Re: Third Party Present Measures [message #27662 is a reply to message #27611] |
Fri, 15 September 2023 17:35 |
Janet-DHS
Messages: 893 Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member, Tom Pullum:
For the Bolivia 2003 survey, using the IR file I entered (in Stata) "tab v501 v812,m" to look at how the responses about the husband being present varied according to the marital status of the woman. Here is the result:
. tab v501 v812,m
current marital | pres. husband (sec 7)
status | yes - lis yes - not . | Total
--------------------+---------------------------------+----- -----
never married | 5 5,587 8 | 5,600
married | 275 6,890 12 | 7,177
living together | 88 3,355 6 | 3,449
widowed | 1 244 0 | 245
divorced | 0 317 0 | 317
not living together | 0 865 1 | 866
--------------------+---------------------------------+----- -----
Total | 369 17,258 27 | 17,654
We see that almost all of the women who are not currently married have a response "yes-not listening", which does not make sense. Here is the value label for v812:
. label list v812
v812:
0 no
1 yes - listening
2 yes - not listening
The label includes 0 for "no" but there is not a single reported case with 0. It's pretty clear that in this survey the "no" cases that should have been given code 0 were actually assigned to code 2. My guess is that there was confusion between two alternative schemes for coding yes/no types of questions. There are many variables for which 1 is "yes" and 2 is "no", just as there are many questions for which 0 is "no" and 1 is "yes". Apparently the label for v812 was not followed correctly. This is intended to be a 3-category variable and in effect it has just 2 categories. I think "no" and "yes-not listening" were consolidated with code 2.
If your Spanish is better than mine, you may find the relevant question(s) in the questionnaire. There may have been some ambiguity there.
You are right, a few surveys do not have the "listening" variables at all.
|
|
|