Exercise_instructions Exercise Instructions

User Manual:

Open the PDF directly: View PDF .
Page Count: 8

Download
Open PDF In Browser	View PDF

Exercise
Below is an exercise to show you how to use the ipacheck package. In the exercise, we
will be working with data collected from a previous IPA project. The data were collected
using SurveyCTO, all PII have been removed and any GPS points have been anonymized.

Instructions
Section 1: Package Overview
1. To download the exercise, start by using the ipacheck command. This will initialize a
folder structure, readme files, all the input sheets, and the data for the exercise:

ipacheck new, exercise

Note: if you are using the ipacheck package for your own project, you can use ipacheck
new to download the file structure and input files without the exercise data. Use help
ipacheck for the full functionality.
2. Now that we're in the proper directory, let's take a look at the HFC files. Start with
the 04_checks/01_inputs folder. You should see these files:
o
o

hfc_inputs.xlsm - this is the input Excel file; inside you'll find a convenient form

for configuring the HFC commands.
hfc_replacements.xlsm - this is the replacement Excel file; it is a running list of
edits/corrections based on HFC outputs; these replacements can be automatically
added to your workflow using readreplace.

Now go back to the main folders, and navigate to 02_dofiles.


master_check.do - this is the master dofile; it reads the inputs, makes any replacements,

runs the HFC checks, runs back checks, outputs violations and produces the dashboard.

Navigate to the 06_media folder and unzip the survey_media zip file. This contains all the
media files for text audits and field comments.
3. Open the master_check.do file and review the major sections. This file is the primary
controller for the HFCs, it is documented to help you understand what it's doing at each
stage. All sections are set up to run according to how you set up the inputs file, but it
helps to have an idea of what is going on to be able to troubleshoot issues.

By pulling from a centralized source you'll always make sure you're starting fresh with
the latest files. There are also a few other utility functions included in
the ipacheck package including:




ipacheck update - downloads the updated ado files directly from GitHub whenever IPA

HQ releases an update so you don't have to go through the above Installation process
again.
ipacheck version - lists the current installed versions of the user-written commands
(useful for verifying you have the latest installed).

Section 2: Configure the inputs
The following section will take you through the stages of setting up your HFCs. After
each stage you are encouraged to run to the corresponding point in
the master_check.do file to see the outputs created.
1. Navigate to the 04_checks/01_inputs folder and open the hfc_inputs.xlsm file in
Excel. This is where you'll configure the HFCs. Each logic check has its own
worksheet. You'll notice, we've tried to make the input file more user-friendly by
adding automatic formatting and help boxes in each sheet. In this exercise we're
going to configure checks for the survey_data.dta dataset.
2. Open the 0. setup sheet. This sheet is where you set global options and link to
the the appropriate files necessary for running the HFCs.


Section 1 of the 0. setup sheet links to all the input and output files (Note: you can
include file paths if the files are in separate folders). The Master Tracking Dataset refers to
a Stata dataset containing your full sample list, either from census or previous survey
waves.



Section 2 specifies the names of the input and replacement files.



Section 3 specifies the names of the output files.




Section 4 specifies key variables in your survey.
Section 5 specifies code values for don't know, missing, and not applicable. Notice, in
this case, this will change all values of -999 to .d, -888 to .r, and -222 to .n. Keep this in
mind when reading outputs and making replacements.
Section 6 includes options for Progress Report using the progreport command, which
compares survey data to a master dataset for summaries of completion.




Section 7 includes specifications for high-frequency checks.



Section 8 specifies options for for back checks.



Section 9 includes your SurveyCTO server and username so you can view observations
from a link on the output sheet.
Section 10 allows you to switch on/off any check. Even if you fill out a sheet or the
specifications for a check, a check will not run if it is not turned on in Section 10. Update



the sheet with the options below.

