is as easy as the specification of the steps themselves. Any
number of PROC steps can be placed between the ODS HTML and ODS HTML CLOSE
statements. The results of these steps will all appear on the same file.
Chapter 10: Solving Other Common Report Problems 363
ods html style=journal
file="&path\results\ch10_5_2.html";
proc report data= . . . .;
proc report data= . . . .;
ods html close;
The NEWFILE= option can be used to force new physical files with each PROC step.
10.5.3 RTF and PDF Reports: Using STARTPAGE=NEVER
For both the RTF and PDF destinations, a new page is generated for each PROC step. This means
that by default, any time a new procedure is executed, a new page is created in the output
document. The STARTPAGE= option can be used to control page generation. The
STARTPAGE=NEVER option tells ODS that a new page should not be generated between
procedures.
ods rtf style=rtf
file="&path\results\ch10_5_3.rtf"
bodytitle
startpage=never;
ods pdf style=printer
file="&path\results\ch10_5_3.pdf"
startpage=never;
title1 'Common REPORT Problems';
title2 'Multiple Reports on a Page';
title3 'Product Sales';
proc report data=sashelp.prdsale
nowd;
column country product,actual;
. . . Portions of the code are not shown . . .
run;
title3 'Shoe Sales';
proc report data=sashelp.shoes
(where=(region in:('Cana', 'West', 'United') &
product in:('Men', 'Boot', 'Sport')))
nowd;
column region product,sales;
. . . Portions of the code are not shown . . .
The PDF document shows the titles once per page.
364 Carpenter’s Complete Guide to the SAS REPORT Procedure
Notice that for the PDF
destination, only the first of
the two TITLE3 definitions
actually shows up in the
report.
Destination: PDF Style: PRINTER
The report for the RTF destination shows all three titles for each report:
In the RTF destination, the
BODYTITLE option places the title
within the body of the report.
All three titles are displayed (TITLE3
correctly changes) for each table.
Destination: RTF Style: RTF
Chapter 10: Solving Other Common Report Problems 365
10.5.4 Aligning Columns across Reports
In the combined reports of the previous example, the left column of the two reports (COUNTRY
and REGION) contains similar values, but because of different character lengths the cells
themselves are not aligned. We can correct this through the use of the CELLWIDTH attribute
modifier.
title1 'Common REPORT Problems';
title2 'Multiple Reports on a Page';
title3 'Aligned Columns';
proc report data=sashelp.prdsale
nowd;
column country product,actual;
define country / group style(column)={cellwidth=25mm};
define product / across ;
define actual / analysis sum
style(column)={cellwidth=20mm}
format=dollar8.
' ';
run;
proc report data=sashelp.shoes
(where=(region in:('Cana', 'West', 'United') &
product in:('Men', 'Boot', 'Sport')))
nowd;
column region product,sales;
define region
/ group style(column)={cellwidth=25mm};
define product / across;
define sales
/ analysis sum
style(column)={cellwidth=20mm}
format=dollar8.
' ';
run;
Because the cell
widths have been
controlled, the
columns are now
aligned.
Destination: PDF Style: PRINTER
366 Carpenter’s Complete Guide to the SAS REPORT Procedure
C h a p t e r
11
Details of the PROC REPORT Process
11.1 Step Sequence Review 368
11.2 Building a Simple Table with Summary Lines 371
11.3 Compute Block Processing 372
11.3.1 Creating a Computed Variable 372
11.3.2 Multiple Compute Blocks 373
11.3.3 Summary Lines and Compute Blocks in the Same Report 374
11.3.4 Using COMPUTE BEFORE and COMPUTE AFTER with Summary
Lines 375
11.4 Using the ACROSS Define Usage 378
An overview of the primary steps that take place during the generation of the report is presented
in Section 1.4.2. The events and process-timing issues that take place during the processing of the
compute block are discussed in Section 7.1. Although these steps and processing events are
discussed in more detail in this chapter, it is assumed that you have read and at least mostly
understood the previous sections.
The examples in this chapter utilize a portion of the data table SASHELP.CLASS.
title1 'Student Weight and Height';
proc print data=sashelp.class
(where=(age in(12,13))
keep=age sex weight height);
var age sex weight height;
run;
368 Carpenter’s Complete Guide to the SAS REPORT Procedure
Student Weight and Height
Obs
Age
Sex
2
3
6
7
9
10
13
16
13
13
12
12
13
12
12
12
F
F
M
F
M
M
F
M
Weight
84.0
98.0
83.0
84.5
84.0
99.5
77.0
128.0
Height
56.5
65.3
57.3
59.8
62.5
59.0
56.3
64.8
This data set contains age, height,
and weight information on a set of
middle school students. This chapter
only uses 12 and 13 year olds.
SEE ALSO
Although SAS Technical Report P-258 (1993, Chapter 10), SAS Guide to the REPORT
Procedure, Reference, Release 6.11 (1995, Chapter 5) and Lavery (2003) each discuss the
sequencing of events in detail, the concepts and terminology have not been updated to the reflect
the current versions of PROC REPORT. Russ Lavery’s “An Animated Guide to the SAS
REPORT Procedure,” which is included on the CD that accompanies this book, discusses the
REPORT processes in detail.
11.1 Step Sequence Review
Understanding the PROC REPORT process requires that we look at the sequence of processing
events. For simple tables, this understanding does not need to be terribly detailed. As the PROC
REPORT step becomes more complex, however, the details are critical if we are to successfully
code complicated compute blocks.
The following PROC REPORT step has two variables with a usage of GROUP, but does not
include any compute blocks or summary lines.
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_1 nowd;
column age sex n weight height;
define age
/ group;
define sex
/ group 'Gender' format=$6.;
define n
/ 'N' format=2.;
define weight / analysis mean 'Mean Weight' format=6.1;
define height / analysis mean 'Mean Height' format=6.2;
run;
When this code is submitted, a series of internal steps (or phases) occurs, resulting in the final
report. The following diagram provides an overview of this process that takes place in memory.
Chapter 11: Details of the PROC REPORT Process 369
Evaluation Phase
PROC REPORT begins by evaluating all of the PROC REPORT step statements. If any compute
blocks are present (there are none in this example), the SAS language elements and LINE
statements are set aside for later.
Setup Phase
Next, after the statements have been evaluated, the setup phase uses the MEANS/SUMMARY
engine to sort the input data for ORDER and GROUPS (AGE and SEX) and computes any
summarizations (the MEAN statistic for HEIGHT and WEIGHT). When summarizations or
statistics are calculated, the results are held in the computed summary information area in
memory.
370 Carpenter’s Complete Guide to the SAS REPORT Procedure
Report Row Phase
After PROC REPORT is done with these preliminary setup phase tasks, the report can be built
row by row in the report row phase. In this phase, PROC REPORT builds each report row using
data from the input data set or, when needed, the computed summary information. If any compute
blocks are present, they are executed during this phase. It is during this phase that REPORT sends
each completed row to all the ODS destinations (LISTING, PDF, etc.) that are currently open.
The OUT= option can be used to generate an output data set that gives us a glimpse of how the
report rows are generated.
Student Weight and Height
Obs
Age
Sex
n
Weight
Height
1
2
3
4
12
12
13
13
F
M
F
M
2
3
2
1
80.75
103.50
91.00
84.00
58.0500
60.3667
60.9000
62.5000
_BREAK_
Here is the final report:
Student Weight and Height
Age
12
13
Gender
F
M
F
M
N
2
3
2
1
Mean
Weight
80.8
103.5
91.0
84.0
Mean
Height
58.05
60.37
60.90
62.50
This simple report contains no summary
lines and no computed variables.
When there are no summary lines and no compute blocks, the movement of values from the input
data set to the final report is very straightforward, with all of the summarization taking place
during the setup phase, which is prior to the creation of even the first report row.
MORE INFORMATION
Section 7.1 discusses timing issues when compute blocks are included in the step.
Chapter 11: Details of the PROC REPORT Process 371
11.2 Building a Simple Table with Summary Lines
When summary lines are added to a simple (no compute block) model, the process is only slightly
modified.
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_2 nowd;
column age sex n weight height;
define age / group;
define sex / group 'Gender' format=$6.;
define n
/ 'N' format=2.;
define weight / analysis mean 'Mean Weight' format=6.1;
define height / analysis mean 'Mean Height' format=6.2;
break after age / summarize suppress skip;
rbreak after / summarize;
run;
The summary rows for this report are created in the setup phase of the PROC REPORT step, and
the resulting values are stored in memory in the computed summary information area. The
computed summary information includes summary rows that will be needed by the BREAK
statement (_BREAK_=‘Age’) and the RBREAK statement (_BREAK_= ‘_RBREAK_’). These
summary rows can be seen in the data set created by the OUT= option (Obs 3, 6, and 7).
Simple Table with Summary Lines
Output Data Set
Obs
1
2
3
4
5
6
7
Age
Sex
n
Weight
Height
12
12
12
13
13
13
.
F
M
2
3
5
2
1
3
8
80.750
103.500
94.400
91.000
84.000
88.667
92.250
58.0500
60.3667
59.4400
60.9000
62.5000
61.4333
60.1875
F
M
_BREAK_
Age
Age
_RBREAK_
All the summary information needed by a BREAK or RBREAK statement is generated in the
setup phase and is held in the computed summary information area of memory. The computed
summary information can also contain rows or values that exist because of COMPUTE BEFORE
or COMPUTE AFTER logic. However, this summary information is not written to the report
unless it is also tied to a BREAK or RBREAK statement with a SUMMARIZE option.
372 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here is the final report:
Simple Table with Summary Lines
Age
12
13
Gender
F
M
N
2
3
5
Mean
Weight
80.8
103.5
94.4
Mean
Height
58.05
60.37
59.44
F
M
2
1
3
91.0
84.0
88.7
60.90
62.50
61.43
8
92.3
60.19
This report contains a
summary line for each
age as well as a reportwide summary across
all ages at the bottom of
the report.
11.3 Compute Block Processing
When compute blocks are added to the PROC REPORT step, the required processing steps
become more complex, but the fundamental process of creating one report row at a time during
the report row phase is the same.
11.3.1 Creating a Computed Variable
In this PROC REPORT step, the Body Mass Index (BMI) is calculated as a computed variable
from the values of the report items HEIGHT and WEIGHT (see Section 7.2.1).
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_3_1 nowd;
column age sex n weight height bmi;
define
define
define
define
define
define
age / group;
sex / group 'Gender' format=$6.;
n
/ 'N' format=2.;
weight / analysis mean 'Mean Weight' format=6.1;
height / analysis mean 'Mean Height' format=6.2;
bmi / computed format=4.1 'BMI';
compute bmi;
bmi = weight.mean / (height.mean*height.mean) * 703;
endcomp;
run;
During the setup phase, the MEAN statistics for HEIGHT and WEIGHT are calculated and stored
in the computed summary information. Then, during the report row phase, each report row is
processed one row at a time. First the report items, including the computed variable BMI, are
initialized to missing values. After row initialization, the report items are processed one at a time
from left to right (the order is taken from the COLUMN statement). It is during this process that
Chapter 11: Details of the PROC REPORT Process 373
the report items are written to the report, again from left to right one item at a time. If a report
item has an associated compute block, the compute block is executed when that report item is
processed.
At the point where the compute block for BMI is processed, the mean information for HEIGHT
and WEIGHT is available in the computed summary information. This information is retrieved
and used in the calculation of BMI.
compute bmi;
bmi = weight.mean / (height.mean*height.mean) * 703;
endcomp;
The process is repeated for each report row, and as each report row is completed the report items
are written to the final report (and to the output data set if an OUT= option has been specified).
All of the variables in this PROC REPORT step appear on the COLUMN statement, and the only
variable defined in the compute block (BMI) is a computed variable. Consequently, this report
step does not define any temporary variables.
11.3.2 Multiple Compute Blocks
In this REPORT step, the measures for HEIGHT and WEIGHT are converted to centimeters and
kilograms respectively, and then the value of BMI is calculated based on the converted values.
Each of these three compute blocks is associated with a report item, and each is executed (from
left to right) when that report item is encountered during the report row phase.
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_3_2 nowd;
column age sex n weight height bmi;
. . . DEFINE statements are not shown . . .
compute weight;
weight.mean = weight.mean/2.2;
endcomp;
compute height;
height.mean = height.mean*2.54;
endcomp;
compute bmi;
bmi = weight.mean / (height.mean*height.mean) * 1e4;
endcomp;
run;
At the end of the setup phase, the computed summary information contains the untransformed
mean values for WEIGHT and HEIGHT. Since each of these report items also has a compute
block associated with it, the compute block is executed during the report row phase as its
associated report item is encountered. As the report row is built, each of the compute blocks is
processed, and the results are written to the computed information summary as well as to the
report row that is being created. Because the value for BMI depends on the results of the first two
compute blocks, it is very important that the variables HEIGHT.MEAN and WEIGHT.MEAN
reflect the result of the preceding compute blocks’ calculations.
374 Carpenter’s Complete Guide to the SAS REPORT Procedure
In this example, if BMI were to the left of either HEIGHT or WEIGHT on the COLUMN
statement, the calculation of BMI would fail. The following COLUMN statement would not
work:
column age sex n weight bmi height;
The value of HEIGHT (converted or otherwise) would not yet be available for use in the BMI
compute block. In fact, it would be missing altogether, and the user would receive an
“uninitialized variable” error.
11.3.3 Summary Lines and Compute Blocks in the Same
Report
When summary lines are requested in the PROC REPORT step, the summarization itself takes
place in the setup phase. During processing of the BREAK or RBREAK statements in the report
row phase, this summary information is retrieved in the same manner as in the previous examples.
When a compute block for a computed variable also appears in a PROC REPORT step, the report
row creation is handled in the same manner as in the example in Section 11.3.2. The following
example combines both a computed variable and summary (BREAK and RBREAK) statements.
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_3_3 nowd;
column age sex n weight height bmi;
. . . DEFINE statements are not shown . . .
break after age / summarize suppress skip;
rbreak after / summarize;
compute bmi;
bmi = weight.mean / (height.mean*height.mean) * 703;
endcomp;
run;
A review of the output data set (WORK.OUT11_3_3) shows that the automatic temporary
variable _BREAK_ is set to ‘Age’ on those report rows where the BREAK AFTER AGE
statement was executed (Obs 3 and 6) and the value of ‘_RBREAK_’ where the RBREAK
statement was executed (Obs 7).
BMI a Computed Variable
With Summary Lines
Obs
Age
Sex
n
Weight
Height
bmi
1
2
3
4
5
6
7
12
12
12
13
13
13
.
F
M
2
3
5
2
1
3
8
80.750
103.500
94.400
91.000
84.000
88.667
92.250
58.0500
60.3667
59.4400
60.9000
62.5000
61.4333
60.1875
16.8459
19.9665
18.7832
17.2489
15.1173
16.5161
17.9023
F
M
_BREAK_
Age
Age
_RBREAK_
Chapter 11: Details of the PROC REPORT Process 375
The report contains the correctly calculated value for BMI on the rows resulting from the BREAK
and RBREAK statements.
BMI a Computed Variable
With Summary Lines
Age
12
13
Gender
F
M
N
2
3
5
Mean
Weight
80.8
103.5
94.4
Mean
Height
58.05
60.37
59.44
BMI
16.8
20.0
18.8
F
M
2
1
3
91.0
84.0
88.7
60.90
62.50
61.43
17.2
15.1
16.5
8
92.3
60.19
17.9
The compute block for
BMI has been executed on
the summary rows
generated by the BREAK
and RBREAK statements.
11.3.4 Using COMPUTE BEFORE and COMPUTE AFTER with
Summary Lines
The processing steps become a bit more interesting when there are both COMPUTE BEFORE or
COMPUTE AFTER statements along with BREAK and RBREAK statements. Remember that
although the COMPUTE BEFORE or COMPUTE AFTER statement generates a summary row in
the computed summary information, that row is not written to the table unless there is a
corresponding BREAK or RBREAK statement with a SUMMARIZE option.
As you read through this example, keep in mind that a compute block associated with a summary
row, e.g., COMPUTE BEFORE or COMPUTE AFTER, is only executed when its corresponding
row in the computed summary information is being processed. Compute blocks associated with
report items, including computed variables, are executed for all report rows, including summary
rows. When two or more compute blocks are to be executed for any given report row, the report
item compute blocks are always executed first (left to right). For a summary row, any compute
blocks associated with that summary row are executed only after all the report items (and their
associated compute blocks) have been processed.
The following somewhat contrived example demonstrates both the timing and the relationships
between compute blocks and report items. In this report, percentages are calculated for a grouping
variable and, although it is a bit silly here, percentages are also calculated across the entire report
as well. In order to do this, two compute blocks are used to create two temporary variables that
hold the denominator for the percentage calculation. TOTN n holds the overall number of
students while TOTAGE o holds the total number of students in each age group.
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_3_4a nowd;
column age sex n percent;
define
define
define
define
age / group;
sex / group 'Gender' format=$6.;
n
/ 'N' format=2.;
percent/ computed format=percent8. 'Percent';
376 Carpenter’s Complete Guide to the SAS REPORT Procedure
break after age / summarize suppress skip;
rbreak after / summarize;
compute before;
totn = n; n
endcomp;
compute before age;
totage = n; o
endcomp;
compute percent; Z
if _BREAK_='_RBREAK_' then percent=n/totn;
else percent = n/ totage;
endcomp;
run;
n The overall number of students is stored in the temporary variable TOTN. This compute block
is executed only once.
o The temporary variable TOTAGE is assigned the value of the number of students in this age
group. This compute block is executed at the start of each age group.
Z This compute block is used to calculate the computed variable PERCENT. Because PERCENT
is a report item, this compute block is executed for every report row.
The incoming data is summarized and stored in the computed summary information during the
setup phase. Therefore, at the beginning of the report row phase, the summary values are already
available to be retrieved from memory when each of the compute blocks execute.
Summary Lines with a Percentage
Obs
Age
1
.
2
3
4
5
12
12
12
12
6
7
8
9
10
13
13
13
13
.
Sex
n
8
F
M
F
M
percent
.
_BREAK_
_RBREAK_
5
2
3
5
.
0.40000
0.60000
1.00000
Age
3
2
1
3
8
0.60000
0.66667
0.33333
1.00000
1.00000
Age
o
Age
o
n
During the setup phase the
incoming data is read,
summarized, and stored in
the computed summary
information area of
memory. This information
is processed during the
report row phase.
Age
_RBREAK_
By reviewing the output data set, we can see that the presence of the COMPUTE BEFORE block
caused the _RBREAK_ information to be captured at the top of the report n during the setup
phase. During report row phase, this compute block is executed, and the previously summarized
value of N is retrieved from memory (the computed summary information area) and then used to
create the temporary variable TOTN. It is at this time that this row is also written to the output
data set.
Chapter 11: Details of the PROC REPORT Process 377
In the same fashion, the COMPUTE BEFORE AGE block o caused the setup phase to
summarize age-group data into the computed summary information area. During the report row
phase, this compute block is executed when the second and sixth report rows are processed. It is at
this time that N is retrieved from memory and used in the calculation of the temporary variable
TOTAGE.
During the report row phase, the COMPUTE PERCENT block p is executed for each report row,
and the value of PERCENT is computed based on the values of the two temporary variables
TOTN and TOTAGE. This of course means that the value for PERCENT will be missing if either
of these two temporary variables are missing. When the first report row is processed, two compute
blocks will execute (n and p).
When more than one compute block is executed for a given report row, it can sometimes be
important to understand their order of execution. For compute blocks associated with report items,
the compute blocks are always executed from left to right—in the same order as the variables on
the COLUMN statement. It is also important to keep in mind that COMPUTE BEFORE and
COMPUTE AFTER blocks always execute after any compute blocks associated with report
items. This means that the results of all report item compute blocks are available for use in
COMPUTE BEFORE and COMPUTE AFTER blocks. This also means that these same report
item values can be altered and overwritten in these compute blocks.
When the first report row (Obs 3 in the previous output data set) is written, the count of 2 for the
12-year-old females is divided by 5, which is the value for the TOTAGE temporary variable.
Then the format of PERCENT8. is applied to the result of the calculation and the first report row
is completed and sent to all open ODS destinations (in this case the LISTING destination is the
only open destination).
Here is the final report as shown in the LISTING destination:
Summary Lines with a Percentage
Age
12
13
Gender
F
M
N
2
3
5
Percent
40%
60%
100%
F
M
2
1
3
67%
33%
100%
8
100%
Processing and division happen the same way for the next report row (males’ summary
information). Their count of 3 is divided by the TOTAGE (5), and the results are formatted with
the PERCENT8. format and written to the report row. Finally, the BREAK AFTER AGE
statement executes and the total count of 5 is divided by TOTAGE (also 5). After the format is
applied, the calculated value of 100% is written to the report row.
This same processing is repeated for the 13-year-old section. Then on the last report row of the
report, when it is time to process the report row generated by the RBREAK statement, the
summarized value for TOTN is retrieved from memory and used in the division for PERCENT.
378 Carpenter’s Complete Guide to the SAS REPORT Procedure
It is tempting to want to assign sequential processing like takes place in the DATA step to the
report row phase. However, PROC REPORT holds all the needed summary information in
memory, so it is available for both COMPUTE BEFORE and COMPUTE AFTER processing
during the construction of each report row. The process is different enough from the DATA step
that we can confuse what is actually happening if we depend too much on our DATA step
processing knowledge. This becomes even more apparent as we add a report item with a define
usage of ACROSS.
11.4 Using the ACROSS Define Usage
Although adding a variable with a define usage of ACROSS changes how we address the columns
of the report, the addition of an ACROSS variable causes only a minor change in the timing of
compute block execution. Because PROC REPORT assigns absolute column numbers at the end
of the setup phase, they are always available to be used in a compute block. Most people do not
use these absolute column names for simple reports. However, when a PROC REPORT step has a
compute block that is meant to be used on the ACROSS variable, absolute column names must be
used. The absolute column names ensure that there is no ambiguity about how report items or
summarized information should be used in the creation of the final report.
The following example is only slightly modified from the one discussed in Section 11.3.4. The
variable SEX now has a define usage of ACROSS, and the BREAK statement has been removed.
Here is the REPORT step:
proc format ;
value $gender
'F' = 'Female'
'M' = 'Male';
run;
proc report data=sashelp.class(where=(age in(12,13)))
out=out11_4 nowd;
column age sex,(n percent);
define
define
define
define
age / group;
sex / across 'Gender' format=$gender.;
n
/ 'N' format=2.;
percent/ computed format=percent8. 'Percent';
rbreak after / summarize dol;
compute before; n
totn = _c2_ + _c4_;
endcomp;
compute before age; o
totage = _c2_ + _c4_;
endcomp;
Chapter 11: Details of the PROC REPORT Process 379
compute percent; Z
if _BREAK_='_RBREAK_' then do; q
* Percentage for a given age;
_c3_ = _c2_/totn;
_c5_ = _c4_/totn;
end;
else do; r
* Percentages for a given age;
_c3_ = _c2_/totage;
_c5_ = _c4_/totage;
end;
endcomp;
run;
This generates the following report:
Summary Lines with a Percentage
Report with an ACROSS Variable
Age
12
13
N
2
2
==
4
Gender
Female
Male
Percent
N Percent
40%
3
60%
67%
1
33%
======== == ========
50%
4
50%
_C1_ _C2_
_C3_
_C4_
The number of students for each age group
is presented, and the percentages within age
group and across all ages are calculated.
Absolute column numbers are added for
reference.
_C5_
n The overall total number of students is determined.
o The total number of students within this age is determined.
p Percentages in this report item compute block are calculated.
q The percentages across all ages are calculated.
r The percentages within each age are calculated.
In the previous example (see Section 11.3.4), we were able to use simple assignment statements to
create TOTN and TOTAGE. With the introduction of the ACROSS variable, PROC REPORT
must now retrieve the N for genders individually on the same report row (n and o) in order to
acquire the correct total for the division. The absolute column names prevent any ambiguity about
exactly which value(s) of the summarized N is being used in the compute block.
380 Carpenter’s Complete Guide to the SAS REPORT Procedure
The values for N (_C2_ and _C4_) were determined during the data summarization process and
have already been stored in the computed summary information. The values for PERCENT (_C3_
and _C5_) are determined in a compute block p. Here is a view of the output data set:
Summary Lines with a Percentage
Report with an ACROSS Variable
Obs
1
2
3
4
5
6
Age
.
12
12
13
13
.
_C2_
_C3_
4
2
2
2
2
4
_C4_
_C5_
_BREAK_
4
3
3
1
1
4
.
.
0.60000
0.20000
0.33333
0.50000
_RBREAK_
Age
.
.
0.40000
0.40000
0.66667
0.50000
Age
_RBREAK_
Notice that columns _C3_ and _C5_have missing values for Obs 2 and not for Obs 4, even though
both are summary rows that were generated as a result of the COMPUTE BEFORE AGE block
Y. Initially this may seem confusing; however, the COMPUTE BEFORE AGE block executes
after the compute block for the report item PERCENT. Consequently, for Obs 2 the value for the
temporary variable TOTAGE has not yet been assigned. Since Obs 1, 2 and 4 are not report rows
(they will not appear in the final report), the only relevant columns on these rows are _C3_ and
_C5_ (the 2 columns that represent the N for every gender).
The PERCENT values (_C3_ and _C5_) are only of importance on report rows that are actually
written to the final report. The final report results would be the same if the code for PERCENT
were changed to the following:
compute percent; p
if _BREAK_='_RBREAK_' then do;
_c3_ = _c2_/totn;
_c5_ = _c4_/totn;
end;
else if _BREAK_=' ' then do;
_c3_ = _c2_/totage;
_c5_ = _c4_/totage;
end;
endcomp;
The resulting SAS data set shows that the calculation for PERCENT was not done on the BREAK
BEFORE AGE observations created from the report rows.
Summary Lines with a Percentage
Report with an ACROSS Variable
Obs
1
2
3
4
5
6
Age
.
12
12
13
13
.
_C2_
4
2
2
2
2
4
_C3_
.
.
0.40000
.
0.66667
0.50000
_C4_
4
3
3
1
1
4
_C5_
.
.
0.60000
.
0.33333
0.50000
_BREAK_
_RBREAK_
Age
Age
_RBREAK_
Chapter 11: Details of the PROC REPORT Process 381
Before each report row is processed, the report row items are initialized to missing. Then during
the processing of the report row, the calculation for PERCENT is performed. Both _C3_ and
_C5_ represent PERCENT, so when either of these report items is encountered, the PERCENT
compute block is executed.
MORE INFORMATION
The SAS program S11_4c.sas demonstrates that the PERCENT compute block is executed
multiple times for each report row.
382 Carpenter’s Complete Guide to the SAS REPORT Procedure
A p p e n d i x
1
Exercise Solutions
A1.1 Solutions to Chapter 1 Exercises 383
A1.2 Solutions to Chapter 2 Exercises 384
A1.3 Solutions to Chapter 3 Exercises 389
A1.4 Solutions to Chapter 4 Exercises 391
A1.5 Solutions to Chapter 5 Exercises 394
A1.6 Solutions to Chapter 6 Exercises 397
A1.7 Solutions to Chapter 7 Exercises 402
A1.8 Solutions to Chapter 8 Exercises 408
A1.1 Solutions to Chapter 1 Exercises
1. What are the three processing phases of a PROC REPORT step?
Evaluation
Setup
Report row
Determine what needs to be done.
Process and summarize incoming data as required. Create the computed
summary information table.
Build the report table one row at a time.
384 Carpenter’s Complete Guide to the SAS REPORT Procedure
2. What is the difference between temporary variables and report variables?
Report variables (report items) are noted on the COLUMN statement and may or may not
appear on the report itself.
Temporary variables never appear on the COLUMN statement and never appear in the report.
3. What two PROC REPORT statement options will you use on virtually every PROC
REPORT statement?
DATA=
declares the name of the incoming data table.
NOWD
closes the interactive report system. Alternates include NOFS and
NOWINDOWS.
A1.2 Solutions to Chapter 2 Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor sports
clothing and equipment store. Generate a report that lists total PROFIT for each YEAR.
Apply the DOLLAR. format to the total PROFIT.
title1 'Total profit per year';
proc report data=sashelp.orsales nowd;
column year profit;
define year / group;
define profit / analysis sum format=dollar15.2;
run;
Total profit per year
Year
1999
2000
2001
2002
Profit in USD
$13,272,490.82
$16,159,587.28
$13,754,040.02
$15,898,930.35
2. Building on the solution for Exercise 1:
a. List the profit for each PRODUCT_LINE as a subgroup of YEAR.
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales nowd;
column year product_line profit;
define year / group;
define product_line / group ;
define profit / analysis sum format=dollar15.2;
run;
Appendix 1: Exercise Solutions 385
Total profit per year
Separated by Product Line
Year Product Line
Profit in USD
1999 Children
$541,552.09
Clothes & Shoes
$4,189,628.63
Outdoors
$2,887,433.36
Sports
$5,653,876.74
2000 Children
$618,803.21
Clothes & Shoes
$5,009,983.42
Outdoors
$3,585,371.04
. . . Portions of the table are not shown . . .
b. Repeat with PRODUCT_LINE as an ACROSS variable.
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales nowd;
column year product_line,profit;
define year / group;
define product_line / across ;
define profit / analysis sum format=dollar15.2;
run;
Total profit per year
Separated by Product Line
Year
1999
2000
2001
2002
Product Line
Children Clothes & Shoes
Outdoors
Profit in USD
Profit in USD
Profit in USD
$541,552.09
$4,189,628.63
$2,887,433.36
$618,803.21
$5,009,983.42
$3,585,371.04
$538,462.06
$4,158,502.80
$3,222,898.41
$718,302.42
$4,739,806.53
$3,704,810.34
Sports
Profit in USD
$5,653,876.74
$6,945,429.61
$5,834,176.75
$6,736,011.06
3. Building on the solution for Exercise 2a, add the following to the report:
Specify text for at least one column.
Use the SPLIT= option.
Use the LINE statement to create a break after each year.
Create a user-defined format that redisplays the product line “Sports” to
“Sports Equipment”.
Add a line of text after the report using the LINE statement.
386 Carpenter’s Complete Guide to the SAS REPORT Procedure
proc format;
value $sequip
'Sports' = 'Sports Equipment';
run;
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit;
define year
/ group;
define product_line
/ group
f=$sequip.
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
compute after year;
line ' ';
endcomp;
compute after;
line @25 'Profits in US dollars';
endcomp;
run;
Total profit per year
Separated by Product Line
Product
Year Groups
1999 Children
Clothes & Shoes
Outdoors
Sports Equipment
Annual
Profit
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
2000 Children
Clothes & Shoes
Outdoors
Sports Equipment
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
2001 Children
Clothes & Shoes
Outdoors
Sports Equipment
$538,462.06
$4,158,502.80
$3,222,898.41
$5,834,176.75
(continued)
Appendix 1: Exercise Solutions 387
2002 Children
Clothes & Shoes
Outdoors
Sports Equipment
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
Profits in US dollars
4. The following step contains no typos and all the variables exist; why will it fail?
proc report data=sashelp.class nowd;
column age sex,height;
define age / group;
define sex /across;
define height/display;
run;
HEIGHT is nested within SEX; consequently PROC REPORT needs a statistic to
display. None are available because HEIGHT has a usage of DISPLAY. Changing the
usage of HEIGHT to ANALYSIS allows the default statistic, SUM, to be used.
5. We would like to create a numeric counter (CNT) for each age group, and then we
want to display the counter along with the age group. We are expecting the
following code to result in a table with one row per age group. What goes wrong and
how can it be fixed?
proc sort data=sashelp.class out=cl1;
by age;
run;
data class;
set cl1;
by age;
if first.age then cnt+1;
run;
title 'Count the Age Groups';
proc report data=class nowd;
column cnt age sex,height;
define cnt / order;
define age / group;
define sex /across;
define height/analysis mean;
run;
388 Carpenter’s Complete Guide to the SAS REPORT Procedure
Count the Age Groups
Sex
F
M
cnt
Age
Height
Height
1
11
51.3
.
57.5
2
12
59.8
.
56.3
.
.
57.3
.
59
.
64.8
3
13
56.5
.
65.3
.
.
62.5
. . . Portions of the table are not shown . . .
The counter, CNT, is built successfully in the DATA step (an example in Section 7.2.1 builds
this counter in a compute block, therefore avoiding the DATA step). However, because the
ORDER usage implies a detail report, the GROUP usage cannot be applied. Groups can be
nested, so changing the define usage for CNT to GROUP solves the problem.
proc sort data=sashelp.class out=cl1;
by age;
run;
data class;
set cl1;
by age;
if first.age then cnt+1;
run;
title 'Count the Age Groups';
proc report data=class nowd;
column cnt age sex,height;
define cnt / group;
define age / group;
define sex /across;
define height/analysis mean;
run;
Appendix 1: Exercise Solutions 389
Count the Age Groups
cnt
1
2
3
4
5
6
Age
11
12
13
14
15
16
Sex
F
M
Height
Height
51.3
57.5
58.05 60.366667
60.9
62.5
63.55
66.25
64.5
66.75
.
72
A1.3 Solutions to Chapter 3 Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Create a report that shows total PROFIT for
each PRODUCT_LINE within each year. You can build on the results of Exercise 3
in Chapter 2.
Using the BREAK and RBREAK statements:
Additionally summarize across product lines and across years.
Experiment with the SUPPRESS, SUMMARIZE, and SKIP options.
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Profit Summaries';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit;
define year / group;
define product_line
/ group
'Product*Groups';
define profit/ analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress ol skip;
rbreak after
/ summarize dol skip;
compute after;
line @25 'Profits in US dollars';
endcomp;
run;
390 Carpenter’s Complete Guide to the SAS REPORT Procedure
Although the SKIP statement is included on the RBREAK statement, it does nothing.
Both the SKIP and SUPPRESS options are ignored on the RBREAK statement. Here is
the table:
The PDF destination, using the
JOURNAL style here, does not utilize
the OL or the DOL options that have
been specified in the BREAK and
RBREAK statements respectively. Nor
does the PDF destination make use of
the SKIP option in the BREAK
statement.
All of the destinations ignore the SKIP
option on the RBREAK statement.
2. The data table SASHELP.RETAIL contains quarterly sales information. For each
YEAR use the quarterly sales to calculate the following sales statistics:
number of quarters (why might we need this statistic, when there are always
four quarters in a year?)
mean quarterly sales
standard deviation of the quarterly sales
title1 'Quarterly Retail Sales';
title2 'Quarterly Sales Statistics';
proc report data=sashelp.retail nowd;
column year sales,(n mean std);
define year / group width=4;
define sales/ analysis;
format sales;
run;
Appendix 1: Exercise Solutions 391
The format for SALES has been turned off. Otherwise all three statistics would receive
the same format. Notice the N size for 1994. The data only contains the first two quarters
for that year.
Quarterly Retail Sales
Quarterly Sales Statistics
YEAR
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
Retail sales in millions of $
n
mean
std
4
257.5 30.621343
4
287 31.208973
4
313 24.508502
4
348.25 35.71531
4
382 29.473152
4
399 46.238512
4
480.5 50.348784
4
541 45.350487
4
603 40.963398
4
648 36.147845
4
683.5 60.135957
4
736.75 48.030372
4
801 82.190835
4
894.5 97.988095
2
937 86.267027
A1.4 Solutions to Chapter 4 Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Generate a report that lists total PROFIT for
each PRODUCT_LINE within each YEAR. List the products ACROSS the report.
You might want to build on the results of Exercise 2b in Chapter 2.
Include the following:
HEADLINE and HEADSKIP options
repeated characters in the spanning header for product line
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales
headline headskip nowd;
column year product_line,profit;
define year / group;
define product_line / across
'- Products -';
define profit / analysis sum format=dollar15.2;
run;
392 Carpenter’s Complete Guide to the SAS REPORT Procedure
The following LISTING table uses the SAS Monospace font so that the horizontal lines
are correctly rendered.
Total profit per year
Separated by Product Line
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ Products ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Children Clothes & Shoes
Year
Profit in USD
Profit in USD
Outdoors
Sports
Profit in USD
Profit in USD
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
1999
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
2000
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
2001
$538,462.06
$4,158,502.80
$3,222,898.41
$5,834,176.75
2002
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
2. The data table SASHELP.RETAIL contains quarterly sales information. List the
columns YEAR, DATE, and SALES. Do the following:
Use YEAR as a grouping variable.
Use the PANELS=, BOX, and PSPACE= options.
title1 'Quarterly Retail Sales';
title2 'Creating Multiple Panels';
proc report data=sashelp.retail
nowd panels=99 box pspace=10;
column year date sales;
define year / group;
define date / display;
define sales/ analysis;
run;
Because the usage of DATE is DISPLAY, a detail-level report is created, and groups are
not formed. Even though groups are not formed, GROUP still causes the rows to be ordered
on YEAR and dictates that YEAR is displayed only on the first row of each year.
Appendix 1: Exercise Solutions 393
. . . Portions of the table are not shown . . .
3. Building on Exercise 2 in this section, use the WIDTH= and SPACING= options.
title1 'Quarterly Retail Sales';
title2 'Creating Multiple Panels';
title3 'Using DEFINE Statement Options';
proc report data=sashelp.retail
nowd panels=99 box pspace=10;
column year date sales;
define year / group width=4;
define date / display spacing=7;
define sales/ analysis width=8;
run;
. . . Portions of the table are not shown . . .
394 Carpenter’s Complete Guide to the SAS REPORT Procedure
A1.5 Solutions to Chapter 5 Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Create a report that shows total PROFIT for
each PRODUCT_LINE within each YEAR. You might wish to build on the results
of Exercise 3 in Chapter 2
Compute the percentage of annual sales that can be attributed to each product line.
proc format;
value $sequip
'Sports' = 'Sports Equipment';
run;
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit percent;
define year / group;
define product_line
/ group
f=$sequip.
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
define percent/ computed 'Product*Percentage'
format=percent10.2;
compute before year;
total = profit.sum; n
endcomp;
compute after year;
line ' '; o
endcomp;
compute percent;
percent = profit.sum/total; Z
endcomp;
compute after;
line ' ';
line @25 'Profits in US dollars';
endcomp;
run;
n The total profits for the year are saved in the temporary variable TOTAL. This value is not
displayed on the table.
o A blank line is written after each year.
Z The percentage is calculated using the annual total n as the denominator.
Appendix 1: Exercise Solutions 395
Total profit per year
Separated by Product Line
Product
Year Groups
1999 Children
Clothes & Shoes
Outdoors
Sports Equipment
Annual
Profit
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
Product
Percentage
4.08%
31.57%
21.76%
42.60%
2000 Children
Clothes & Shoes
Outdoors
Sports Equipment
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
3.83%
31.00%
22.19%
42.98%
2001 Children
$538,462.06
3.91%
. . . Portions of the table are not shown . . .
2. Building on the solution to Exercise 1 in this section, add a summary line for each
year using a BREAK statement.
Do you need to do anything ‘extra’ for the percentage on this summary line?
What if you also used an RBREAK statement?
The following BREAK statement can be used.
break after year/ summarize suppress skip;
Total profit per year
Separated by Product Line
Product
Year Groups
1999 Children
Clothes & Shoes
Outdoors
Sports Equipment
Annual
Profit
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
$13,272,490.82
Product
Percentage
4.08%
31.57%
21.76%
42.60%
100.00%
2000 Children
Clothes & Shoes
Outdoors
Sports Equipment
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
$16,159,587.28
3.83%
31.00%
22.19%
42.98%
100.00%
2001 Children
$538,462.06
3.91%
. . . Portions of the table are not shown . . .
396 Carpenter’s Complete Guide to the SAS REPORT Procedure
The percentage on the report-wide summary line will be calculated incorrectly if an
RBREAK statement is added.
break after year/ summarize suppress skip;
rbreak after / summarize;
. . . Portions of the table are not shown . . .
2002
Children
Clothes & Shoes
Outdoors
Sports Equipment
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
$15,898,930.35
4.52%
29.81%
23.30%
42.37%
100.00%
$59,085,048.46
371.63%
Profits in US dollars
The overall summary line shows a silly percentage. This is because the overall total profit
is being divided by TOTAL, which in this case is the total from 2002.
3. Building on the solution to Exercise 2 in this section, use the OUT= option to see the
final output data table.
proc report data=sashelp.orsales
out= yrdat nowd split='*';
column year product_line profit percent;
define year / group;
define product_line
. . . Portions of the code are not shown . . .
run;
title3 'Glimpse of the Output Data Table';
proc print data=yrdat;
run;
Appendix 1: Exercise Solutions 397
Here is the result of the PROC PRINT:
Total profit per year
Separated by Product Line
Glimpse of the Output Data Table
Obs
Year
1
2
3
4
5
6
7
8
9
10
1999
1999
1999
1999
1999
1999
2000
2000
2000
2000
. . .
Product_Line
Profit
13272490.82
541552.09
4189628.63
2887433.36
5653876.74
13272490.82
16159587.28
Children
618803.21
Clothes & Shoes
5009983.42
Outdoors
3585371.04
Portions of the table are not
Children
Clothes & Shoes
Outdoors
Sports
percent
.
0.04080
0.31566
0.21755
0.42598
1.00000
1.21752
0.03829
0.31003
0.22187
shown . .
_BREAK_
Year
Year
Year
.
A1.6 Solutions to Chapter 6 Exercises
1. Using the SASHELP.RETAIL quarterly sales data, for each year calculate the
following sales statistics: N, MEAN, SUM, and STDERR. Use aliases and a DEFINE
statement for each statistic. You might want to build on the results of Exercise 2 in
Chapter 3.
title1 'Yearly Statistics';
title2 'Based on Quarterly Sales';
proc report data=sashelp.retail nowd split='*';
column year sales sales=s_mean sales=s_n sales=s_se;
define year / group;
define sales / sum 'Total*Sales';
define s_mean / mean 'Average*Sales';
define s_n / n
'Quarters' f=8.;
define s_se / stderr 'Standard*Error';
run;
398 Carpenter’s Complete Guide to the SAS REPORT Procedure
Yearly Statistics
Based on Quarterly Sales
YEAR
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
Total
Sales
$1,030
$1,148
$1,252
$1,393
$1,528
$1,596
$1,922
$2,164
$2,412
$2,592
$2,734
$2,947
$3,204
$3,578
$1,874
Average
Sales
$258
$287
$313
$348
$382
$399
$481
$541
$603
$648
$684
$737
$801
$895
$937
Quarters
4
4
4
4
4
4
4
4
4
4
4
4
4
4
2
Standard
Error
$15
$16
$12
$18
$15
$23
$25
$23
$20
$18
$30
$24
$41
$49
$61
A very similar report could have been generated without the use of aliases.
title1 'Yearly Statistics';
title2 'Based on Quarterly Sales';
proc report data=sashelp.retail nowd split='*';
column year sales,(sum mean n stderr);
define year / group;
define sum / 'Total*Sales';
define mean / 'Average*Sales';
define n
/ 'Quarters' f=8.;
define stderr / 'Standard*Error';
run;
Appendix 1: Exercise Solutions 399
Yearly Statistics
Based on Quarterly Sales
YEAR
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
Retail sales in millions of $
Total
Average
Standard
Sales
Sales Quarters
Error
$1,030
$258
4
$15
$1,148
$287
4
$16
$1,252
$313
4
$12
$1,393
$348
4
$18
$1,528
$382
4
$15
$1,596
$399
4
$23
$1,922
$481
4
$25
$2,164
$541
4
$23
$2,412
$603
4
$20
$2,592
$648
4
$18
$2,734
$684
4
$30
$2,947
$737
4
$24
$3,204
$801
4
$41
$3,578
$895
4
$49
$1,874
$937
2
$61
2. Using the SASHELP.RETAIL quarterly sales data, for each year display the
SALES amount for each quarter (DATE) in a separate column (ACROSS). Why is a
format such as QTR. needed on DATE?
title1 'Quarterly Sales';
proc report data=sashelp.retail nowd;
column year date,sales;
define year / group;
define date / across f=qtr.;
define sales / sum;
run;
400 Carpenter’s Complete Guide to the SAS REPORT Procedure
Quarterly Sales
YEAR
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1
Retail
sales in
millions
of $
$220
$247
$284
$299
$342
$337
$419
$484
$546
$594
$606
$703
$692
$758
$876
DATE
2
3
Retail
Retail
sales in sales in
millions millions
of $
of $
$257
$258
$292
$286
$307
$318
$351
$359
$388
$385
$399
$412
$472
$490
$543
$542
$607
$616
$666
$662
$674
$705
$709
$728
$797
$826
$909
$920
$998
.