3. Open the 1. incomplete sheet and review the help boxes. This check verifies that
all surveys have been completed. It corresponds with
the ipacheckcomplete command in the master_check.do file.
The ipacheckcomplete command can also check that each submission meets a
minimum nonmissing entry threshold by specifying a threshold value in
the complete_percent column. For our data set, a value of 2 in the
variable intstatus indicates a complete interview. Update
the variable and complete_value columns to intstatus and 2 and update
the complete_percent column to 40 indicating that we want to flag any
submission that has less than 40% of entries as nonmissing.
4. Open the 2. duplicates sheet and review the help boxes. This check verifies that
there are no duplicate surveys. The inputs are loaded to
the ipacheckdups command in the master_check.do file. For our data set, the
variables gpsLatitude and gpsLongitude should contain no duplicates. Update
the variable column to reflect this.
5. Open the 3. consent sheet and review the help boxes. This check verifies that all
surveys have consent. The inputs are loaded to the ipacheckconsent command in
the master_check.do file. For our data set, the value 1 for
variable consent indicate consent. Update
the variableand consent_value columns to reflect this.
6. Open the 4. no miss sheet and review the help boxes. This check verifies that
certain variables have no missing values. The inputs are loaded to
the ipachecknomiss command in the master_check.do file. For our dataset, the
variables gpsLatitude, gpsLongitude, enumid, consent, consentsign, ward, gender,
and age should have no missing values. Update the variable column to reflect
this.
7. Open the 5. follow up sheet and review the help boxes. This check verifies that
respondent data at follow up matches data in the master list. The inputs are
loaded to ipacheckfollowup in the master_check.do file. For our dataset, we want to
verify the consistency of gender and age between the master list and the current
dataset. Add these variables to the variablecolumn.
8. Open the 6. logic sheet and review the help boxes. This check verifies survey
logic and skip patterns. The inputs are loaded to
the ipachecklogic command. Update the variable, assert ,
and if_condition columns with the logic checks in the table
below.

9. Open the 7. all miss sheet and review the help boxes. This check verifies that
certain variables are not all missing. The inputs are loaded to
the ipacheckallmiss command in the master_check.do file. For our data set, check
all survey variables to see if any are all missing.You can use the Stata
wildcard * or _all to do it more efficiently!
10. Open the 8. constraints sheet and review the help boxes. This check verifies hard
and soft constraints. The inputs are loaded to the ipacheckconstraints command
in the master_check.do file. Update the variable, soft_min, soft_max, hard_min,
and hard_maxcolumns with the logic checks in the table below. Notice that
you can use Stata the wildcard * to specify constraints for all copies of
variables in a repeat
group.
11. Open the 9. specify sheet and review the help boxes. This check lists all
nonmissing specify other values to identify possible recodes or new categories.
The inputs are loaded to the ipacheckspecify command. Update the child and
parent column with all specify other variable combinations (hint: use ds
*_other in the command window)
12. Open the 10. dates sheet and review the help boxes. This check looks for
common survey date errors. The inputs are loaded to
the ipacheckdates command. Update the startdate, enddate ,
and surveystart columns with the data in the table
below.

13. Open the 11. outliers sheet and review the help boxes. This check looks for
potential outliers in continuous variable values. The inputs are loaded to
the ipacheckoutliers command. For our data set, we define a value 3.0 times the
SD as an outlier for the variables salary and childnum. Update
the variable and multiplier columns to reflect this.
14. Open the 12. field comments sheet and review the help boxes. Notice none of
these are required since we specified the variable name of the comments in
the 0. setup sheet. The comments files are in the 06_media/survey_media folder.
15. Open the 13. text audit sheet and review the help boxes. The group_name refers
to the groups coded in the SurveyCTO xlsform. For this exercise, we will look at
the consent_grpgroup to review duration of the survey once consent has been
confirmed. Enter consent_grp in the group_name column. The text audit files are in
the 06_media/survey_media folder.

16. Open the enumdb sheet and review the help boxes. This check creates the
enumerator dashboard: hfc_enumerators.xlsx. It compiles productivity, missing,
and nonresponse rates by surveyor and checks for the time spent surveying. The
inputs are loaded to the ipacheckenumcommand. Update the columns with the
data in the table
below.

17. Open the research oneway sheet and review the help boxes. This check creates
table summaries of key research variables and outputs them to the research
file: hfc_research.xlsx. The type of summary (means, medians, response
frequencies, etc) is determined by the variable type specified (e.g. continuous,
categorical, binary). The inputs are loaded to the first instance of
the ipacheckresearch command. Update the columns with the data in the table
below.