4
Retail
sales in
millions
of $
$295
$323
$343
$384
$413
$448
$541
$595
$643
$670
$749
$807
$889
$991
.
Because DATE does not repeat across years, dates tend not to work well as ACROSS
variables. We get around this by assigning a format that results in a value that does
repeat. In this case the QTR. format maps dates into a value of 1, 2, 3, or 4.
3. Building on the results of Exercise 2 in this section, add a spanning header for
DATE and modify the column labels. Is it necessary to nest SALES within DATE?
title1 'Quarterly Sales';
proc report data=sashelp.retail nowd;
column year date,sales;
define year / group;
define date / across f=qtr. '- Quarter -';
define sales / sum 'Sales';
run;
Appendix 1: Exercise Solutions 401
Quarterly Sales
ƒƒƒƒƒƒƒƒƒƒƒƒƒ Quarter ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
1
2
3
4
YEAR
Sales
Sales
Sales
Sales
1980
$220
$257
$258
$295
1981
$247
$292
$286
$323
1982
$284
$307
$318
$343
1983
$299
$351
$359
$384
1984
$342
$388
$385
$413
1985
$337
$399
$412
$448
1986
$419
$472
$490
$541
1987
$484
$543
$542
$595
1988
$546
$607
$616
$643
1989
$594
$666
$662
$670
1990
$606
$674
$705
$749
1991
$703
$709
$728
$807
1992
$692
$797
$826
$889
1993
$758
$909
$920
$991
1994
$876
$998
.
SALES are nested
within DATE so that
the sales values will
appear in the columns
under each quarter
number. This nesting
would not be
necessary if we had
not assigned a usage of
ACROSS to DATE.
4. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Create a separate report BY YEAR that shows
total PROFIT for each PRODUCT_LINE and PRODUCT_CATEGORY.
Turn off the BYLINE option and place the year in the title using the #BYVAL title
option.
options nobyline; n
title1 'Total Profit';
title2 '#byval1'; o
proc report data=sashelp.orsales nowd split='*';
by year; Z
column product_line product_category profit;
define product_line
/ group
'Product*Groups';
define product_category
/ group
'Product*Category';
define profit / analysis
sum format=dollar15.2
'Profit';
run;
n The BYLINE is suppressed.
o The #BYVAL1 option is replaced by the value of the first BY variable (TITLE).
Notice that the TITLE statement does not have to appear within the step that defines
the BY variable.
Z For the LISTING destination the BY statement generates a page for each value of
YEAR.
402 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here is the report for YEAR=2000:
Total Profit
2000
o Z
Product
Groups
Children
Clothes & Shoes
Outdoors
Sports
Product
Category
Children Sports
Clothes
Shoes
Outdoors
Assorted Sports Articles
Golf
Indoor Sports
Racket Sports
Running - Jogging
Swim Sports
Team Sports
Winter Sports
Profit
$618,803.21
$2,582,318.17
$2,427,665.25
$3,585,371.04
$2,799,503.47
$1,022,274.78
$414,078.13
$540,104.52
$626,185.56
$185,326.66
$262,704.77
$1,095,251.73
A1.7 Solutions to Chapter 7 Exercises
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Create a report that shows total PROFIT and
percentage of annual sales for each PRODUCT_LINE within each year. You might
want to build on the results of Exercise 1 in Chapter 5.
Include a summary line (BREAK) for each product line and a total summary line
(RBREAK). Use a compute block to change the percentage on the report summary to
missing.
proc format;
value $sequip
'Sports' = 'Sports Equipment';
run;
title1 'Total profit per year';
title2 'Separated by Product Line';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit percent;
define year
/ group;
define product_line
/ group
f=$sequip.
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
define percent/ computed 'Product*Percentage'
format=percent10.2;
Appendix 1: Exercise Solutions 403
break after year/ summarize suppress skip;
rbreak after / summarize;
compute before year;
total = profit.sum;
endcomp;
compute percent;
percent = profit.sum/total;
endcomp;
compute after;
percent = .; n
line ' ';
line @25 'Profits in US dollars';
endcomp;
run;
n For the report summary line (RBREAK), PERCENT is set to missing in the
COMPUTE AFTER block to eliminate the bogus calculation for PERCENT that took
place in the COMPUTE PERCENT block (overall profits divided by the total for the
last year—2002).
Total profit per year
Separated by Product Line
Product
Year Groups
1999 Children
Clothes & Shoes
Outdoors
Sports Equipment
Annual
Product
Profit Percentage
$541,552.09
4.08%
$4,189,628.63
31.57%
$2,887,433.36
21.76%
$5,653,876.74
42.60%
$13,272,490.82
100.00%
2000 Children
Clothes & Shoes
Outdoors
Sports Equipment
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
$16,159,587.28
3.83%
31.00%
22.19%
42.98%
100.00%
2001 Children
Clothes & Shoes
Outdoors
Sports Equipment
$538,462.06
$4,158,502.80
$3,222,898.41
$5,834,176.75
$13,754,040.02
3.91%
30.23%
23.43%
42.42%
100.00%
(continued)
404 Carpenter’s Complete Guide to the SAS REPORT Procedure
2002 Children
Clothes & Shoes
Outdoors
Sports Equipment
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
$15,898,930.35
4.52%
29.81%
23.30%
42.37%
100.00%
$59,085,048.46
.
Profits in US dollars
2. Building on the results of Exercise 1 in Chapter 7, add ‘Total’ as text in the YEAR
column for each of the annual summary lines, and ‘Overall’ on the report summary
line.
Use the OUT= option to examine the output data set, and note the values taken on
by the _BREAK_ variable.
proc report data=sashelp.orsales nowd
out=rptdat split='*';
column year yeartxt product_line profit percent;
define year / group noprint; o
define yeartxt/ f=$5. 'Year';
define product_line
/ group
f=$sequip.
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
define percent/ computed 'Product*Percentage'
format=percent10.2;
n
break after year/ summarize suppress skip;
rbreak after / summarize;
compute before year;
total = profit.sum;
endcomp;
compute percent;
percent = profit.sum/total;
endcomp;
compute yeartxt / char length=7; p
if _break_ = ' ' and year ne . then yeartxt = put(year,4.);
else if _break_ = ' ' and year = . then yeartxt = ' ';
else if _break_='_RBREAK_' then yeartxt = 'Overall';
else yeartxt = 'Total';
endcomp;
compute after;
Appendix 1: Exercise Solutions 405
percent = .;
line ' ';
line @25 'Profits in US dollars';
endcomp;
run;
n The computed variable YEARTXT is added to the COLUMN statement. This variable is
used to hold the year and the summary line text.
o The variable YEAR is used to form groups, but is not printed.
p Text is assigned to YEARTXT through the use of the _BREAK_ temporary variable.
Total profit per year
Separated by Product Line
Product
Groups
Children
Clothes & Shoes
Outdoors
Sports Equipment
Annual
Profit
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
$13,272,490.82
Product
Percentage
4.08%
31.57%
21.76%
42.60%
100.00%
Children
Clothes & Shoes
Outdoors
Sports Equipment
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
$16,159,587.28
3.83%
31.00%
22.19%
42.98%
100.00%
Children
Clothes & Shoes
Outdoors
Sports Equipment
$538,462.06
$4,158,502.80
$3,222,898.41
$5,834,176.75
$13,754,040.02
3.91%
30.23%
23.43%
42.42%
100.00%
Children
Clothes & Shoes
Outdoors
Sports Equipment
Total
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
$15,898,930.35
4.52%
29.81%
23.30%
42.37%
100.00%
Overall
$59,085,048.46
.
Year
1999
Total
2000
Total
2001
Total
2002
Profits in US dollars
406 Carpenter’s Complete Guide to the SAS REPORT Procedure
Here is the LISTING of the output data set corresponding to this table:
Total profit per year
Separated by Product Line
Obs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Year
yeartxt
1999 Total
1999 1999
1999
1999
1999
1999 Total
2000 Total
2000 2000
2000
2000
2000
2000 Total
2001 Total
2001 2001
2001
2001
2001
2001 Total
2002 Total
2002 2002
2002
2002
2002
2002 Total
. Overall
Product_Line
Children
Clothes & Shoes
Outdoors
Sports
Children
Clothes & Shoes
Outdoors
Sports
Children
Clothes & Shoes
Outdoors
Sports
Children
Clothes & Shoes
Outdoors
Sports
Profit
percent
_BREAK_
13272490.82
541552.09
4189628.63
2887433.36
5653876.74
13272490.82
16159587.28
618803.21
5009983.42
3585371.04
6945429.61
16159587.28
13754040.02
538462.06
4158502.80
3222898.41
5834176.75
13754040.02
15898930.35
718302.42
4739806.53
3704810.34
6736011.06
15898930.35
59085048.46
.
0.04080
0.31566
0.21755
0.42598
1.00000
1.21752
0.03829
0.31003
0.22187
0.42980
1.00000
0.85114
0.03915
0.30235
0.23432
0.42418
1.00000
1.15595
0.04518
0.29812
0.23302
0.42368
1.00000
.
Year
Year
Year
Year
Year
Year
Year
Year
_RBREAK_
3. Instead of using the RBREAK statement to calculate the overall total profit,
calculate the value in a compute block and display it using a LINE statement.
The COMPUTE AFTER block is used instead of an RBREAK statement. During the
processing of this block, the overall total profits are available in PROFIT.SUM.
title1 'Total profit per year';
title2 'Grand Total without RBREAK';
proc report data=sashelp.orsales nowd
split='*';
column year yeartxt product_line profit percent;
define year
/ group noprint;
define yeartxt/ f=$5. 'Year';
define product_line
/ group
f=$sequip.
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
Appendix 1: Exercise Solutions 407
define percent/ computed 'Product*Percentage'
format=percent10.2;
break after year/ summarize suppress skip;
compute before year;
total = profit.sum;
endcomp;
compute percent;
percent = profit.sum/total;
endcomp;
compute yeartxt / char length=7;
if _break_ = ' ' and year ne . then yeartxt = put(year,4.);
else if _break_ = ' ' and year = . then yeartxt = ' ';
else if _break_='_RBREAK_' then yeartxt = 'Overall';
else yeartxt = 'Total';
endcomp;
compute after;
line ' ';
line @5 'Total Profits in US dollars was ' profit.sum dollar14.2;
endcomp;
run;
The last few lines of the report show the overall summary line, which has been generated
with a LINE statement.
. . . Portions of the report are not shown . . .
Sports Equipment
$5,834,176.75
42.42%
Total
$13,754,040.02
100.00%
2002
Total
Children
Clothes & Shoes
Outdoors
Sports Equipment
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
$15,898,930.35
4.52%
29.81%
23.30%
42.37%
100.00%
Total Profits in US dollars was $59,085,048.46
408 Carpenter’s Complete Guide to the SAS REPORT Procedure
A1.8 Solutions to Chapter 8 Exercises
The problems for the Chapter 8 exercises utilize the program E8_0Basic.sas. It generates the
following report for the LISTING destination.
Total profit per year
Separated by Product Line
Profit Summaries
Product
Year Groups
1999 Children
Clothes & Shoes
Outdoors
Sports
Annual
Profit
$541,552.09
$4,189,628.63
$2,887,433.36
$5,653,876.74
$13,272,490.82
2000 Children
Clothes & Shoes
Outdoors
Sports
$618,803.21
$5,009,983.42
$3,585,371.04
$6,945,429.61
$16,159,587.28
2001 Children
Clothes & Shoes
Outdoors
Sports
$538,462.06
$4,158,502.80
$3,222,898.41
$5,834,176.75
$13,754,040.02
2002 Children
Clothes & Shoes
Outdoors
Sports
$718,302.42
$4,739,806.53
$3,704,810.34
$6,736,011.06
$15,898,930.35
$59,085,048.46
1. The data table SASHELP.ORSALES contains sales data from a retail outdoor
sports clothing and equipment store. Using the BREAK and RBREAK statements
summarize across product lines and years. You can build on the results of Exercise
1 in Chapter 3 or start with program E8_0Basic.sas .
Using the STYLE= option, change attributes on the following:
column headers
row header
Appendix 1: Exercise Solutions 409
data values
something on a summary row
Experiment with different components on different PROC REPORT step
statements.
The following solution writes to the HTML destination.
options center;
ods listing close;
ods html style=default
path="&path\results"
body='E8_1.html';
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Profit Summaries';
proc report data=sashelp.orsales nowd split='*'
style(header)={background=white}
style(column)={background=pink};
column year product_line profit;
define year
/ group
style(header)={background=yellow}
style(column)={background=cyan};
define product_line
/ group
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress skip
style(summary)={background=green};
rbreak after / summarize
style(summary)={background=red};
run;
ods _all_ close;
410 Carpenter’s Complete Guide to the SAS REPORT Procedure
If the same component / attribute combination appears with different attribute
characteristics on both the DEFINE statement and the PROC REPORT statement,
which receives precedence?
Here is the top of the HTML report:
The colors specified for the
headers on the REPORT
statement have been overridden on
the DEFINE statement for
PRODUCT_LINE and PROFIT.
Attributes specified on the
DEFINE statement override those
specified on the REPORT
statement.
2. Repeat Exercise 1 in this section using the CALL DEFINE statement. Is the order
that the attributes are applied the same? Which portions of the report are not
controllable through the CALL DEFINE statement?
ods html style=default
path="&path\results"
body='E8_2.html';
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Profit Summaries';
proc report data=sashelp.orsales nowd split='*';
* Call define does not offer control of the header spaces
* style(header)={background=white}
* style(column)={background=pink};
column year product_line profit;
define year
/ group;
define product_line
/ group
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress skip;
rbreak after / summarize;
Appendix 1: Exercise Solutions 411
compute year;
call define(_col_,'style','style={background=cyan}');
endcomp;
compute product_line;
call define(_col_,'style','style={background=pink}');
endcomp;
compute after year;
call define(_row_,'style','style={background=green}');
endcomp;
compute after;
call define(_row_,'style','style={background=red}');
endcomp;
run;
ods _all_ close;
The attributes of the header areas are
controlled through the STYLE= option
rather than through the CALL
DEFINE routine. Notice that the order
of the application of the background
colors is different from that in Exercise
1 in this section. This is because the
CALL DEFINE routine is executed
during a compute block. STYLE= is
applied first and then potentially
overridden by the result of the CALL
DEFINE routine. _COL_ takes
precedence over _ROW_.
3. In the basic table from Exercise 1 in this section, add trafficlighting to alert the
reader to profits that are less than $1 million. Solve using both the STYLE= option
and the CALL DEFINE statement.
proc format;
value lowval
low - <1000000 = 'yellow';
run;
options center;
ods listing close;
ods html style=default
path="&path\results"
body='E8_3a.html';
412 Carpenter’s Complete Guide to the SAS REPORT Procedure
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Profit Summaries';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit;
define year
/ group;
define product_line
/ group
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit'
style(column)={background=lowval.};
break after year / summarize suppress skip;
rbreak after / summarize;
run;
ods _all_ close;
Apply the trafficlighting format using the CALL DEFINE routine by removing the
STYLE= option on the DEFINE statement and by adding a compute block with a CALL
DEFINE routine.
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress skip;
rbreak after / summarize;
Appendix 1: Exercise Solutions 413
compute profit;
call define(_col_,'style','style={background=lowval.}');
endcomp;
run;
4. In the basic table, create links for the ‘Sports’ product line that drill down to a
detail table that shows PRODUCT_CATEGORY within PRODUCT_LINE.
Extra Credit:
Make the links and detail tables year-specific.
The link to the individual product lines is built in the temporary variable LINK and then
established with CALL DEFINE.
ods html style=default
path="&path\results"
body='E8_4a.html';
title1 'Total profit per year';
title2 'Separated by Product Line';
title3 'Links to Sports Detail';
proc report data=sashelp.orsales nowd split='*';
column year product_line profit;
define year
/ group;
define product_line
/ group
'Product*Groups';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
break after year / summarize suppress skip;
rbreak after / summarize;
compute before year;
yr = year;
endcomp;
compute product_line;
* For SPORTS create a link;
if product_line = 'Sports' then do;
link = 'Sports'||trim(left(put(yr,4.)))||'.html';
call define(_col_, 'url', link);
end;
endcomp;
run;
ods _all_ close;
The preceding PROC REPORT step creates the following primary table. The
PRODUCT_LINE of ‘Sports’ is shown as a link to the detail report for that particular
year.
414 Carpenter’s Complete Guide to the SAS REPORT Procedure
The detail report for ‘Sport’ is created for each specific year. The following step
generates the detail table for YEAR=1999.
* Sports Detail Report for 1999;
ods html style=default
path="&path\results"
body='Sports1999.html';
title2 'Sports Detail for 1999';
title3 "Return to Full Report";
proc report data=sashelp.orsales(where=(product_line='Sports' &
year=1999))
nowd split='*';
column product_category profit;
define product_category
/ group
'Product*Category';
define profit / analysis
sum format=dollar15.2
'Annual*Profit';
rbreak after / summarize skip;
run;
ods _all_ close;
Appendix 1: Exercise Solutions 415
SEE ALSO
The hardcoded generation of the detail reports, as was done in the preceding example, is
not very practical in the production environment. The programs E8_4b.SAS and
E8_4c.SAS both use the macro language to automate the process.
5. Using the E8_0Basic.sas program, use three or more title statement options to
change text attributes.
The following TITLE statements change a few of the title attributes with TITLE
statement options.
title1 c=red h=2'Total profit per year';
title2 c=blue j=l 'Separated by' j=r color=green 'Product Line';
title3 f='times new roman' h=2 'Profit' h=4 ' Summaries';
416 Carpenter’s Complete Guide to the SAS REPORT Procedure
A p p e n d i x
2
Syntax and Example Index
A2.1 PROC REPORT Step 418
A2.1.1 Primary Statements 418
A2.1.2 PROC REPORT Statement Options 418
A2.1.3 BY Statement Options 419
A2.1.4 COLUMN Statement Options 420
A2.1.5 DEFINE Statement Options 420
A2.1.6 BREAK Statement Options 422
A2.1.7 RBREAK Statement Options 422
A2.1.8 COMPUTE Statement Options 422
A2.1.9 In the Compute Block 423
A2.1.10 Other PROC REPORT Step Statements 423
A2.2 Output Delivery System 423
A2.2.1 ODS Destinations 424
A2.2.2 ODS Statements and Options 424
A2.2.3 HTML Destination Options 425
A2.2.4 PDF Destination Options 425
A2.2.5 RTF Destination Options 426
A2.3 Attribute Control and Modification 426
A2.3.1 STYLE= Option 426
A2.3.2 CALL DEFINE Routine 427
A2.3.3 Attribute Modifiers 428
A2.4 System Options 429
418 Carpenter’s Complete Guide to the SAS REPORT Procedure
The following tables indicate where you can find various statements and options that relate to the
report process. These are not meant to be an exhaustive list, but rather an indication of possible
examples that contain the item of interest.
A2.1 PROC REPORT Step
A2.1.1 Primary Statements
These statements are used throughout the book.
Statement
Syntax
Discussion
Examples
PROC REPORT
1.1
Throughout the book
COLUMN
2.1
2.1.1, throughout the book
DEFINE
2.2
All examples following Section 2.2.
BREAK
3.2
3.2.1 through 3.2.4, 4.4.8, 5.3, 5.5, 7.9.1, 10.3.2,
11.2, 11.3.3, 11.3.4, A1.5.2, A1.7.1, A1.7.2
RBREAK
3.3
6.6.2, 7.1.3, 7.2.4, 7.2.5, 7.4. 7.4.2, 11.2, 11.3.3,
11.3.4, A1.5.2, A1.7.1, A1.7.2
COMPUTE
2.6
Especially Chapters 5 and 7, throughout the
remainder of the book
A2.1.2 PROC REPORT Statement Options
Many of these options are used infrequently or are applicable only to the LISTING destination.
Option
Syntax
Discussion
Examples
BOX
4.4.1
4.4.7, A1.4.2, A1.4.3
CENTER
4.4.2
COLWIDTH=
4.4.3
4.4.4, 4.4.5, 4.4.6, 4.5.1
COMPLETEROWS
6.4.2
6.4.3
COMPLETECOLS
6.4.2
CONTENTS=
8.5.5
DATA=
1.1
FORMCHAR
4.4.7
Throughout the book
(continued)
Appendix 2: Syntax and Example Index 419
A2.1.2 PROC REPORT Statement Options (continued)
Option
Syntax
Discussion
Examples
HEADLINE
4.1
A1.4.1
HEADSKIP
4.1
A1.4.1
LINESIZE=
4.4.6
4.4.8
LIST
6.5.3
MISSING
6.5.4
NAMED
6.5.2
NOHEADER
6.5.1
6.5.2
NOWD
1.1
Throughout the book
NOFS, NOWINDOWS
1.1
As the preferred option, NOWD is used
throughout the book
OUT=
5.5
7.9.3, A1.5.3
PAGESIZE=
4.4.6
PANELS=
4.4.4
4.4.5, 4.4.6, A1.4.2, A1.4.3
PSPACE=
4.4.5
A1.4.2, A1.4.3
SPLIT=
2.5.3, 9.3.3
4.3.2, 4.4.1, 4.4.4, 4.4.5, 4.4.6, 4.4.7, 4.5.1,
5.1, 5.2, 5.3, 9.3.1, A1.2.3, A1.5.3
WRAP
4.4.8
6.5.2
A2.1.3 BY Statement Options
Syntax
Discussion
Examples
BY group
processing
2.2.2
6.6.3, 8.6.2, 8.6.3 , A1.6.4
#BYVAL
6.6.3
8.6.2, 8.6.3 , A1.6.4
#BYVAR
6.6.3
Option
420 Carpenter’s Complete Guide to the SAS REPORT Procedure
A2.1.4 COLUMN Statement Options
Option
Syntax
Discussion
Examples
Alias
specification
6.2
6.3, 6.5.4, 7.2.3, 7.2.5, 7.3, 7.7.2, 7.9.1, 7.9.2, 8.8.1,
A1.6.1
(Grouping)
2.3.3
2.5.1, 4.3
Nesting
(comma)
2.3.1, 6.3
2.3.2, 2.3.3, 2.3.4, 6.1.6, 7.2.4, 7.5, A1.6.2, A1.6.3
Statistics
2.3.2
2.3.3, 2.3.4, 4.3, A1.3.2, A1.6.1
N
6.1.3, 6.5.1, 7.8.1, A1.3.2
Text
2.5.1
2.5.3, throughout the remainder of the book
Repeated text
4.3, 9.3.3
6.4, A1.4.1, A1.6.3
A2.1.5 DEFINE Statement Options
Option
Syntax
Discussion
ACROSS
2.2.4, 11.4
2.2, 2.2.4, 2.3.4, 2.4.5, 2.5.1, 4.1, throughout the
remainder of the book
ANALYSIS
2.2.1
2.2, 2.2.1, 2.3.3, throughout the remainder of the book
COMPUTED
Chapters 5
and 7
2.2, 5.1, 5.2, throughout the remainder of the book
DISPLAY
2.2.1
2.2, 2.2.1, occasionally throughout the book
GROUP
2.2.3
2.2, 2.2.3, 2.2.4, 2.3.3, 2.4.5, 2.5.1, throughout the
remainder of the book
ORDER
2.2.2
2.2, 2.2.2, 3.3.2, 5.1, 5.3, throughout the remainder of
the book
Examples
(continued)
Appendix 2: Syntax and Example Index 421
A2.1.5 DEFINE Statement Options (continued)
Option
Syntax
Discussion
Examples
Statistics
2.4.1
2.4.2, 2.4.3, A1.6.1
MEAN
2.4.1
2.4.2, throughout the remainder of the book
N
2.4.4
2.4.5, 7.5, 7.9.3, 11.3.4
STD
SUM
7.9.2, 7.9.3
2.4.1
A1.4.1, A1.6.2
DESCENDING
2.4.3
EXCLUSIVE
6.4.2
FLOW
4.5.2
4.5.3
FORMAT=
2.4.2
2.4.3, 2.4.5, 2.5.1, 5.2, 10.1.2, 10.1.4, A1.2.3, A1.6.2
ID
6.1.6
6.1.7
Justification
6.1.2
7.9.1
MISSING
6.1.3
NOPRINT
6.1.5
NOZERO
6.1.4
ORDER=
2.4.3
PAGE
6.1.7
PRELOADFMT
6.4.2
6.4.3
SPACING=
4.5.3
A1.4.3
Text
2.5.2
2.5.3, 5.1, 5.2, 8.6.1, A1.4.1
WIDTH=
4.5.1
4.5.2, 4.5.3, 5.1, A1.3.2, A1.4.3
7.4.3, 7.6.2, 7.7.2, 7.9.1, 7.9.3, 10.1.4, A1.7.2
10.1.2, 10.1.4
422 Carpenter’s Complete Guide to the SAS REPORT Procedure
A2.1.6 BREAK Statement Options
Option
Syntax
Discussion
Examples
DOL
4.2
5.3, 5.5
DUL
4.2
OL
4.2
A1.3.1
PAGE
3.2.4
7.9.1
SKIP
3.2.1
3.2.2, 3.2.3, 3.2.4, 4.4.8, 5.3, 5.5, 10.3.2, 11.2, 11.3.4,
A1.3.1, A1.7.1
SUMMARIZE
3.2.2
3.2.3, 3.2.4, 3.3.2, 3.3.3, 4.1, 5.3, 5.5, 7.9.1, 8.2.2, 8.3.1,
8.3.2, 8.3.4, A1.7.1, throughout the remainder of the
book
SUPPRESS
3.2.3
3.2.4, 5.3, 5.5, 8.2.2, 8.3.1, 8.3.2, 8.3.4, throughout the
remainder of the book
UL
4.2
A2.1.7 RBREAK Statement Options
Option
Syntax
Discussion
Examples
DOL
4.2
6.6.2, 7.2.4, 7.2.5, 7.4.2, 7.4.3, 7.4, A1.3.1
SUMMARIZE
3.3.1
3.3.2, 3.3.3, 6.6.2, 7.1.3, 7.2.4, 7.2.5, 7.4, 7.4.3, 8.2.2,
8.4.4, 8.5.1, 8.5.3, 8.5.5, A1.7.1, throughout the
remainder of the book
A2.1.8 COMPUTE Statement Options
Option
Syntax
Discussion
AFTER/BEFORE
2.7, 7.3
5.3, 5.4, Chapter 7, throughout the remainder of the
book
CHARACTER
7.6.2
7.9.2, 10.1.4, 10.3.1, A1.7.2
LEFT
7.6.1
8.2.2
LENGTH=
7.6.2
7.9.2, 10.1.4, 10.3.1, A1.7.2
_PAGE_
5.4
7.6.1, 8.2.2
Examples
Appendix 2: Syntax and Example Index 423
A2.1.9 In the Compute Block
Statement
Syntax
Discussion
_BREAK_
7.1.3, 7.2.5
7.3, 7.4.3, 7.7.3, 7.8.1, 8.4.5, 8.4.6, 8.5.3, 11.3.4,
11.4, A1.7.2
CALL DEFINE
7.5, 8.3
8.3, throughout the remainder of Chapter 8,
10.2.2, 10.4, A1.8.2, A1.8.3, A1.8.4
LINE
2.6.1
2.6.2, 2.6.3, 8.2.2, 8.6.1 through 8.6.6, 10.1.1,
10.1.2, 10.1.4, A1.2.3, A1.3.1, A1.5.1, A1.7.1,
A1.7.2
SAS language
elements
2.6.4, 7.7
5.1, 5.2, 5.3, 5.5, throughout Chapter 7
(especially Section 7.7), 8.3.3, 8.4.4, 8.4.5,
10.1.4, 11.3.1 through 11.3.4, 11.4, A1.7.1,
A1.7.2, A1.8.4
Examples
A2.1.10 Other PROC REPORT Step Statements
Statement
Syntax
Discussion
Examples
FREQ
6.7
7.9.4
A2.2 Output Delivery System
Since the focus of this book in not on the Output Delivery System, not all of the ODS options and
statements have been fully explained. This is especially true of the primary destination statements.
The reader is encouraged to read one of the texts on ODS, such as Haworth (2001a) and Gupta
(2003), for more information.
424 Carpenter’s Complete Guide to the SAS REPORT Procedure
A2.2.1 ODS Destinations
The LISTING destination is used implicitly in a majority of the examples in the first seven
chapters.
Destination
Examples
LISTING
Throughout the book
HTML
1.2, 2.5, 3.2.2, 3.3.1, 6.6.4, most examples in Chapter 8, 9.3,
10.5.2, A1.8.1, A1.8.3
HTML3
9.3.1
MARKUP
TAGSET=EXCELXP
9.3.1, 9.3.2
MSOFFICE2K
9.3.1
PDF
8.2.4, 8.5.1, 8.5.5, 8.6.1, 8.6.2, 8.6.3, 8.6.5, 8.9, 8.11, 9.11, 10.3.3,
10.5.3
RTF
8.5.6, 8.5.7, 8.6.2, 8.6.5, 8.6.6, 8.9, 9.1, 10.5.3
A2.2.2 ODS Statements and Options
(Nominally Destination Independent)
Although included in this table, these options do not necessarily have the same effect in all
destinations, and in some cases might not even be available for all destinations.
Option
Syntax
Discussion
Examples
ANCHOR=
8.5.3
8.5.5
BOOKMARKLIST=
8.5.5
COLUMNS=
8.9
ESCAPECHAR=
8.6
FILE=
8.6.1 through 8.6.5
1.2, 3.2.2, 3.3.1, 3.3.2, 8.2.3, 8.2.4, most ODS
examples except for the HTML destination
LAYOUT
10.5.1
LINK=
8.5.1
NEWFILE=
6.6.4
OPTIONS
9.3.1
PROCLABEL=
8.5.5
(continued)
Appendix 2: Syntax and Example Index 425
A2.2.2 ODS Statements and Options (continued)
Option
Syntax
Discussion
REGION
10.5.1
STARTPAGE=
8.5.5
Examples
8.10, 8.11
STYLE=
3.2.3, 3.3.1, 3.3.2, 8.2.1, 8.2.2, A1.8.1, most
ODS examples
TEXT=
8.10
8.11
Title statement options
8.7
Question 5 in the Chapter 8 exercises
A2.2.3 HTML Destination Options
Option
Syntax
Discussion
Examples
HTML tags
8.5.1
8.5.2, 8.5.7, 9.3.1, 9.3.3, A1.8.4
PATH=
8.2.1, 8.2.2, 8.2.3, 8.2.4, A1.8.1, most HTML examples
BODY=
8.2.1, 8.2.2, 8.2.3, 8.2.4, A1.8.1, most HTML examples
URL=none
8.5.1
8.5.2, 8.5.3
A2.2.4 PDF Destination Options
Option
Syntax
Discussion
AUTHOR=
9.2.1
KEYWORDS=
9.2.1
NOTOC
8.5.5
SUBJECT=
9.2.1
TITLE=
9.2.1
Examples
8.11
426 Carpenter’s Complete Guide to the SAS REPORT Procedure
A2.2.5 RTF Destination Options
Option
Syntax
Discussion
Examples
BODYTITLE
9.1.1
8.6.2, 8.6.5, 9.1.2, 9.1.4, 10.5.3
Raw RTF
codes
8.6.6
RTF control
words
9.1.2
A2.3 Attribute Control and Modification
A2.3.1 STYLE= Option
The primary introduction to the STYLE= attribute modification option is in Section 8.1.
Component
Discussion
Examples
COLUMN
8.1
8.2.1, 8.2.3, 8.2.4, 8.4.2, 8.4.4, 8.6.6, 8.11, 9.3.2, 10.4,
10.5.4, A1.8.1, A1.8.3
CALLDEF
8.1
HEADER
8.1
8.2.1, 8.5.4, 8.5.5, 8.5.6, 8.6.6, 8.8.2, A1.8.1
LINES
8.1
8.2.2, 8.2.4
REPORT
8.1
8.2.1, 8.2.3
SUMMARY
8.1
8.2.1, 8.4.2, 8.4.4, A1.8.1
Appendix 2: Syntax and Example Index 427
A2.3.2 CALL DEFINE Routine
Initial discussions of the CALL DEFINE routine appear in Sections 7.5 and 8.3.
The first argument of the CALL DEFINE routine is the location.
Location
(column id)
Discussion
Examples
Report item
name
7.5
10.3.1
_COL_
7.5
8.4.3, 8.4.5, 8.5.3, 8.5.5, 8.5.7, 8.8.1, 8.9, 10.2.2,
A1.8.2, A1.8.3, A1.8.4
_Cxx_
7.5
7.5, 8.3.3, 8.4.5, 8.4.6, 8.4.7
_ROW_
7.5
8.3.1, 8.3.2, 8.4.6, 8.4.7, 10.2.2, 10.4
The second argument of the CALL DEFINE routine is the attribute type.
Attribute
Type
Discussion
Examples
FORMAT
7.5
7.5
GRSEG
10.3.1
STYLE
7.5, 8.3
8.3.1 through 8.3.3, 8.4.3, 8.4.5 through 8.4.7, 8.8.1,
10.2.2, 10.4, throughout the remainder of the book
URL
7.5
8.5.3, 8.5.5, 8.5.7, A1.8.4
428 Carpenter’s Complete Guide to the SAS REPORT Procedure
A2.3.3 Attribute Modifiers
These style attribute modifiers can be used to override the attributes specified in the ODS style.
Most can be used on both the STYLE= option and in the CALL DEFINE routine.
Attribute Name
Syntax
Discussion
BACKGROUND
8.1
8.2.1, 8.2.4, 8.3.1 through 8.3.3, 8.4.1 through
8.4.7, 10.2.2, A1.8.1, A1.8.2, A1.8.3
BORDERCOLOR
8.1
8.2.4
CELLHEIGHT
8.1
CELLWIDTH
8.1
CELLPADDING
8.1
CELLSPACING
8.2.4
HTMLSTYLE
8.8.1
8.8.3
FLYOVER
8.8.1
8.8.2, 8.8.3, 8.9, 10.2.2
FONT
8.2.3
FONT_FACE
8.1
8.2.1, 8.2.2, 8.3.1, 8.3.2, 8.6.4, 8.6.6
FONT_SIZE
8.1
8.2.1, 8.2.2, 8.2.3, 8.2.4
FONT_STYLE
Examples
8.2.4, 8.11, 10.1.3, 10.5.4
8.6.4
FONT_WEIGHT
8.1
8.2.1, 8.2.2, 8.3.1, 8.3.2, 8.6.4, 8.6.6
FOREGROUND
8.1
8.2.1, 8.3.3, 8.4.1 through 8.4.7
HEADERFIXED
8.6.4
HEADERSTRONG
8.6.4
HTMLSTYLE
10.4
JUST
8.1
POSTIMAGE
10.3.1
PREIMAGE
8.2.2
RIGHTMARGIN
8.1
8.11
TAGATTR
9.3.2
10.4
URL
8.5.4
8.5.5, 8.5.6
8.2.1, 8.2.2, 10.1.3
Appendix 2: Syntax and Example Index 429
A2.4 System Options
Option
Syntax
Discussion
CENTER
1.1, 4.4.2,
7.6.1
LABEL
2.0
LEFTMARGIN
9.1.4
NOBYLINE
TOPMARGIN
Examples
8.2.1, and in various examples throughout the book.
Most of the LISTING examples have this option set to
NOCENTER.
8.6.2, A1.6.4
9.1.4
430 Carpenter’s Complete Guide to the SAS REPORT Procedure
A p p e n d i x
3
Example Locator
A3.1 Combination Detail and Summary Reports 431
A3.1.1 Transposing Rows and Columns 432
A3.1.2 Specifying and Calculating Statistics 432
A3.1.3 Enhancing Tables 432
A3.1.4 Controlling Pages 433
A3.1.5 Controlling the Order of the Report’s Rows 433
A3.2 Calculating Percentages 433
A3.3 Processing Weighted Means and Totals 433
A3.4 Understanding Processing Phases and Event Sequencing 434
Most of the examples contained in this book are primarily designed to demonstrate a statement,
option, or technique. However, a number of these examples are also notable for their approach to
specific tasks. Whereas Appendix 2 enables you to locate examples that are specific to a particular
statement or option, this appendix is arranged according to tasks that you might need to
accomplish. Often additional examples of a given technique are highlighted in the MORE
INFORMATION section in the main part of the book or can be found by looking up specific
options and statements in Appendix 2.
A3.1 Combination Detail and Summary Reports
Most of the examples in this book contain various combinations of detail and summary
information. However, a number of techniques are associated with this type of report.
432 Carpenter’s Complete Guide to the SAS REPORT Procedure
A3.1.1 Transposing Rows and Columns
7.2.4
10.1
A2.1.5
Using absolute column addresses
Using vertically concatenated tables
Using the ACROSS option on the DEFINE statement (see the examples for this
option)
A3.1.2 Specifying and Calculating Statistics
2.3.2
2.4.1
2.4.5
5.2
7.9.4
Specifying statistics on the COLUMN statement (A2.1.4)
Specifying statistics on the DEFINE statement (A2.1.5)
Using the N statistic without an analysis variable
Performing calculations based on statistics
Calculating the weighted average on a summary line
A3.1.3 Enhancing Tables
3.2.1
3.3
6.6.3
7.2.1
7.4
7.6.1
7.7.1
7.7.2
7.7.3
7.8.1
7.9.1
7.9.2
8.2.2
8.3.2
8.4
8.5
8.6.5
8.7
8.8
10.1.3
10.3
10.5.2
Skipping lines between groups
Adding summary rows to detail reports
Using #BYVAL title options
Adding an OBS column
Changing the text associated with summary lines
Justifying text
Counting items (rows) within groups
Changing text associated with GROUP and ORDER variables
Counting pages and items across pages
Adding lines of text for group summaries
Creating report-wide summaries on each page
Combining multiple values into one field
Importing a graphic or logo into text
Shading table rows
Using trafficlighting effects
Creating linked tables
Controlling line breaks and wrapping
Using title and footnote statement options
Creating tip and flyover text
Aligning text and numbers
Importing graphics
Aligning text across reports
Appendix 3: Example Locator 433
A3.1.4 Controlling Pages
5.4
6.1.7
7.7.3
8.6.2
10.5
Forcing pages using _PAGE_
Using the PAGE option on the DEFINE statement to create vertical page breaks
Counting pages and items across pages
Using in-line formatting sequences to display page numbers
Placing multiple tables on the same page
A3.1.5 Controlling the Order of the Report’s Rows
2.2.2
6.4.3
10.1.2
Using the ORDER= option on the DEFINE statement (A2.1.5)
Using formats built with the NOTSORTED option
Using dummy variable values and formats to order rows
A3.2 Calculating Percentages
Percentage calculations require the use of a temporary variable to hold either the numerator, the
denominator, or both prior to the percentage calculation.
5.3
7.1.3
7.3
10.1
11.3.4
Calculating percentages within groups
Calculating percentages within groups
Calculating percentages within and across groups
Calculating group percentages in a vertically concatenated table
Calculating percentages using both BEFORE and AFTER summary lines
A3.3 Processing Weighted Means and Totals
Weighted statistics require additional processing. This processing can be accomplished using
specific statements as well as techniques.
6.7
7.9.4
Using the FREQ statement
Calculating the weighted average on a summary line
434 Carpenter’s Complete Guide to the SAS REPORT Procedure
A3.4 Understanding Processing Phases and Event
Sequencing
As your knowledge of the PROC REPORT process deepens, it becomes more and more important
to understand how PROC REPORT works. These sections are designed to help you learn the
processing phases of PROC REPORT.
1.4.2
2.7
7.1
11.1
Understanding the PROC REPORT step processing phases
Understanding compute block sequences
Understanding the PROC REPORT step processing phases
Reviewing the processing step sequence
Russ Lavery’s “An Animated Guide to the SAS REPORT Procedure,” which is included on the
CD that accompanies this book, is another excellent resource for information on event
sequencing.
References
Online Notes, Sample Programs, and FAQs by
SAS Institute
FAQs
The index for SAS Technical Support FAQs can be found at
http://support.sas.com/faq/
At this location you can search for FAQs by topic and number. All FAQs are published by SAS
Institute Inc., Cary, NC.
FAQ #4007: What release of Microsoft Word is recommended for reading RTF files created by
ODS?
FAQ #4010: In ODS RTF, can I get my page numbers in page X of Y format?
FAQ #4012: Just when I thought the ODS RTF destination was boring, you add animation.
(Demonstrates text animations in RTF.)
FAQ #4019: In ODS RTF, how can I generate HTML hyperlinks?
FAQ #4148: How do links work under PDF in ODS?
FAQ #4450: In SAS 9.1, are there easier ways to customize page numbers in RTF output?
FAQ #4473: How can I use PROC REPORT to link from one page in my PDF file to another?
Online Samples
The index for SAS online sample programs can be found at
http://support.sas.com/ctx/samples/index.jsp?sxf=adv
At this location you can search the library of sample programs by topic, title, or number. All
samples are published by SAS Institute Inc., Cary, NC.
Sample 607: Printing text prior to the headers on every page but the first page using
PROC REPORT
Sample 608: In PROC REPORT, COMPUTE statements for a COMPUTED variable under an
ACROSS when the number of columns is unknown
Sample 608: Printing CONTINUED or other text at the bottom of pages of PROC REPORT
output
Sample 611: Showing the ORDER or GROUP variable value on every row in PROC REPORT
436 Carpenter’s Complete Guide to the SAS REPORT Procedure
Sample 634: Wrapping values of a long variable onto multiple lines within the width of the
variable's column using PROC REPORT
Sample 635: Specify one or more ID variables to print on each page in PROC REPORT
Sample 637: Wrapping observations that are too long to fit in PROC REPORT
Sample 770: Using the SCAN Function
SAS Documentation
Although the following manuals provide a great deal of information about the REPORT
procedure and its statements and options, they each reflect the procedure in its SAS 6 operating
environment. This means that although the general information that they contain is still accurate,
they generally do not reflect the current terminologies or internal processes of PROC REPORT.