18. Open the research twoway sheet and review the help boxes. This check is the same
as the previous but allows you to summarize key outcomes by another variable
(e.g. treatment status, enumerator, region, etc.) specified in
the by column. Update the columns with the data in the table just as you did
with research oneway, but include treatment in the `by' column.
19. Open the backchecks sheet and review the help boxes. The
columns okrange_min, okrange_max, ttest, and reliability allow for different
specifications and tests, and the type column lets you specify what type of
question each variable is. Update the columns with the data in the table
below.

Section 3: Run and Review the Output

1. Before running your checks, make sure you have unzipped
the 06_media/survey_media folder.Navigate to the 02_dofiles folder to
open master_check.do and make sure it references the correct location and input
file in line 17. Run the whole do file. Once master_check.do has finished running,
you should have an updated hfc_outputs.xlsx available. This file contains lists of
check violations encountered by the HFC program. Open this file and inspect the
contents. You'll notice it is arranged in the same format as the input with a
separate sheet for each check. The output also includes a summary with overall
violation counts.
2. Navigate to the 04_checks/02_outputs folder and open the hfc_outputs.xlsx file.
Answer the following questions:
o

How many interviews have been conducted?

o

Are we missing any submissions that we planned?

o
o

Is everyone using the latest form version?
How many incomplete interviews are there? Inspect any incomplete
observations and see if you can figure out what is going on. (Hint use
the list or browse commands)

o

How many duplicates are there?

o

How may variables have missing values that shouldn't be missing?

o

How many skip pattern/logic violations are there? What can be done to
prevent/resolve these?

o

How many constraint violations are there? Do any values appear to be
nonsensical? What should be done?

o

Do you see any specify options that could be recoded or new categories?

o

Do all surveys have appropriate dates? What could be going on if not?

3. Open the hfc_enumerators.xlsx file and inspect the contents. Do you notice any
significant differences between the enumerators? What should be done in
response to these findings?
4. Open the hfc_research.xlsx file and inspect the contents. Do the entries make
sense? How would you summarize this for your PIs? Is there anything else you
might like to check?
5. Open the hfc_duplicates.xlsx file and inspect the contents. Why do you think the
duplicate occurred?
6. Navigate to the 03_tracking/02_outputs folder to open the hfc_tracking.xlsx file
and inspect the contents. How far along is survey progress? Is progress consistent
by day and by ward?

Section 4: Make Replacements

1. Open the hfc_replacements.xlsm workbook (in the 04_checks/01_inputs folder) and
read the help boxes. This file is used to make batch corrections/edits to the
survey dataset based on errors or violations found via the HFC template. You can
either drop an observation, replace a value in an observation, or mark an
observation as okay once you have confirmed the value and no longer want it to
show up in your output files. Create a new sheet using the instruction sheet with
the sheetname as survey_data, the ID variable as key, and the enumerator variable
as enumid. When filling out this sheet, it is important to use key instead of the ID
variable since there can be possible duplicates in your ID variable.
2. After reviewing all your output files and communicating the output of the HFCs
with your field teams you discover that one of the duplicate observations is a
duplicate, but has the correct values for the variables relationship, pregnant,
and childnum. Use the replacements sheet to change relationship from .d to 1,
change pregnant from .d to 0, change childnumfrom .d to 0 for the observation
with the key value uuid:faddd692-de86-11e8-9f32-f2801f111128, and drop the
duplicate with the key value uuid:fade4d02-de86-11e8-9f32-f2801f197841. Make
sure to correctly specify the action as drop, replace, or okay for each
change. When dropping an observation, use id in the variable column
and 1201 in the value column. This program confirms it is dropping the
correct observation by ensuring the values of variable and value are correct
for the key it is dropping.
3. Add the file hfc_replacements.xlsm and the corresponding
sheetname survey_data to the 0. setup sheet of the inputs and name the file for
the replacements log in Section 3. Verify that the replacements were made and
the errors no longer appear in the output file. Inspect the output file for other
potential replacements that can be made and add them to the list.

Section 5: Rerun the HFCs
1. Rerun the master_check.do and verify that the replacements were made and the errors
no longer appear in the output file. Inspect the output file for other potential
replacements that can be made and add them to the list.



---

Source Exif Data:

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.7
Linearized                      : No
Warning                         : Info object (115 0 obj) not found at 1272139

EXIF Metadata provided by EXIF.tools