They also have no information about the Output Delivery System.
SAS Institute Inc. 1990. SAS Guide to the REPORT Procedure: Usage and Reference, Version 6.
1st ed. Cary, NC: SAS Institute Inc.
SAS Institute Inc. 1993. SAS Technical Report P-258: Using the REPORT Procedure in a
Nonwindowing Environment, Release 6.07. Cary, NC: SAS Institute Inc.
Although somewhat dated, this manual contains a wealth of pre-ODS information.
SAS Institute Inc. 1995. SAS Guide to the REPORT Procedure, Reference, Release 6.11. Cary,
NC: SAS Institute Inc.
This reference manual contains some of the same information as can be found in P-258,
although in a more abbreviated form.
Books, Papers, and Presentations
A number of the following papers have been included on the CD that accompanies this book.
Included papers are indicated by a bolding of the author’s name and the date. The PDF file name
(also bolded in the bibliographic reference) matches the citation in the book, and the name is also
a combination of the last name of the author(s) and the first year of publication. Unfortunately, it
was not possible to contact all of the authors, and as a consequence not all of the cited papers are
included on the CD.
Albarran, Luis F. 2003. “Creating “That Other” Laboratory Listing: Let PROC TRANSPOSE,
FREQ, CONTENTS and REPORT Do the Job for You.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2003), paper TU091. Cary,
NC: SAS Institute Inc.
Allen, Richard R. 2002. “Recovering Lost Lines: Using the Whole Page with Proc Report.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2002),
paper CC08. Cary, NC: SAS Institute Inc. (Allen_2002.pdf)
References 437
Baroud, Thaer, John W. Senner, and Paul W. Johnson. 2006. “Incorporating PROC
FORMAT, PROC REPORT, and ODS Style Definitions to Track Sources of Records
with Poor Data.” Proceedings of the Thirty-first Annual SAS Users Group International
Conference, paper 129-31. Cary, NC: SAS Institute Inc.
(Baroud_Senner_Johnson_2006.pdf)
®
Brown, David. 2005. “%sas2xl: A Flexible SAS Macro That Uses Tagsets to Produce Complex,
Multi-Tab Excel Spreadsheets with Custom Formatting.” Proceedings of the Thirtieth
Annual SAS Users Group International Conference, paper 092-30. Cary, NC: SAS
Institute Inc.
Bruns, Dan, Ray Pass, and Alan Eaton. 2002. “Battle of the Titans: REPORT vs TABULATE.”
Proceedings of the Twenty-seventh Annual SAS Users Group International Conference,
paper 133-27. Cary, NC: SAS Institute Inc. (Bruns_Pass_Eaton_2002.pdf)
Bruns, Dan and Ray Pass. 2004. “To REPORT or to TABULATE?—That is the Question!”
Proceedings of the Twenty-ninth Annual SAS Users Group International Conference,
paper 122-29. Cary, NC: SAS Institute Inc. (Bruns_Pass_2004.pdf)
Buck, Deborah Babcock. 1999. “So Many PROCs, So Little Time: A Primer of Efficient Report
®.
Writing in Base SAS ” Proceedings of the SouthEast SAS Users Group Conference
(SESUG 1999), paper 015. Cary, NC: SAS Institute Inc.
_______. 2004. “Here’s the Data, Here’s the Report I Want—How Do I Get There?” Proceedings
of the SouthEast SAS Users Group Conference (SESUG 2004), paper IN08. Cary, NC:
SAS Institute Inc.
Burlew, Michele M. 1998. SAS Macro Programming Made Easy. Cary, NC: SAS Institute Inc.
®
_______. 2005. SAS Guide to Report Writing: Examples. 2nd ed. Cary, NC: SAS Institute Inc.
Carpenter, Art. 1999. Annotate: Simply the Basics. Cary, NC: SAS Institute Inc.
®
_______. 2004a. Carpenter’s Complete Guide to the SAS Macro Language. 2nd ed. Cary, NC:
SAS Institute Inc.
Carpenter, Arthur L. 2004b. “Building and Using User Defined Formats.” Proceedings of the
Twenty-ninth Annual SAS Users Group International Conference, paper TU02. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Pharmaceutical SAS
Users Group Conference (PharmaSUG 2004), paper TU02. Cary, NC: SAS Institute Inc.
(Carpenter_2004b.pdf)
_______. 2005. “PROC REPORT Basics: Getting Started with the Primary Statements.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2005),
paper HW07. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Thirteenth Western Users of SAS Software Conference (WUSS 2005), paper how_procreport_basics. Cary, NC: SAS Institute Inc. (Carpenter_2005.pdf)
®
_______. 2006a. “Data Driven Annotations: An Introduction to SAS/GRAPH’s Annotate
Facility.” Proceedings of the Thirty-first SAS Users Group International Conference,
paper 108-31. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2006), paper HW03. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Fourteenth Western
Users of SAS Software Conference (WUSS 2006), paper HOW_Carpenter. Cary, NC:
SAS Institute Inc. (Carpenter_2006a.pdf)
438 Carpenter’s Complete Guide to the SAS REPORT Procedure
_______. 2006b. “In The Compute Block: Issues Associated with Using and Naming Variables.”
Proceedings of the Fourteenth Western Users of SAS Software Conference (WUSS 2006),
paper DPR-Carpenter. Cary, NC: SAS Institute Inc. (Carpenter_2006b.pdf)
_______. 2006c. “Advanced PROC REPORT: Traffic Lighting—Controlling Cell Attributes
With Your Data.” Proceedings of the Fourteenth Western Users of SAS Software
Conference (WUSS 2006), paper TUT-Carpenter. Cary, NC: SAS Institute Inc.
(Carpenter_2006c.pdf)
®
Carpenter, Arthur L. and Charles E. Shipp. 1995. Quick Results with SAS/GRAPH Software.
Cary, NC: SAS Institute Inc.
Carpenter, Arthur L. and Richard O. Smith. 2000. “Clinical Data Management: Building a
Dynamic Application.” Proceedings of the Pharmaceutical SAS Users Group
Conference (PharmaSUG 2000), paper DM10. Cary, NC: SAS Institute Inc. Also
published as “Data Management: Building a Dynamic Application” in the Proceedings
of the Tenth Western Users of SAS Software Conference (WUSS 2002), pp. 487–492.
Cary, NC: SAS Institute Inc. (Carpenter_Smith_2000.pdf)
_______. 2001. “Library and File Management: Building a Dynamic Application.” Proceedings
of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2001), paper DM02.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Ninth Western
Users of SAS Software Conference (WUSS 2001), pp. 266–272. Cary, NC: SAS Institute
Inc. Also published in the Proceedings of the Twenty-seventh Annual SAS Users Group
International Conference, paper 21-27. Cary, NC: SAS Institute Inc.
(Carpenter_Smith_2001.pdf)
_______. 2003a. “ODS and Web Enabled Device Drivers: Displaying and Controlling Large
Numbers of Graphs.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2003), paper TT026. Cary, NC: SAS Institute Inc. Also published in the
Proceedings of the Tenth Western Users of SAS Software Conference (WUSS 2002),
pp. 182–189. Cary, NC: SAS Institute Inc. (Carpenter_Smith_2003a.pdf)
_______. 2003b. “Controlling Graph Size: Building Thumbnails and GIF Files Using
®.
SAS/GRAPH ” Proceedings of the Eleventh Western Users of SAS Software Conference
(WUSS 2003). Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Twenty-ninth Annual SAS Users Group International Conference, paper 87-29. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Pharmaceutical SAS
Users Group Conference (PharmaSUG 2004), paper TT05. Cary, NC: SAS Institute Inc.
(Carpenter_Smith_2003b.pdf)
Casas, Angelina Cecilia. 2002. “Data Dependent Footnotes using PROC REPORT.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2002),
paper TT20. Cary, NC: SAS Institute Inc. (Casas_2002.pdf)
Chapman, David D. 2002. “Using PROC REPORT To Produce Tables With Cumulative Totals
and Row Differences.” Proceedings of the Twenty-seventh Annual SAS Users Group
International Conference, paper 120-27. Cary, NC: SAS Institute Inc.
_______. 2003. “Using Formats and Other Techniques to Complete PROC REPORT Tables.”
Proceedings of the Twenty-eighth Annual SAS Users Group International Conference,
paper 132-28. Cary, NC: SAS Institute Inc.
Chung, Chang Y. and Toby Dunn. 2005. “Page X of Y with PROC REPORT.” Proceedings of
the Pharmaceutical SAS Users Group Conference (PharmaSUG 2005), paper CC31.
Cary, NC: SAS Institute Inc. (Chung_Dunn_2005.pdf)
References 439
Cisternas, Miriam and Ricardo Cisternas. 2003. “Reading and Writing XML Files from
®
SAS .” Proceedings of the Eleventh Western Users of SAS Software Conference (WUSS
2003), paper i-reading_and_writing_xml. Cary, NC: SAS Institute Inc. Also published in
the Proceedings of the Twelfth Western Users of SAS Software Conference (WUSS 2004),
paper i_how_reading_and_writing_xm. Cary, NC: SAS Institute Inc.
(Cisternas_Cisternas_2003.pdf)
Cochran, Ben T. 2005. “A Gentle Introduction to the Powerful REPORT Procedure.”
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
259-30. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the SouthEast
SAS Users Group Conference (SESUG 2005), paper DP04-05. Cary, NC: SAS Institute
Inc.
®
Cody, Ronald P. 2004. SAS Functions by Example. Cary, NC: SAS Institute Inc.
Collins, Linda M. and Alan Hopkins. 2005. “Comparison of Techniques for XML and RTF
Tables.” Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG
2005), paper AD18. Cary, NC: SAS Institute Inc.
DeAngelis, Roger J. 2005. “Program Templates for Report Generation.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2005), paper PO15. Cary,
NC: SAS Institute Inc.
Delaney, Kevin P. 2003a. “ODS PDF: It’s Not Just for Printing Anymore.” Proceedings of the
Twenty-eighth Annual SAS Users Group International Conference, paper 146-28. Cary,
NC: SAS Institute Inc. (Delaney_2003a.pdf)
_______. 2003b. “Bookmarks, Links and It Looks Great Printed Too. Come In and See What
ODS PDF Can Do for You.” Proceedings of the SouthEast SAS Users Group Conference
(SESUG 2003), paper DP05. Cary, NC: SAS Institute Inc. (Delaney_2003b.pdf)
®
DelGobbo, Vincent. 2005. “Moving Data and Analytical Results between SAS and Microsoft
Office.” Proceedings of the Thirtieth Annual SAS Users Group International Conference,
paper 136-30. Cary, NC: SAS Institute Inc. (DelGobbo_2005.pdf)
_______. 2006. “Creating AND Importing Multi-Sheet Excel Workbooks the Easy Way with
®
SAS .” Proceedings of the Fourteenth Western Users of SAS Software Conference
(WUSS 2006), paper HOW-DelGobbo. Cary, NC: SAS Institute Inc. Also published in
the Proceedings of the Twenty-third Pacific Northwest SAS Users Group Conference
(PNWSUG 2006), paper DelGobbo-PNWSUG06. Cary, NC: SAS Institute Inc.
®
Delwiche, Lora D. and Susan J. Slaughter. 2003. The Little SAS Book: A Primer. 3rd ed. Cary,
NC: SAS Institute Inc.
Downing, Daniel K. 2004. “How To Get Ten Pounds Of Data Into A Five Pound Can – Using
Flyover Technique In HTML Output.” Proceedings of the Twelfth Western Users of SAS
Software Conference (WUSS 2004), paper c_dp_how_to_get_ten_pounds_o. Cary, NC:
SAS Institute Inc. (Downing_2004.pdf)
Dunn, Sharon. 2004. “Using Compute Blocks in PROC REPORT.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2004), paper CC07. Cary,
NC: SAS Institute Inc.
Ewing, Daphne and Ray Pass. 2000. “So Now You’re Using PROC REPORT. Is It Pretty and
Automated?” Proceedings of the Twenty-fifth Annual SAS Users Group International
Conference, paper 148-25. Cary, NC: SAS Institute Inc. Also published in the
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
244-30. Cary, NC: SAS Institute Inc. (Ewing_Pass_2000.pdf)
440 Carpenter’s Complete Guide to the SAS REPORT Procedure
®
Fairfield-Carter, Brian, Tracy Sherman, and Stephen Hunt. 2005. “Instant SAS Applications
With VBScript, Jscript, and dHTML.” Proceedings of the Thirtieth Annual SAS Users
Group International Conference, paper 009-30. Cary, NC: SAS Institute Inc.
Fan, Zizhong. 2005. “Make the Invisible Visible—A Case Study of Using ODS Inline Formatting
Style.” Proceedings of the Thirtieth Annual SAS Users Group International Conference,
paper 042-30. Cary, NC: SAS Institute Inc.
Feder, Steven. 2004. “Perfecting REPORT Output to RTF.” Proceedings of the Twenty-ninth
Annual SAS Users Group International Conference, paper 088-29. Cary, NC: SAS
Institute Inc.
®
_______. 2005. “Reporting and Interacting with Excel spreadsheets with ODS and DDE”,
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
026-30. Cary, NC: SAS Institute Inc.
Feng, Guanjun. 2006. “Font Color or Highlight? You Name It!” Proceedings of the Fourteenth
Western Users of SAS Software Conference (WUSS 2006), paper COD-Feng. Cary, NC:
SAS Institute Inc. (Feng_2006.pdf)
Flavin, Justina M. 1996. “Using PROC REPORT to Summarize Clinical Safety Data.”
Proceedings of the Fourth Western Users of SAS Software Conference (WUSS 1996),
pp125–127. Cary, NC: SAS Institute Inc. (Flavin_1996.pdf)
Foley, Malachy J. 2001. “PROC REPORT: How To Get Started.” Proceedings of the Southern
SAS Users Group Conference (SSU 2001), paper P-819. Cary, NC: SAS Institute Inc.
Foose, Diane E. 2002. “Report Macro - A Tool to Generate Flexible Summary Reports.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2002),
paper P06. Cary, NC: SAS Institute Inc. (Foose_2002.pdf)
Frey, Gerald D. 2005. “SAS Excels!” Proceedings of the Thirtieth Annual SAS Users Group
International Conference, paper 238-30. Cary, NC: SAS Institute Inc. Also published in
the Proceedings of the Sixteenth MidWest SAS Users Group Conference (MWSUG
2005), paper IS200. Cary, NC: SAS Institute Inc.
Gebhart, Eric A. 2005. “ODS MARKUP: The SAS® Reports You've Always Dreamed Of.”
Proceedings of the Thirtieth Annual SAS Users Group International Conference, Cary,
paper 085-30. NC: SAS Institute Inc. (Gebhart_2005.pdf)
_______. 2006. “The Beginners Guide to ODS MARKUP: Don't Panic!” Proceedings of the
Fourteenth Western Users of SAS Software Conference (WUSS 2006), paper TUTGephart. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Twentythird Pacific Northwest SAS Users Group Conference (PNWSUG 2006), paper GephartPNWSUG06. Cary, NC: SAS Institute Inc.
Gianneschi, Dennis. 2006. “General Methods to Use Special Characters.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2006), paper CC06. Cary,
NC: SAS Institute Inc.
Gilbert, Jeffery D. 2005. “Web Reporting Using the ODS.” Proceedings of the Thirtieth Annual
SAS Users Group International Conference, paper 095-30. Cary, NC: SAS Institute Inc.
(Gilbert_2005.pdf)
Girgis, Hanaa M. 2006 “Custom Template for Reporting Interim Analyses Using PROC
REPORT and Output Delivery System (ODS).” Proceedings of the Thirty-first Annual
SAS Users Group International Conference, paper 052-31. Cary, NC: SAS Institute Inc.
(Girgis_2006.pdf)
References 441
Godard, Peter and Cyndi Williamson. 2004. “Using Microsoft Excel® for Data Presentation.”
Proceedings of the Twelfth Western Users of SAS Software Conference (WUSS 2004),
paper i_dp_using_microsoft_excel_f. Cary, NC: SAS Institute Inc.
Godard_Williamson_2004.pdf)
Gonzalez, Sharon. 2003. “Using Compute Blocks in Proc Report.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2003), paper CC101. Cary,
NC: SAS Institute Inc.
Gupta, Sunil. 2003. Quick Results with the Output Delivery System. Cary, NC: SAS Institute Inc.
Guttadauro, Lara E. H. 2003. “Automated Page Breaking and Numbering For Output.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2003),
paper TT094. Cary, NC: SAS Institute Inc.
Gwet, Bikila bi. 2003. “Pivoting Data. An Alternative to ACROSS Variables of the REPORT
Procedure.” Proceedings of the SouthEast SAS Users Group Conference (SESUG 2003),
paper DP02. Cary, NC: SAS Institute Inc.
Hadden, Louise. 2006. “STOP! WAIT! GO!: See What Traffic-Lighting Can Do For You!”
Proceedings of the Thirty-first Annual SAS Users Group International Conference,
paper 142-31. Cary, NC: SAS Institute Inc. (Hadden_2006.pdf)
Hagendoorn, Michiel, Jonathan Squire, and Johnny Tai. 2006. “Save Those Eyes: A QualityControl Utility for Checking RTF Output Immediately and Accurately.” Proceedings of
the Thirty-first Annual SAS Users Group International Conference, paper 066-31. Cary,
NC: SAS Institute Inc. (Hagendoorn_Squire_Tai_2006.pdf)
®
Hamilton, Jack. 2004. “Master/Detail Reporting in Base SAS .” Proceedings of the Twelfth
Western Users of SAS Software Conference (WUSS 2004), paper
i_dp_masterdetail_reporting. Cary, NC: SAS Institute Inc.
Hamilton, Paul. 2003. “ODS to RTF: Tips and Tricks.” Proceedings of the Twenty-eighth Annual
SAS Users Group International Conference, paper 024-28. Cary, NC: SAS Institute Inc.
Also published in the Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2002), paper FDA06. Cary, NC: SAS Institute Inc. Also published in the
Proceedings of the Pacific Northwest SAS Users Group Conference (PNWSUG 2005).
Cary, NC: SAS Institute Inc.
Harper, Jennifer and Lisa Pappas. 2005. “Coding with Compassion – Giving Everyone Access to
The Power to Know™.” Proceedings of the Thirteenth Western Users of SAS Software
Conference (WUSS 2005), paper dp_coding_with_compassion. Cary, NC: SAS Institute
Inc.
Haworth, Lauren E. 1999. PROC TABULATE by Example. Cary, NC: SAS Institute Inc.
_______. 2001a. Output Delivery System: The Basics. Cary, NC: SAS Institute Inc.
_______. 2001b. “ODS for PRINT, REPORT and TABULATE.” Proceedings of the Twentysixth Annual SAS Users Group International Conference, paper 003-26. Cary, NC: SAS
Institute Inc. Also published in the Proceedings of the Southern SAS Users Group
Conference (SSU 2001), paper p-828. Cary, NC: SAS Institute Inc.
(Haworth_2001b.pdf)
442 Carpenter’s Complete Guide to the SAS REPORT Procedure
®
_______. 2002. “SAS with Style: Creating your own ODS Style Template.” Proceedings of the
Twenty-seventh Annual SAS Users Group International Conference, paper 186-27. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Eleventh Western Users
of SAS Software Conference (WUSS 2003), paper i-sas_with_style. Cary, NC: SAS
Institute Inc. (Haworth_2002.pdf)
®
_______. 2003. “SAS Reporting 101: REPORT, TABULATE, ODS, and Microsoft Office.”
Proceedings of the Twenty-eighth Annual SAS Users Group International Conference,
paper 071-28. Cary, NC: SAS Institute Inc. (Haworth_2003.pdf)
®
_______. 2004. “SAS with Style: Creating your own ODS Style Template for RTF Output.”
Proceedings of the Twenty-ninth Annual SAS Users Group International Conference,
paper 125-29. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2004), paper HOW04.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Twelfth Western
Users of SAS Software Conference (WUSS 2004), paper i_how_sas_with_style_creatin.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Pacific Northwest
SAS Users Group Conference (PNWSUG 2004), Cary, NC: SAS Institute Inc.
(Haworth_2004.pdf)
_______. 2005a. “Applying Microsoft Word Styles to ODS RTF Output.” Proceedings of the
Thirtieth Annual SAS Users Group International Conference, paper 043-30. Cary, NC:
SAS Institute Inc. (Haworth_2005a.pdf)
®
_______. 2005b. “SAS with Style: Creating your own ODS Style Template for PDF Output.”
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
132-30. Cary, NC: SAS Institute Inc. (Haworth_2005b.pdf)
Hester, Wayne. 2006. “Teaching Your RTF Tagset to Do Clever Tricks.” Proceedings of the
Thirty-first Annual SAS Users Group International Conference, paper 067-31. Cary, NC:
SAS Institute Inc. (Hester_2006.pdf)
Hull, Bob. 2001. “Now There Is an Easy Way to Get to Word, Just Use PROC TEMPLATE,
PROC REPORT, and ODS RTF.” Proceedings of the Twenty-sixth Annual SAS Users
Group International Conference, paper 163-26. Cary, NC: SAS Institute Inc.
(Hull_2001.pdf)
Humphreys, Suzanne M. 2006. “Page Breaks: Simple and Effective Ways to Neatly Present
Data.” Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG
2006), paper CC04. Cary, NC: SAS Institute Inc.
Huntley, Scott. 2006. “Let the ODS PRINTER Statement Take Your Output into the TwentyFirst Century.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2006), paper SA03. Cary, NC: SAS Institute Inc. (Huntley_2006.pdf)
Izard, David C. and David C. Chen. 2005. “A Technique for Producing Sorted Columns with a
Hanging First Line Using PROC REPORT.” Proceedings of the Pharmaceutical SAS
Users Group Conference (PharmaSUG 2005), paper CC16. Cary, NC: SAS Institute Inc.
Jiang, Songtao and Daniel Boisvert. 2006. “Effective Strategy to Set Page Breaks for ODS RTF
®
Output.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2006), paper TT12. Cary, NC: SAS Institute Inc.
Kumar, Sanjaya. 2006. “Show Me the Big Picture.” Proceedings of the Thirty-first Annual SAS
Users Group International Conference, paper 093-31. Cary, NC: SAS Institute Inc.
(Kumar_2006.pdf)
References 443
Karunasundera, Tikiri. 2006. “Creating and Modifying PDF Bookmarks.” Proceedings of the
Fourteenth Western Users of SAS Software Conference (WUSS 2006), paper DPRKarunasundera. Cary, NC: SAS Institute Inc.
Lafler, Kirk Paul. 2005. “Creating HTML Output with Output Delivery System.” Proceedings
of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2005), paper CC07.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Twelfth Western
Users of SAS Software Conference (WUSS 2004), paper c_cc_creating_html_output_wi.
Cary, NC: SAS Institute Inc. (Lafler_2005.pdf)
Lavery, Russell. 2003. “An Animated Guide©: Proc Report: The File behind the Scenes.”
Proceedings of the SouthEast SAS Users Group Conference (SESUG 2003), paper TU08.
Cary, NC: SAS Institute Inc. (Lavery_2003.pdf)
LeBouton, Kimberly Stinson. 2004. “Getting Up to Speed with PROC REPORT.” Proceedings
of the Twenty-ninth Annual SAS Users Group International Conference, paper 242-29.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Twenty-eighth
Annual SAS Users Group International Conference, paper 070-28. Cary, NC: SAS
Institute Inc. Also published in the Proceedings of the Eleventh Western Users of SAS
Software Conference (WUSS 2003), paper i-getting_up_to_speed_with_proc_report.
Cary, NC: SAS Institute Inc. Also published in the Proceedings of the Twelfth Western
Users of SAS Software Conference (WUSS 2004), paper i_tut_getting_up_to_speed_wi.
Cary, NC: SAS Institute Inc. (LeBouton_2004.pdf)
Leprince, Daniel J. and Elizabeth Li. 2003. “A Plot and a Table per Page Times Hundreds in a
Single PDF File.” Proceedings of the Twenty-eighth Annual SAS Users Group
International Conference, paper 141-28. Cary, NC: SAS Institute Inc.
(Leprince_li_2003.pdf)
Levin, Lois. 2005. “PROC FORMAT - Not Just Another Pretty Face.” Proceedings of the
Thirtieth Annual SAS Users Group International Conference, paper 001-30. Cary, NC:
SAS Institute Inc.
®
Li, Don (Dongguang). 2006. “Create Descriptive Documentation with Hyper-links for SAS
Database.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2006), paper CC12. Cary, NC: SAS Institute Inc.
®
Li, Na. 2006. “Using SAS to Manage and Report Long Text Fields in a Clinical DBMS.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2006),
paper TT15. Cary, NC: SAS Institute Inc. (Li_Na_2006.pdf)
Lund, Pete. 2005. “PDF Can Be Pretty Darn Fancy: Tips and Tricks for the ODS PDF
Destination.” Proceedings of the Pacific Northwest SAS Users Group Conference
(PNWSUG 2005). Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Thirty-first Annual SAS Users Group International Conference, paper 092-31. Cary,
NC: SAS Institute Inc. (Lund_2005.pdf)
Ma, J. Meimei and Sandra Schlotzhauer. 2000. “Fast Track to PROC REPORT Results.”
Proceedings of the Twenty-fifth Annual SAS Users Group International Conference,
paper 67-25. Cary, NC: SAS Institute Inc.
Ma, J. Meimei, Sandra Schlotzhauer, and Maria Ilieva. 2002. “Quick Results in PROC
REPORT.” Proceedings of the Twenty-seventh Annual SAS Users Group International
Conference, paper 59-27. Cary, NC: SAS Institute Inc.
444 Carpenter’s Complete Guide to the SAS REPORT Procedure
McNeill, Sandy. 2002. “What's New in the Output Delivery System, Version 9.0.” Proceedings
of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2002), paper SAS03.
Cary, NC: SAS Institute Inc. (McNeill_2002.pdf)
McQuown, Gary. 2004. “Exceptional Exception Reports.” Proceedings of the SouthEast SAS
Users Group Conference (SESUG 2004), paper AD11. Cary, NC: SAS Institute Inc.
(McQuown_2004.pdf)
Michel, Denis. 2005. “CALL EXECUTE: A Powerful Data Management Tool.” Proceedings of
the Thirtieth Annual SAS Users Group International Conference, paper 027-30. Cary,
NC: SAS Institute Inc. (Michel_2005.pdf)
Mitchell, Rick M. 2005. “Stranded on a Deserted Island With Nothing But TITLE Statements.”
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
069-30. Cary, NC: SAS Institute Inc. (Mitchell_2005.pdf)
_______. 2006. “Really Cool Graphics With SAS/GRAPH -Plotting and Embedding Likert
Scales in Your Reports.” Proceedings of the Thirty-first Annual SAS Users Group
International Conference, paper 086-31. Cary, NC: SAS Institute Inc.
(Mitchell_2006.pdf)
Molter, Mike. 2006. “The REPORT Procedure’s Temporary Variable: What is it? Why Do I
Care?” Proceedings of the Thirty-first Annual SAS Users Group International
Conference, paper 060-31. Cary, NC: SAS Institute Inc. (Molter_2006.pdf)
Morgan, Derek. 2006. “Inline Formatting + Long Character Variables = ‘I Didn't Know You
®
Could Do That with SAS !’” Proceedings of the Thirty-first Annual SAS Users Group
International Conference, paper 051-31. Cary, NC: SAS Institute Inc.
(Morgan_2006.pdf)
®
Okerson, Barbara B. 2004. “Evaluating Hospital Performance: Using SAS ODS to Create a
Hospital Scorecard.” Proceedings of the Twenty-ninth Annual SAS Users Group
International Conference, paper 157-29. Cary, NC: SAS Institute Inc.
Osowski, Scott and Thomas Fritchey. 2006. “Hyperlinks and Bookmarks with ODS RTF.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2006),
paper TT21. Cary, NC: SAS Institute Inc. (Osowski_Fritchey_2006.pdf)
®
Pagé, Jacques. 2004. “Automated Distribution of SAS Results.” Proceedings of the Twenty-ninth
Annual SAS Users Group International Conference, paper 039-29. Cary, NC: SAS
Institute Inc.
Parker, Chevell. 2003. “Generating Custom Excel Spreadsheets using ODS.” Proceedings of the
Pharmaceutical SAS Users Group Conference (PharmaSUG 2003), paper SAS129. Cary,
NC: SAS Institute Inc. (Parker_2003.pdf)
Parker, Susan. 2005. “A Microsoft Word Macro for Post-Processing ODS RTF Files.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2005),
paper PO38. Cary, NC: SAS Institute Inc.
®
Parsons, Lori S. 2005. “Using SAS 9 ODS Features to Present Table and Graph Data in an
Adobe PDF File.” Proceedings of the Thirtieth Annual SAS Users Group International
Conference, paper 172-30. Cary, NC: SAS Institute Inc. (Parsons_2005.pdf)
_______. 2006. “Enhancing RTF Output with RTF Control Words and In-Line Formatting.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2006),
paper PO04. Cary, NC: SAS Institute Inc. (Parsons_2006.pdf)
References 445
Pass, Ray. 2000. “PROC REPORT - Land of the Missing OBS Column.” Proceedings of the
Twenty-fifth Annual SAS Users Group International Conference, paper 105-25. Cary,
NC: SAS Institute Inc. (Pass_2000.pdf)
Pass, Ray and Daphne Ewing. 2000. “So You’re Still Not Using PROC REPORT. Why Not?”
Proceedings of the Twenty-fifth Annual SAS Users Group International Conference,
paper 147-25. Cary, NC: SAS Institute Inc. Also published in the Proceedings of the
Twenty-sixth Annual SAS Users Group International Conference, paper 149-26. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Twenty-eighth Annual
SAS Users Group International Conference, paper 196-28. Cary, NC: SAS Institute Inc.
Also published with Ewing as the lead author in the Proceedings of the Pharmaceutical
SAS Users Group Conference (PharmaSUG 2004), paper HW01. Cary, NC: SAS
Institute Inc. Also published in the Proceedings of the Thirty-first Annual SAS Users
Group International Conference, paper 235-31. Cary, NC: SAS Institute Inc.
(Pass_Ewing_2000.pdf)
Pass, Ray and Sandy McNeill. 2003. “PROC REPORT: Doin’ It in Style!” Proceedings of the
Twenty-eighth Annual SAS Users Group International Conference, paper 015-28. Cary,
NC: SAS Institute Inc. Also published in the Proceedings of the Twenty-seventh Annual
SAS Users Group International Conference, paper 187-27. Cary, NC: SAS Institute Inc.
Also published in the Proceedings of the Thirty-first Annual SAS Users Group
International Conference, paper 116-31. Cary, NC: SAS Institute Inc. Also published in
the Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG
2006), paper SA04. Cary, NC: SAS Institute Inc.
Peterson, Donald W. and Taghi R. Garacani. 2005. “User-Specified Text Flow Inside the
Report Procedure.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2005), paper PO17. Cary, NC: SAS Institute Inc.
(Peterson_Garacani_2005.pdf)
Peterson, Donald W. and John R. Gerlach. 1999. “User-Specified Text Flow Inside the Report
Procedure.” Proceedings of the Twenty-fourth Annual SAS Users Group International
Conference, paper 78. Cary, NC: SAS Institute Inc. Also published in the Proceedings of
the SouthEast SAS Users Group Conference (SESUG 1999), paper 031. Cary, NC: SAS
Institute Inc.
Ping, Fan and Richard Schiefelbein. 2006. “Dynamic Column Breaking Macro and Other
Applications.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2006), paper CC16. Cary, NC: SAS Institute Inc.
Poppe, Frank. 2005. “Creating Complicated Word Documents from an Intranet Application.”
Proceedings of the Thirtieth Annual SAS Users Group International Conference, paper
093-30. Cary, NC: SAS Institute Inc. (Poppe_2005.pdf)
®
Qi, Eric and Liping Zhang. 2003. “%RTFTable—A Powerful SAS Tool to Produce Rich Text
Format Tables.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2003), paper TT038. Cary, NC: SAS Institute Inc. (Qi_Zhang_2003.pdf)
Rafee, Dana. 1999. “Batch Proc Report.” Destiny Corporation, Wethersfield, CT.
(www.destinycorp.com)
Rhodes, Phil. 2005. “Automating the Drudgery Away: Using Macros and ODS to Produce
(Almost) Complete Reports.” Proceedings of the Thirtieth Annual SAS Users Group
International Conference, paper 011-30. Cary, NC: SAS Institute Inc.
446 Carpenter’s Complete Guide to the SAS REPORT Procedure
Rohowsky, Nestor. 2005. “Reporting Statistical Results with PROC REPORT.” Proceedings of
the Thirtieth Annual SAS Users Group International Conference, paper 036-30. Cary,
NC: SAS Institute Inc. (Rohowsky_2005.pdf)
Rowell, Richard and Jim Lenihan. 2004. “More Customization?: Creating Symbols In RTF Files
Using ODS.” Proceedings of the Pharmaceutical SAS Users Group Conference
(PharmaSUG 2004), paper TT12. Cary, NC: SAS Institute Inc. Also published in the
Proceedings of the Eleventh Western Users of SAS Software Conference (WUSS 2003),
paper c-more_customization. Cary, NC: SAS Institute Inc.
Sevick, Carter. 2006. “‘Table 1’ in Scientific Manuscripts: Using PROC REPORT and the ODS
System.” Proceedings of the Fourteenth Western Users of SAS Software Conference
(WUSS 2006), papers POS-Sevick and TUT-Sevick. Cary, NC: SAS Institute Inc.
Shah, Varsha C. 2003. “Queries by Ods Beginners.” Proceedings of the Pharmaceutical SAS
Users Group Conference (PharmaSUG 2003), paper CC060. Cary, NC: SAS Institute
Inc.
Shannon, David. 2002. “To ODS RTF and Beyond.” Proceedings of the Twenty-seventh Annual
SAS Users Group International Conference, paper 001-27. Cary, NC: SAS Institute Inc.
(Shannon_2002.pdf)
Smith, Curtis A. 2003. “Creating Output Data Sets from Almost Any Procedure Using the
Output Delivery System.” Proceedings of the Eleventh Western Users of SAS Software
Conference (WUSS 2003), paper c-creating_output_data_sets_from_almost_any_
procedure. Cary, NC: SAS Institute Inc. (Smith_2003.pdf)
Smith, Edward R. 2000. “Table Generation Using the PROC REPORT Feature.” Proceedings of
the Pharmaceutical SAS Users Group Conference (PharmaSUG 2000), paper DM07.
Cary, NC: SAS Institute Inc.
Smith, Richard O. and Arthur L. Carpenter. 2004. “Connecting Your Output: Using Drilldown Techniques to Link Reports, Graphics, and Tables.” Proceedings of the Twelfth
Western Users of SAS Software Conference (WUSS 2004), paper i_post_connecting_
your_outpu. Cary, NC: SAS Institute Inc. (Smith_Carpenter_2004.pdf)
Smoak, Carey G. 2004. “Creating Word Tables using PROC REPORT and ODS RTF.”
Proceedings of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2004),
paper TT02. Cary, NC: SAS Institute Inc. (Smoak_2004.pdf)
Squire, Jonathan. 2003. “Hot Links: Creating Embedded URLs using ODS.” Proceedings of the
Twenty-eighth Annual SAS Users Group International Conference, paper 023-28. Cary,
NC: SAS Institute Inc.
Squire, Jonathan and Johnny Tai. 2005. “It’s the Lines Per Page That Counts” Proceedings of the
Thirteenth Western Users of SAS Software Conference (WUSS 2005), paper cc_its the
lines per page. Cary, NC: SAS Institute Inc.
Stewart, Larry and Marje Fecht. 2002. “Tips and Tricks for Easier Reporting.” Proceedings of the
Twenty-seventh Annual SAS Users Group International Conference, paper 184-27. Cary,
NC: SAS Institute Inc.
Stroupe, Cindy. 2002. “PROC REPORT and ODS.” Proceedings of the Pharmaceutical SAS
Users Group Conference (PharmaSUG 2002), paper CC15. Cary, NC: SAS Institute Inc.
References 447
Taylor, Liz. 2005. “Generating Microsoft Word Macros that Automate the Organization and
Maintenance of SAS Tables, Listings and Figures.” Proceedings of the Pharmaceutical
SAS Users Group Conference (PharmaSUG 2005), paper CC14. Cary, NC: SAS Institute
Inc.
Thornton, Patrick. 2006. “Essential ODS Techniques for Creating Reports in PDF.”
Proceedings of the Fourteenth Western Users of SAS Software Conference (WUSS 2006),
paper COD-Thornton#1. Cary, NC: SAS Institute Inc. (Thornton_2006.pdf)
Trenery, David. 1999. “Jazzing up Your Reports—Some Tricks with PROC REPORT.”
Proceedings of the Twenty-fourth Annual SAS Users Group International Conference,
paper 80-24. Cary, NC: SAS Institute Inc.
Tsykalov, Eugene and Shi-Tao Yeh. 2006. “The Invisible Character Alt-255 – Hidden Dragon:
®
Hiding And Aligning Text in SAS Output.” Proceedings of the Thirty-first Annual SAS
Users Group International Conference, paper 055-31. Cary, NC: SAS Institute Inc.
(Tsykalov_Yeh_2006.pdf)
Wade, David and Shi-Tao Yeh. 2000. “Generating Case Report Form Tabulation.” Proceedings
of the Twenty-fifth Annual SAS Users Group International Conference, paper 30-25.
Cary, NC: SAS Institute Inc. (Wade_Yeh_2000.pdf)
Watts, Perry. 2002. Multiple-Plot Displays: Simplified with Macros. Cary, NC: SAS Institute Inc.
Whitlock, Ian. 2000. “Macro Design Considerations for a Word Wrapping Macro.” Proceedings
of the SouthEast SAS Users Group Conference (SESUG 2000), paper p-411. Cary, NC:
SAS Institute Inc.
®
®
Yeh, Eugene. 2004. “A SAS Database Search Engine Gives Everyone the Power to Know .”
Proceedings of the Twenty-ninth Annual SAS Users Group International Conference,
paper 19-29. Cary, NC: SAS Institute Inc.
Young, Walter R. 2003. “Creating a Compact Columnar Output with PROC REPORT.”
Proceedings of the SouthEast SAS Users Group Conference (SESUG 2003), paper TU16.
Cary, NC: SAS Institute Inc.
Yu, Hsiwei and Dong-Min Shen. 2006. “Dynamic Drill Down on Numeric Data.” Proceedings
of the Pharmaceutical SAS Users Group Conference (PharmaSUG 2006), paper TT08.
Cary, NC: SAS Institute Inc. (Yu_Shen_2006.pdf)
Zhang, Lei. 2005. “Constructing Stack Tables With Proc Report and ODS RTF.” Proceedings of
the Pharmaceutical SAS Users Group Conference (PharmaSUG 2005), paper AD16.
Cary, NC: SAS Institute Inc. (Zhang_2005.pdf)
Zhang, Lei and George Li. 2006. “Constructing Stack Tables with Magic %GluePlus (Another
User-Defined Macro for Creating Stack Tables).” Proceedings of the Thirty-first Annual
SAS Users Group International Conference, paper 089-31. Cary, NC: SAS Institute Inc.
448 Carpenter’s Complete Guide to the SAS REPORT Procedure
Index
A
absolute column numbers 161–164
absolute column references 161–164
ACROSS variables 18
CALL DEFINE routine and 177
columns and 155
combining values with 204–206
define type of 24–25, 34, 110, 155, 204
define usage of 26, 150, 161, 378–381
derived columns and 161
DESCENDING option 110
GROUP variables and 25, 36
missing values and 113
nesting 122, 204–206
nesting procedure code 36
ORDER=DATA option 35
preloading formats for 126, 129
repeat characters in 124
reversing order of columns 110–111
setup phase and 150
statistics and 29–30, 37–39
AFTER option, BREAK statement 58–60, 166,
200
AFTER option, COMPUTE statement
153–154, 166–169
adding lines of text 47–49
calculating weighted mean 208
coordinating statements 98
counting items across page breaks 190–191
_PAGE_ option and 103–104
paging issues 145
report row phase and 150
sequencing step events 52–54
summary lines and 375–378
syntax 45–46
writing formatted values 49–50
AFTER option, RBREAK statement 58
aliases
for variables 121–122
in COLUMN statement 155, 160–161
specifying statistics 31
aligning
columns 337
columns across reports 365–366
numbers and text in columns 335–338
analysis statistic 31–33
ANALYSIS variables 17
compound variable names 159
concatenating reports 30
define type of 18, 58, 156–157
missing values and 115–118
nesting 122
numeric variables and 156–157
statistics and 22, 31–33, 36–37, 99, 155
totaling variables 22
ANCHOR= option 264
anchor tags
See HTML anchor tags
annotate facility (SAS/GRAPH) 355–357
arrays, and compute blocks 51, 182
assignment statements
compute blocks and 51, 182–183
LINE statement execution 197–198
asterisk (*) 44, 79, 194
at sign (@) 48
attributes
See also style attributes
See also specific attributes
cell 216–219, 355
conditional assignment of 231–232
setting for tagsets 322–323
text 216–219
AUTHOR= option, ODS PDF statement
314–315
automating reports 341–345
B
BACKGROUND attribute 215
backslash (\) 79
BColor= option, TITLE/FOOTNOTE statements
292
BEFORE option, BREAK statement 58, 200,
202
BEFORE option, COMPUTE statement 150,
153–154, 166–169
adding lines of text 47–49
adding repeat characters 195
coordinating statements 98
counting items across page breaks 190–191
GROUP variables and 102
_PAGE_ option and 103–104, 202
paging issues 145
report row phase and 150
sequencing step events 52–54
summary lines and 375–378
450
Index
BEFORE option, COMPUTE statement
(continued)
syntax 45–46
writing formatted values 49–50
BEFORE option, RBREAK statement 58
blank lines
BREAK/RBREAK statement options 78
HEADSKIP option, REPORT procedure
76–77
inserting 46–47
LISTING destination 78–79
BMI (Body Mass Index) 156–159
BODYTITLE option, ODS RTF statement
311–312
BOLD option, TITLE/FOOTNOTE statements
292
BOOKMARKLIST= option, ODS PDF
statement 264
BORDERCOLOR attribute 215
BOX option 83–84
%BQUOTE macro function 291
_BREAK_ variable 164–165, 173–174, 193
BREAK statement 59
See also SKIP option, BREAK statement
AFTER option 58–60, 166, 200
BEFORE option 58, 200, 202
blank lines 78
_BREAK_ automatic variable and 164–166
BY groups and 138
combining summary reports with detail
reports 68–69
compute blocks process and 150
define types and 18
DOL/DUL options 59, 78
generating pages for group levels 67–68
grouping variables and 58–59, 65, 72
in detail reports 70–71
OL option 59, 78
order of applying options 59
PAGE option 58–59, 67–68, 145
paging issues 145
sequencing step events 52, 54
skipping lines between groups 59–61, 102
STYLE= option 213–215
SUMMARIZE option 54, 58–59, 61–67,
169, 201
summarizing across groups 61–65
summary reports and 71–72
SUPPRESS option 58, 169
suppressing summarization label 65–67
syntax 57–58
UL option 59, 78
BY group processing 136–143
BY groups
BREAK statement and 138
GROUP variables and 138
ODS and 141–143, 145
ODS destinations and 141, 145
ORDER variables and 138
RBREAK statement and 138–139
BY statement
DESCENDING option 34, 136
NOTSORTED option 136–137
ODS destinations and 137
sorting multiple variables 20
sorting with 21
SUM statement and 8
TITLE/FOOTNOTE statements and
139–141
#BYLINE option 140
#BYVAL option 139–141
#BYVAR option 139–141
C
CALL DEFINE routine 174–179
ACROSS variables and 177
adding logos to reports 222
changing style attributes 227–232
COLUMN statement and 178
COMPUTE statement and 178
conditional assignment of attributes
231–232
establishing links with 257–260
flyover text and 293–295
FORMAT attribute 175
importing graphics with 345–352
in reports 228–229
LISTING destination and 175
ODS destinations and 175
STYLE= option 175, 213–215
trafficlighting effects 233, 236–249
URL attribute 175, 257–260
URLBP attribute 175
URLP attribute 175
CALL SYMPUT routine 345
CALLDEF component, STYLE= option
214–215
CATS function 203
cell attributes 216–219, 355
CELLHEIGHT attribute 215, 346
CELLPADDING attribute 215
CELLSPACING attribute 225–226
Index 451
CELLWIDTH attribute 224
aligning columns across reports 365–366
component support 215
graphics and 346
ODS destination support 33
RIGHTMARGIN attribute and 301–303
CENTER option
COMPUTE statement 179–180, 222
DEFINE statement 110–112
PROC REPORT statement 85, 88
CGM file type 346
character formats 181
CHARACTER option, COMPUTE statement
174
creating character variables 180–182
example 203
repeating variables in rows 186
character variables
creating with CHARACTER option
180–182
creating with LENGTH= option 180–182
default define types for 18
LINE statement execution 197
classification variables
ACROSS define type 204
missing values for 113–118, 134–136
code
nesting 36
raw destination-specific 289–291
_COL_ keyword 174
colon (:) 79
Color= option, TITLE/FOOTNOTE statements
292
COLUMN component, STYLE= option
214–215
column headers 5, 14
column numbers 161–164
column references 160–164
COLUMN statement 16–17
adding report headers 41
adding text via 41–43
aliases in 155, 160–161
attaching statistics with comma 26–28
CALL DEFINE routine and 178
compound variable names 159
compute blocks process and 150–154
creating vertically concatenated tables 326
DEFINE display attribute and 59
DEFINE statement and 16, 18–19, 38,
98–99
derived columns and 161
dummy columns 173
forming groups 28–29
FREQ statement and 144
grouping variables and 70
N statistic 36–37, 193–194
nested associations 26
nesting statistics 29–30, 155
NOHEADER option and 131
nonprinting variables and 118
ordering columns 20
processing phases and 11
repeat characters in spanning headers
80–82
report row phase and 150
report variables in 10
sequencing step events 53
syntax 4
temporary variables in 10, 157
variable aliases 121
column width
adjusting 85–86
scaling 78
specifying 93, 358
columns
ACROSS variables and 155
adding space between 94–95
aligning 337
aligning across reports 365–366
calculating in setup phase 150
combining values in 202–204
concatenating 204
designating in LINE statement 48
dummy 173–174
identification 119–120
justification for 111–112, 358
modifying using compute blocks 97–106
number alignment in 335–338
OBS column 158
ordering 20
referencing in compute blocks 154–165
reversing order of 110–111
spacing 358
specifying multiple 298–299
text alignment in 335–338
trafficlighting effects 240–245
COLUMNS= option, ODS statement 298–299
COLWIDTH option 33, 85–86, 93
comma (,) 26–28, 205
comments, in compute blocks 182
COMPLETECOLS option 127, 129
COMPLETEROWS option 127–129
452
Index
compound variable names 159–161
COMPUTE AFTER statement 103
COMPUTE BEFORE _PAGE_ technique 202
compute blocks
See also COMPUTE statement
adding lines of text 47–49, 170–171
arrays and 51, 182
assignment statements and 51, 182–183
CALL DEFINE routine 174–179
comments in 182
conditional processing and 46, 51
DATA step and 182–183
DEFINE statement and 45
DO loops and 51, 183
examples of common tasks 199–209
functions and 182
GROUP variables and 154
grouping variables in 169–174
HTML anchor tags 255–257
IF-THEN/ELSE processing and 46, 51, 183
inserting blank lines 46–47
LINE statement and 154, 191–199
logic in 182–191
macro variables and 51
modifying columns using 97–106
multiple 373–374
nonprinting variables and 118
process overview 148–154, 372–378
referencing items in 154–165
report variables in 10
SAS language elements and 51–52,
182–191
sequencing step events 52–54
SUM statement and 183–185
summary lines and 374–375
syntax 45–46
temporary variables in 10, 152, 155,
157–158, 183–185
writing formatted values 49–50
COMPUTE statement
See also AFTER option, COMPUTE
statement
See also BEFORE option, COMPUTE
statement
See also compute blocks
See also _PAGE_ option, COMPUTE
statement
CALL DEFINE routine and 178
CENTER option 179–180, 222
CHARACTER option 174, 180–182, 186,
203
column justification 358
FORMAT= option and 181
LEFT option 179–180, 222
LENGTH= option 174, 180–182, 186, 203
options and switches 179–182
RIGHT option 179, 222
STYLE= option 213–215
syntax 4
computed summary information 10
COMPUTED variables 18, 155, 203
concatenating
CATS function and 203
columns 204
reports 8, 30
tables vertically 326–341
conditional assignment of attributes 231–232
conditional processing, and compute blocks 46,
51
CTRL key 266
D
{DAGGER} function 277–278
daggers, generating 277–278
dash (-) 79
data lines, wrapping 91–92
DATA _NULL_ step vs. REPORT procedure
8–9
DATA= option 4, 14
DATA step
compute blocks and 182–183
HTML anchor tags 255–257
LINE statement execution 197
outdated terminology 10
DATA Variable Table (DVT) 10
debugging 134
default margins, settings 315–316
DEFINE CALL routine
See CALL DEFINE routine
DEFINE display attribute 59
DEFINE statement 17–25
See also FORMAT= option, DEFINE
statement
See also ORDER= option, DEFINE
statement
adding report headers 41
adding text via 43–44
aliases as column references 160
analysis statistic 31–33
assigning define types 10
assigning define usage 10
associating statistics with 37–39
Index 453
CELLWIDTH attribute 33
CENTER option 110–112
character formats and 181
column justification 358
COLUMN statement and 16, 18–19, 38,
98–99
compute blocks and 45
DESCENDING option 21–22, 34–35,
110–111
EXCLUSIVE option 127–130
FLOW option 92–94, 282
formatting values 33
grouping variables 59
ID option 92, 110, 119–120
LEFT option 110–112
LISTING destination and 92–95
MISSING option 110, 113–115, 134–136
N statistic 36–37
nesting variables 122–123
NOHEADER option and 131
NOPRINT option 110, 118, 161, 205
NOZERO option 110, 115–118, 161
options 110–120
ordering displayed values 34–36
PAGE option 110, 119–120
PRELOADFMT option 127, 129–131
repeat characters in 80–81
RIGHT option 110–112
SPACING= option 92, 94–95, 126, 358
specifying statistics 31
statement options 92–95
STYLE= option 213–215
syntax 4
user-defined formats 123–125
variable aliases 121–122
WIDTH= option 33, 92–93
define types 10, 17–18
ACROSS variables 24–25, 34, 110, 155,
204
ANALYSIS variables 18, 58, 156–157
DISPLAY variables 18, 23, 59, 68, 99, 155
GROUP variables 22–25, 34, 110, 155
ORDER variables 34, 110, 155
RBREAK statement and 18
define usage 10, 17–18
ACROSS variables 26, 150, 161, 378–381
DISPLAY variables 27, 58
ORDER variables 19–22
derived columns 161
DESCENDING option, BY statement 34, 136
DESCENDING option, DEFINE statement 110
changing display order with 110–111
example 34–35
GROUP define type and 22
ORDER define usage and 21
detail reports
BREAK statement in 70–71
combining with summary reports 68–69
defined 9, 18
DISPLAY variables and 59
RBREAK statement in 69–71
support for 8
device drivers 347–351
DEVICE (DEV) option 346
direct variable names and references 156–159
DISPLAY define type 23
Display Manager 83
DISPLAY variables 17
combining summary reports with detail
reports 68
compound variable names 159
define type of 18, 59, 68, 99, 155
define usage of 27, 58
detail reports and 59
direct variable name references 156–159
GROUP variables and 23
summary reports and 59
DO loops
automation with macro language 266–268
compute blocks and 51, 183
LINE statement execution 197–198
DOL option
BREAK statement 59, 78
RBREAK statement 59, 78
DROP statement 182
DUL option
BREAK statement 59, 78
RBREAK statement 59, 78
dummy columns 173–174
DVT (DATA Variable Table) 10
E
EMF file type 346–347
ENDCOMP statement 45
equal sign (=) 79–81
error text, inserting 287–289
escape character 270–291
evaluation phase 10–11, 149, 369
EXCELXP tagset 321–322
EXCLUSIVE option, DEFINE statement
127–130
exporting reports to Microsoft Excel 316–322
454
Index
F
fields, combining values in 202–204
file descriptors, PDF 314–315
final report output 10
FLOW option, DEFINE statement 92–94, 282
FLYOVER attribute 293–297
flyover text 293–297
Font= option, TITLE/FOOTNOTE statements
292
FONT_FACE attribute 215
FONT_SIZE attribute 215
FONT_STYLE attribute 215
FONT_WEIGHT attribute 215
fonts
monospace-only 77–78, 84, 357–360
proportional 48–49
FOOTNOTE statement
BColor= option 292
BOLD option 292
BY statement and 139–141
Color= option 292
Font= option 292
Height= option 292
ITALIC option 292
Justify= option 292
LINK= option 254
ODS destinations 292–293
UNDERLINE option 292
footnotes 103, 250–255, 279–282
FOREGROUND attribute 215
FORMAT attribute, CALL DEFINE routine
175
FORMAT= option, DEFINE statement
COMPUTE statement options and 181
example 38, 176
formatting values 33, 172
PRELOADFMT option and 127
STYLE= option and 234
FORMAT procedure 123–125, 130–131,
233–234
VALUE statement 124–125, 130–131
FORMAT statement 33, 123
formats
building links with 268–269
character formats 181
in-line 270–291
in PROC REPORT statement 123–131
ordering by definition 130–131
preloading 126–131
STYLE= option and 234–236
user-defined 123–126, 268–269
variable names in LINE statement 194
formatted values 33, 49–50, 171–172
formatting
missing values and 171–172
S={ } formatting string 279–282, 297
trafficlighting effects 233–234, 249
variable names in LINE statement 194
with escape character 270
FORMCHAR option 84, 89–91
FORMCHAR system option 84, 89–91
FREQ statement 144–145, 206–207
PROC REPORT statement and 144–145
SUMMARY procedure and 144
functions, and compute blocks 182
G
GETOPTION function 196
GIF file type 346–347
GOPTIONS procedure 351–352
XPIXELS option 351
YPIXELS option 351
GPLOT procedure 354
GPRINT procedure 352–357
grand totals 200–202
graphics
CELLWIDTH attribute and 346
coordinating 345–357
importing 345–352
greater than (>) 79, 81
GREPLAY procedure 352–357
GROUP define type 22
GROUP variables 17
ACROSS variables and 25, 36
BEFORE option, COMPUTE statement
102
BY groups and 138
combining summary reports with detail
reports 68
compute blocks and 154
define type of 22–25, 34, 110, 155
DESCENDING option 110
DISPLAY variables and 23
forming groups 136
missing values and 113–114
nesting statistics 29–30
ordering values and 111
preloading formats 126
referencing by name 154
repeating on each row 185–187
report composition and 30
reversing order of columns 110–111
Index 455
grouping variables
BREAK statement and 58–59, 65, 72
COLUMN statement and 70
creating vertically concatenated tables
332–335
DEFINE statement and 59
generating pages for each level of 67–68
in compute blocks 169–174
parentheses for 41
preloading formats and 126
RBREAK statement and 69
summary reports and 72
groups
BY statement processing 136–143
calculating percentages within 101–103,
166
creating summaries for 192–194
forming with GROUP variables 136
forming with parentheses 28–29, 41–42
generating pages for each level 67–68
nested 24
skipping lines between 59–61, 64–65, 102
summarizing across 61–65
GRSEG attribute 345–348, 350, 352
H
HDR component, STYLE= option 214–215
HEADER component, STYLE= option
214–215, 296
headers
adding report headers 41
column 5, 14
removing 131–132
spanning 80–82, 194–196
HEADLINE option 359
adding repeat characters 194–195
LISTING destination 76–77
HEADSKIP option 359
adding repeat characters 194–195
blank lines 76–77
LISTING destination 76–77
Height= option, TITLE/FOOTNOTE statements
292
horizontal spaces 223–227
HTML anchor tags
as data values 255–257
building links with formats 269
IF-THEN/ELSE processing and 255
linking with 250–255
repeat characters and 323–324
HTML destination 316–324, 362–363
adding text to reports 40
BY groups and 141
column justification 112
dagger symbol 278
flyover text for 293–297
formatting with escape character 270
importing graphics to 347, 350
linking with CALL DEFINE routine
257–260
page break determination 104
page breaks in 187
paging issues 145
passing raw codes 289
skipping lines between groups 64–65
SUMMARIZE option, BREAK statement
66–67
superscripts/subscripts 272
HTMLSTYLE= attribute 295
hyperlinks, embedding in tables 249–269
hyphen (-) 79
I
ID option, DEFINE statement 92, 110,
119–120
IF-THEN/ELSE processing
compute blocks and 46, 51, 183
conditional assignment of attributes
231–232
HTML anchor tags and 255
LINE statement execution 197–198
importing graphics 345–352
in-line formats 270–291
%INCLUDE statement 51, 183
indentations, and line breaks 282–284
indirect column references 161
interactive windowing environment 4
ITALIC option, TITLE/FOOTNOTE statements
292
italics 290
J
JPEG file type 346–347
JUST= attribute
adding logos to reports 222
column alignment 337
column justification 358
STYLE= option and 215, 219, 222
justification
column 111–112, 358
LINE statement text 179–180
456
Index
Justify= option, TITLE/FOOTNOTE statements
292
K
KEEP statement 182
KEYWORDS= option, ODS PDF statement
314–315
L
labels 65–67
{LASTPAGE} function 273–277
LEFT option
COMPUTE statement 179–180, 222
DEFINE statement 110–112
LENGTH= option, COMPUTE statement 174
creating character variables 180–182
example 203
repeating variables in rows 186
less than (<) 79, 81
line breaks
indentations and 282–284
wrapping and 282–289
LINE statement
adding lines of text 47–49
adding repeated characters 194–196
compute blocks and 154, 191–199
COMPUTE AFTER statement and 103
creating group summaries 192–194
designating columns in 48
evaluation phase and 11
execution overview 197–199
formatting variable names in 194
inserting blank lines 46–47
justifying text 179–180
PUT statement and 46, 191, 194, 197
SUM statement and 193–194
writing formatted values 49–50
LINES component, STYLE= option 214–215
LINESIZE= system option 88, 145, 196
LINK= option, TITLE/FOOTNOTE statements
254
links
building with formats 268–269
embedding hyperlinks in tables 249–269
establishing with CALL DEFINE routine
257–260
for PDF destination 257, 261–265
for RTF destination 257, 261, 265–266
forming with STYLE= option 260–261
linking with HTML anchor tags 250–255
LIST option 134
LISTING destination 76
adding text to reports 39–41
as default 14
blank lines 78–79
BY groups and 141
BY statement and 137
CALL DEFINE routine attributes 175
column justification 111–112
column width 358
COMPUTE BEFORE _PAGE_ technique
202
counting items across page breaks 187–191
DEFINE statement options 92–95
FLOW option and 282
HEADLINE option 76–77
HEADSKIP option 76–77
NOHEADER option and 132
overlines 78–79
page control 358
page width for 196
paging issues 145
PROC REPORT statement options 83–92
repeat characters 79–83
SKIP option, BREAK statement 62, 64, 72
skipping lines between groups 59–60
SUMMARIZE option, BREAK statement
62, 66
text wrapping 358
underlines 78–79
user-defined formats 124
logical processing 182–191
logos, adding to reports 219–222
LS= option 88, 91
LS= system option 88, 196
M
macro language 266–268
macro variables
compute blocks and 51
quotation marks and 291
repeating characters in 196
resolution issues 343–345
margin marker 283–284
margins, default settings 315–316
MARKUP destination 145
MEAN statistic
compound variable names 159–160
computed variables and 203
example 39
REPORT procedure support 26
specifying 31
Index 457
MEANS procedure 8, 26, 144
MEANS/SUMMARY engine 11, 150
MEDIAN statistic 26, 31
MERGE statement 182
Microsoft Excel, exporting reports to 316–322
MISSING option, DEFINE statement 110,
113–115, 134–136
missing values
classification variables and 113–118,
134–136
formatting and 171–172
GROUP variables and 113–114
monospace-only fonts
SAS Monospace font 77–78, 84
workarounds for 357–360
MSOFFICE2K tagset 320–321
N
N statistic
COLUMN statement and 36–37, 193–194
DEFINE statement and 36–37
REPORT procedure support 26
requesting directly 203
specifying 31
without ANALYSIS variable 36–37
NAMED output 132–133
naming conventions 343
nesting
ACROSS variables 204–206
ANALYSIS variables 122
associations with comma 26, 205
groups 24
procedure code 36
statistics 29–30, 155
style attributes 281–282
variables 122–123
NEWFILE= option, ODS statement 142
NOCENTER option 85, 88
NOCENTER system option 179
NOFS option 4, 14
NOHEADER option 131–132
NOLABEL system option 5, 14
non-breaking spaces in text 284–287
nonprinting variables 118
NOPRINT option, DEFINE statement 110, 118
column numbers and 161
example 205
NOTSORTED option
BY statement 136–137
VALUE statement, FORMAT procedure
130–131
NOWD option 4, 14
NOWINDOWS option 4, 14
NOZERO option, DEFINE statement 110,
115–118, 161
number alignment in columns 335–338
numeric variables 18, 28
ANALYSIS variables and 156–157
O
OBS column, PRINT procedure 158
ODS (Output Delivery System)
adding text with TEXT= option 300
BY groups and 141–143, 145
changing style attributes 227–232
embedding hyperlinks in tables 249–269
escape character for formatting 270–291
RIGHTMARGIN attribute 301–303
specifying multiple columns 298–299
STYLE= option 213–227
trafficlighting effects 232–249
ODS COLUMNS= option 298–299
ODS CONTENTS= option 264–265
ODS destinations
See also HTML destination
See also PDF destination
See also RTF destination
BY groups and 141, 145
CALL DEFINE routine attributes 175
CELLWIDTH attribute support 33
default 14
importing graphics to 347
page breaks in 187
proportional fonts and 48
report row phase and 11
routing reports to 6–7
SKIP option, BREAK statement 60, 64
SUMMARIZE option, BREAK statement
66
TITLE/FOOTNOTE statements and
292–293
ODS ESCAPECHAR= option 270–271
ODS LAYOUT statement 352, 360–362
ODS PDF statement
AUTHOR= option 314–315
BOOKMARKLIST= option 264
KEYWORDS= option 314–315
STARTPAGE= option 264, 363–364
SUBJECT= option 314–315
TITLE= option 314–315
ODS PROCLABEL= option 264
458
Index
ODS RTF statement
BODYTITLE option 311–312
ODS sandwich 6–7
ODS statement
COLUMNS= option 298–299
NEWFILE= option 142
ODS STYLE attributes 214–215
OL option
BREAK statement 59, 78
RBREAK statement 59, 78
ORDER define usage 21
ORDER= option, DEFINE statement 34–36
determining column numbers 161, 164
GROUP define type and 22
ORDER=data 34–35, 162, 334
ORDER define usage and 21
ORDER=formatted 34, 163
ORDER=FREQ 34
ORDER=INTERNAL 34
preloading formats 127–131
ORDER variables 18
BY groups and 138
define type of 34, 110, 155
define usage overview 19–22
DESCENDING option 110
DISPLAY define type and 23
printing 19–21
repeating on each row 185–187
reversing order of columns 110–111
suppressing 88
ordering
columns 20
display of values 34–36
formats by definition 130–131
rows 19–22
values 111
OUT= option 104–106, 164–166, 168
Output Delivery System
See ODS
overlines 78–79
P
page breaks 104, 120, 187–191
page control 358
page numbers 273–277, 358
PAGE option
BREAK statement 58–59, 67–68, 145
DEFINE statement 110, 119–120
RBREAK statement 58–59, 145
_PAGE_ option, COMPUTE statement
adding logos to reports 221
AFTER option and 103–104
BEFORE option and 103–104, 202
breaks with BY groups 139
issues with 187
syntax 46
page width 196
{PAGEOF} function 273–277
pages, generating for group levels 67–68
PAGESIZE= system option 88, 145
paging issues 145
PANELS= option 86–87, 359
parentheses () 28–29, 41–42
&PATH macro variable 7
PATTERN statement 350
PDF destination 314–316
BY statement and 137
creating links for 257, 261–265
flyover text for 293–297
formatting with escape character 270
importing graphics to 347, 351
line breaks and wrapping 282
LINK= option and 254
page break determination 104
page breaks in 187
SKIP option, BREAK statement 60
specifying multiple columns 298–299
STARTPAGE=NEVER option 363–364
PDF file descriptors 314–315
PERCENT8.1 format 102
percentages, calculating within groups
101–103, 166
period (.) 79, 159
plus sign (+) 79
PNG file type 346–347
POSTIMAGE attribute 346
PREIMAGE attribute 219, 346
PRELOADFMT option, DEFINE statement
127, 129–131
preloading formats 126–131
PRINT procedure
as reporting tool 7
OBS column 158
REPORT procedure vs. 4–5, 8
SPLIT= option 44
STYLE= option 7
wrapping data lines 91–92
printing ORDER variables 19–21
PROC REPORT statement
See also REPORT procedure
additional options 131–136
BY GROUP processing 137–143
Index 459
creating vertically concatenated tables
338–341
DEFINE statement options 110–120
formats in 123–131
FREQ statement and 144–145
LISTING destination options 83–92
nesting variables 122–123
paging issues 145
variable aliases 121–122
procedure code, nesting 36
processing phases 11, 313–314, 367–381
proportional fonts 48–49
PROTECTSPECIALCHARS attribute 312
PS= option 88, 187–189
PS= system option 88
PSPACE= option 87
PUT function 173
PUT statement
designating columns in 48
LINE statement and 46, 191, 194, 197
NAMED output and 132
repeated strings and 194
Q
quotation marks 291, 345
R
raw destination-specific codes 289–291
RBREAK statement 69
AFTER option 58
BEFORE option 58
blank lines 78
_BREAK_ automatic variable and 164–166
BY groups and 138–139
compute blocks process and 150, 152
define types and 18
DOL/DUL options 59, 78
grouping variables and 69
in detail reports 69–71
OL option 59, 78
order of applying options 59
PAGE option 58–59, 145
paging issues 145
sequencing step events 52, 54
STYLE= option 213–215
SUMMARIZE option 54, 58–59, 169
summary reports and 69, 71–72
SUPPRESS option 152, 169
syntax 57–58
UL option 59, 78
RDV (Report Data Vector) 10
repeat characters
adding to spanning headers 80–82,
194–196
HTML tags and 323–324
in ACROSS variables 124
LISTING destination 79–83
SPLIT= option 82–83
REPEAT function 194, 196
REPORT component, STYLE= option
214–215
Report Data Vector (RDV) 10
REPORT procedure 14–16
as reporting tool 7
automating 341–345
coordinating graphics with 345–357
DATA _NULL_ step vs. 8–9
ODS sandwich and 6–7
PRINT procedure vs. 4–5, 8
process overview 9–11, 149, 367–381
sequencing step events 52–54
summary statistics support 26
syntax 4–5
TABULATE procedure vs. 8
report row phase 11, 150, 370
_BREAK_ automatic variable 164–166
defined 10, 150
report rows 9–10
report variables 10
reporting tools 7–9
reports
See also detail reports
See also summary reports
adding headers to 41
adding logos to 219–222
adding text to 39–45
aligning columns across 365–366
automating 341–345
CALL DEFINE routine in 228–229
concatenating 8, 30
controlling size of 223
coordinating graphics with 345–357
creating vertically concatenated tables
326–341
exporting to Microsoft Excel 316–322
generating 150
generating multiple reports on same page
360–366
referencing items in compute blocks
154–166
460
Index
reports (continued)
routing to ODS destinations 6–7
workaround for monospace-only options
357–360
RETAIN statement 182
reversing column order 110–111
RIGHT option
COMPUTE statement 179, 222
DEFINE statement 110–112
RIGHTMARGIN attribute 215, 301–303
_ROW_ keyword 174
rows
See also report row phase
GROUP variables repeating on 185–187
ORDER variables repeating on 185–187
ordering 19–22
report 9–10
shaded 229–231
RTF control words 312–313
RTF destination 310–314
BODYTITLE option and 311–312
BY statement and 137
creating links for 257, 261, 265–266
dagger symbol 278
displaying page numbers 273–277
formatting with escape character 270
importing graphics to 347
LINK= option and 254
page breaks in 187
passing raw codes 289
proportional fonts and 48–49
specifying multiple columns 298–299
STARTPAGE=NEVER option 363–364
SUMMARIZE option, BREAK statement
61
superscripts/subscripts 272
S
S={ } formatting string 279–282, 297
SAS/GRAPH 345–347, 352, 355
annotate facility 355–357
SAS language elements
compute blocks and 51–52, 182–191
LINE statement execution 197
SAS Monospace font 77–78, 84
scaling column width 78
SET statement 182
setup phase 11, 149–150, 369
ACROSS variables and 150
calculating columns 150
defined 10, 150
determining column numbers 161
shaded rows 229–231
SKIP option, BREAK statement 58, 59–61, 78
LISTING destination 62, 64, 72
ODS destinations and 60, 64
order of 59, 62–63
PDF destination and 60
SUMMARIZE option and 62
skipping lines between groups 59–61, 64–65,
102
SORT procedure 137, 338
sorting variables 20, 21
space
adding space between columns 94–95
column spacing 358
horizontal 223–227
non-breaking 284–287
skipping lines between groups 59–61,
64–65, 102
vertical 223–227
SPACING= option, DEFINE statement 92,
94–95, 126, 358
spanning headers 80–82, 194–196
SPLIT= option
PRINT procedure 44
REPORT procedure 44–45, 82–83
STARTPAGE= option, ODS PDF statement
264, 363–364
statistics
ACROSS variables and 29–30, 37–39
analysis statistic 31–33
ANALYSIS variables and 22, 31–33,
36–37, 99, 155
associating with DEFINE statement 37–39
attaching with comma 26–28
calculations based on 99–101
CALL DEFINE routine and 177
compound variable names 159–160
creating group summaries 192–194
DEFINE statement and 31
MEAN statistic 26, 31, 39, 159–160, 203
MEDIAN statistic 26, 31
N statistic 26, 31, 36–37, 193–194, 203
nesting 29–30, 155
numeric variables and 28
STD statistic 26, 159, 203
SUM statistic 26, 31, 159
SUMMARY statistic 8
VAR statistic 26
VARIANCE statistic 31
Index 461
STD statistic 26, 159, 203
style attributes
changing with CALL DEFINE routine
227–232
changing with escape character 279–282
changing with STYLE= option 216–227
nesting 281–282
STYLE= option
BREAK statement 213–215
CALL DEFINE routine 175, 213–215
CALLDEF component 214–215
changing attributes 216–227
COLUMN component 214–215
COMPUTE statement 213–215
DEFINE statement 213–215
flyover text and 295–297
FORMAT= option and 234
formats and 234–236
forming links with 260–261
HDR component 214–215
HEADER component 214–215, 296
importing graphics and 345, 348
JUST= style attribute 215, 219, 222
LINES component 214–215
ODS 213–227
ODS STYLE attributes 214–215
PRINT procedure 7
RBREAK statement 213–215
REPORT component 214–215
REPORT procedure 7, 213–215
STYLE=MINIMAL option 66
SUMMARY component 214–215, 238,
241
TABULATE procedure 7
trafficlighting effects 233–249
{STYLE} formatting sequence 279–282, 297
SUB function 271–273
SUBJECT= option, ODS PDF statement
314–315
subscripts 271–273
subtotals, calculating 8
SUM statement
BY statement and 8
compute blocks and 183–185
counting items in groups 187
LINE statement and 193–194
temporary variables and 157–158
SUM statistic 26, 31, 159
summaries, creating for groups 192–194
summarization label 65–67
SUMMARIZE option
BREAK statement 54, 58–59, 61–67, 169,
201
RBREAK statement 54, 58–59, 169
summarizing across groups 61–65
SUMMARY component, STYLE= option
214–215, 238, 241
summary lines 374–378
building tables with 371–372
SUMMARY procedure 8
creating vertically concatenated tables
327–328, 339
FREQ statement and 144
summary statistics support 26
summary reports
BREAK statement and 71–72
combining with detail reports 68–69
computed summary information 10
creating 8
defined 9, 18
DISPLAY variables and 59
grouping variables and 72
RBREAK statement and 69, 71–72
SUMMARIZE option, BREAK statement
65–67
suppressing labels 65–67
SUPER function 271–273
superscripts 271–273
SUPPRESS option
BREAK statement 58, 169
RBREAK statement 152, 169
suppressing
labels 65–67
ORDER variables 88
summarization label 65–67
SYMPUT CALL routine 345
T
tables
building with summary lines 371–372
concatenating vertically 326–341
embedding hyperlinks in 249–269
TABULATE procedure 7–8, 326
STYLE= option 7
tagsets 320–323
Temporary Internal File (TIF) 10
462
Index
temporary variables
_BREAK_ 164–165, 173–174, 193
compute block processing and 10, 152,
155, 157–158, 183–185
defined 10
in COLUMN statement 10, 157
LINE statement execution 197–198
SUM statement and 157–158
text
adding to reports 39–45
adding via COLUMN statement 41–43
adding via compute block 47–49, 170–171
adding via DEFINE statement 43–44
adding with TEXT= option 300
columnar alignment 335–338
error text 287–289
non-breaking spaces in 284–287
tip text 293–297
warning text 287–289
wrapping 93–94, 358
text attributes 216–219
TEXT= option 300
{THISPAGE} function 273–277
TIF (Temporary Internal File) 10
tip text 293–297
TITLE= option, ODS PDF statement 314–315
TITLE statement
BColor= option 292
BOLD option 292
BY statement and 139–141
Color= option 292
Font= option 292
Height= option 292
ITALIC option 292
Justify= option 292
LINK= option 254
ODS destinations 292–293
S={ } formatting string 279
{STYLE} formatting sequence 281
UNDERLINE option 292
underlining in 290
titles 250–255, 279–282
totals
calculating 8
grand totals 200–202
totaling variables 22
writing on every page 200–202
trafficlighting effects 232–249
TRANSPOSE procedure 24, 328, 338–340
U
UL option
BREAK statement 59, 78
RBREAK statement 59, 78
UNDERLINE option, TITLE/FOOTNOTE
statements 292
underlines
BREAK/RBREAK statement options 78
HEADLINE option 76–77
in TITLE statement 290
LISTING destination 78–79
underscore (_) 79, 83
UNIVARIATE procedure 8, 26
UPDATE statement 182
URL attribute, CALL DEFINE routine 175,
257–260
URLBP attribute, CALL DEFINE routine 175
URLP attribute, CALL DEFINE routine 175
user-defined formats 123–126, 268–269
V
VALUE statement, FORMAT procedure 124–
125, 130–131
NOTSORTED option 130–131
values
combining in fields/columns 202–204
combining with nested ACROSS variables
204–206
formatted 33, 49–50, 171–172
HTML anchor tags as 255–257
ordering display of 34–36, 111
VAR statistic 26
variable names 156–161, 194
compound 159–161
variables
See also ACROSS variables
See also ANALYSIS variables
See also character variables
See also define types
See also GROUP variables
See also grouping variables
See also ORDER variables
See also temporary variables
ACROSS define type and 25
aliases for 121–122
ANALYSIS define type and 22
classification 113–118, 134–136, 204
COMPUTED 18, 155, 203
Index 463
DEFINE statement and 18
macro 51, 196, 291, 343–345
nesting 122–123
nonprinting 118
numeric 18, 28, 156–157
report variables 10
sorting 20, 21
totaling 22
trafficlighting effects and 236–240
VARIANCE statistic 31
vertical spaces 223–227
vertically concatenated tables 326
W
warning text 287–289
weighted mean, calculating 206–209
WHERE clause 266
WIDTH= option, DEFINE statement 33, 92–93
windowing environment, interactive 4
WRAP option 91–92, 132
wrapping
data lines 91–92
line breaks and 282–289
text 93–94, 358
X
XLS extension 317–320
XPIXELS option, GOPTIONS procedure 351
Y
YPIXELS option, GOPTIONS procedure 351
Symbols
* (asterisk) 44, 79, 194
@ (at sign) 48
\ (backslash) 79
: (colon) 79
, (comma) 26–28, 205
- (dash) 79
= (equal sign) 79–81
> (greater than) 79, 81
< (less than) 79, 81
() parentheses 28–29, 41–42
. (period) 79, 159
+ (plus sign) 79
_ (underscore) 79, 83
xn sequency 282–284
xz formatting sequences 287–289
464
Index
Books Available from SAS Press
Advanced Log-Linear Models Using SAS ®
Debugging SAS ® Programs: A Handbook of Tools and
Techniques
by Daniel Zelterman
by Michele M. Burlew
Analysis of Clinical Trials Using SAS®: A Practical Guide
by Alex Dmitrienko, Geert Molenberghs, Walter Offen, and
Christy Chuang-Stein
Decision Trees for Business Intelligence and Data Mining: Using
SAS® Enterprise MinerTM
by Barry de Ville
Annotate: Simply the Basics
by Art Carpenter
Efficiency: Improving the Performance of Your SAS ®
Applications
Applied Multivariate Statistics with SAS® Software,
Second Edition
by Robert Virgile
by Ravindra Khattree
and Dayanand N. Naik
Elementary Statistics Using JMP®
Applied Statistics and the SAS ® Programming Language,
Fifth Edition
The Essential Guide to SAS ® Dates and Times
by Sandra D. Schlotzhauer
by Ronald P. Cody
and Jeffrey K. Smith
The Essential PROC SQL Handbook for SAS ® Users
An Array of Challenges — Test Your SAS ® Skills
by Robert Virgile
Building Web Applications with SAS/IntrNet®: A Guide to the
Application Dispatcher
by Don Henderson
by Katherine Prairie
Fixed Effects Regression Methods for Longitudinal Data
Using SAS ®
by Paul D. Allison
Genetic Analysis of Complex Traits Using SAS ®
Carpenter’s Complete Guide to the SAS® Macro Language,
Second Edition
by Art Carpenter
Carpenter’s Complete Guide to the
by Derek P. Morgan
SAS®
REPORT Procedure
Edited by Arnold M. Saxton
A Handbook of Statistical Analyses Using SAS®, Second Edition
by B.S. Everitt
and G. Der
by Art Carpenter
Health Care Data and SAS®
The Cartoon Guide to Statistics
by Marge Scerbo, Craig Dickstein,
and Alan Wilson
by Larry Gonick
and Woollcott Smith
Categorical Data Analysis Using the
Second Edition
The How-To Book for SAS/GRAPH ® Software
SAS ®
System,
by Maura E. Stokes, Charles S. Davis,
and Gary G. Koch
by Thomas Miron
In the Know ... SAS ® Tips and Techniques From
Around the Globe, Second Edition
by Phil Mason
Cody’s Data Cleaning Techniques Using SAS® Software
by Ron Cody
Instant ODS: Style Templates for the Output Delivery System
Common Statistical Methods for Clinical Research with
SAS ® Examples, Second Edition
Integrating Results through Meta-Analytic Review Using
SAS® Software
by Morgan C. Wang
and Brad J. Bushman
by Glenn A. Walker
by Bernadette Johnson
The Complete Guide to SAS ® Indexes
by Michael A. Raithel
Introduction to Data Mining Using SAS® Enterprise MinerTM
by Patricia B. Cerrito
Data Management and Reporting Made Easy with
SAS ® Learning Edition 2.0
Learning SAS® by Example: A Programmer’s Guide
by Sunil K. Gupta
by Ron Cody
Data Preparation for Analytics Using SAS®
Learning SAS ® in the Computer Lab, Second Edition
by Gerhard Svolba
by Rebecca J. Elliott
support.sas.com/pubs
The Little SAS ® Book: A Primer
by Lora D. Delwiche
and Susan J. Slaughter
Professional SAS ® Programmer’s Pocket Reference,
Fifth Edition
The Little SAS ® Book: A Primer, Second Edition
by Lora D. Delwiche
and Susan J. Slaughter
(updated to include SAS 7 features)
Professional SAS ® Programming Shortcuts, Second Edition
The Little SAS ® Book: A Primer, Third Edition
by Lora D. Delwiche
and Susan J. Slaughter
(updated to include SAS 9.1 features)
by Arthur L. Carpenter
and Charles E. Shipp
The Little SAS ® Book for Enterprise Guide® 3.0
by Susan J. Slaughter
and Lora D. Delwiche
The Little SAS ® Book for Enterprise Guide® 4.1
by Susan J. Slaughter
and Lora D. Delwiche
Logistic Regression Using the SAS® System:
Theory and Application
by Paul D. Allison
Longitudinal Data and SAS®: A Programmer’s Guide
by Ron Cody
Maps Made Easy Using SAS®
by Mike Zdeb
by Rick Aster
by Rick Aster
Quick Results with SAS/GRAPH ® Software
Quick Results with the Output Delivery System
by Sunil K. Gupta
Reading External Data Files Using SAS®: Examples Handbook
by Michele M. Burlew
Regression and ANOVA: An Integrated Approach Using
SAS ® Software
by Keith E. Muller
and Bethel A. Fetterman
SAS ® for Forecasting Time Series, Second Edition
by John C. Brocklebank
and David A. Dickey
SAS ® for Linear Models, Fourth Edition
by Ramon C. Littell, Walter W. Stroup,
and Rudolf J. Freund
SAS ® for Mixed Models, Second Edition
Models for Discrete Data
by Daniel Zelterman
by Ramon C. Littell, George A. Milliken, Walter W. Stroup,
Russell D. Wolfinger, and Oliver Schabenberger
Multiple Comparisons and Multiple Tests Using SAS®
Text and Workbook Set
SAS ® for Monte Carlo Studies: A Guide for Quantitative
Researchers
(books in this set also sold separately)
by Peter H. Westfall, Randall D. Tobias,
Dror Rom, Russell D. Wolfinger,
and Yosef Hochberg
Multiple-Plot Displays: Simplified with Macros
by Perry Watts
Multivariate Data Reduction and Discrimination with
SAS ® Software
by Ravindra Khattree
and Dayanand N. Naik
by Xitao Fan, Ákos Felsovályi,
Stephen A. Sivo,
˝
and Sean C. Keenan
SAS ® Functions by Example
by Ron Cody
SAS ® Guide to Report Writing, Second Edition
by Michele M. Burlew
SAS ® Macro Programming Made Easy, Second Edition
by Michele M. Burlew
SAS ® Programming by Example
Output Delivery System: The Basics
by Lauren E. Haworth
Painless Windows: A Handbook for SAS ® Users, Third Edition
by Jodie Gilmore
(updated to include SAS 8 and SAS 9.1 features)
Pharmaceutical Statistics Using SAS®: A Practical Guide
Edited by Alex Dmitrienko, Christy Chuang-Stein,
and Ralph D’Agostino
The Power of PROC FORMAT
by Jonas V. Bilenas
PROC SQL: Beyond the Basics Using SAS®
by Ron Cody
and Ray Pass
SAS ® Programming for Researchers and Social Scientists,
Second Edition
by Paul E. Spector
SAS ® Programming in the Pharmaceutical Industry
by Jack Shostak
SAS ® Survival Analysis Techniques for Medical Research,
Second Edition
by Alan B. Cantor
by Kirk Paul Lafler
SAS ® System for Elementary Statistical Analysis,
Second Edition
PROC TABULATE by Example
by Sandra D. Schlotzhauer
and Ramon C. Littell
by Lauren E. Haworth
support.sas.com/pubs
SAS ® System for Regression, Third Edition
by Rudolf J. Freund
and Ramon C. Littell
SAS ® System for Statistical Graphics, First Edition
by Michael Friendly
The SAS ® Workbook and Solutions Set
(books in this set also sold separately)
JMP® Books
JMP® for Basic Univariate and Multivariate Statistics: A Step-byStep Guide
by Ann Lehman, Norm O’Rourke, Larry Hatcher,
and Edward J. Stepanski
JMP® Start Statistics, Third Edition
by Ron Cody
by John Sall, Ann Lehman,
and Lee Creighton
Selecting Statistical Techniques for Social Science Data:
A Guide for SAS® Users
Regression Using JMP®
by Frank M. Andrews, Laura Klem, Patrick M. O’Malley,
Willard L. Rodgers, Kathleen B. Welch,
and Terrence N. Davidson
by Rudolf J. Freund, Ramon C. LIttell,
and Lee Creighton
Statistical Quality Control Using the SAS ® System
by Dennis W. King
Statistics Using SAS ® Enterprise Guide®
by James B. Davis
A Step-by-Step Approach to Using the SAS ® System
for Factor Analysis and Structural Equation Modeling
by Larry Hatcher
A Step-by-Step Approach to Using SAS ® for Univariate and
Multivariate Statistics, Second Edition
by Norm O’Rourke, Larry Hatcher,
and Edward J. Stepanski
Step-by-Step Basic Statistics Using SAS®: Student Guide
and Exercises
(books in this set also sold separately)
by Larry Hatcher
Survival Analysis Using SAS ®:
A Practical Guide
by Paul D. Allison
Tuning SAS ® Applications in the OS/390 and z/OS
Environments, Second Edition
by Michael A. Raithel
Univariate and Multivariate General Linear Models:
Theory and Applications Using SAS ® Software
by Neil H. Timm
and Tammy A. Mieczkowski
Using SAS ® in Financial Research
by Ekkehart Boehmer, John Paul Broussard,
and Juha-Pekka Kallunki
Using the SAS ® Windowing Environment: A Quick Tutorial
by Larry Hatcher
Visualizing Categorical Data
by Michael Friendly
Web Development with SAS® by Example, Second Edition
by Frederick E. Pratter
Your Guide to Survey Research Using the SAS® System
by Archer Gravely
support.sas.com/pubs
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.6 Linearized : No XMP Toolkit : 3.1-701 Modify Date : 2015:11:02 09:47:13+08:00 Create Date : 2008:02:15 23:58:52+03:00 Metadata Date : 2015:11:02 09:47:13+08:00 Creator Tool : PScript5.dll Version 5.2.2 Format : application/pdf Title : Creator : . Document ID : uuid:9c34d7d8-c946-4883-9a18-25f851c10f68 Instance ID : uuid:490a4a5b-109e-42b4-a59c-efe4865ec86d Producer : Acrobat Distiller 7.0.5 (Windows) Has XFA : No Page Count : 497 Page Layout : SinglePageEXIF Metadata provided by EXIF.tools