| HIV testing: Ethiopia 2005 [message #25501] |
Sun, 30 October 2022 21:07  |
salome
Messages: 19
Registered: August 2021
|
Member |
|
|
Hello
I'd appreciate any help in explaining the large number of NA/missing values for v781 (ever-testing) in Ethiopia 2005 women's recode. The documentation says that v781 was created as a combination of S910J (Test as part of antenatal care), S910M(Test since test during pregnancy) and ever tested. However I am unable to find the latter "ever tested" variable in the dataset. The only one is v781 and that too with lots of missing values.
My hunch is that v781 currently represents ever tested among women who had delivered after 'MESKEREM 1995'. Other women should have been asked about ever testing Q910O but this data are not represented. I was wondering if this is the case?
Thank you so much in advance.
Salome
[Updated on: Sun, 30 October 2022 21:10] Report message to a moderator |
|
|
|
| Re: HIV testing: Ethiopia 2005 [message #25542 is a reply to message #25501] |
Mon, 07 November 2022 16:36  |
Janet-DHS
Messages: 888
Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member Tom Pullum:
In this survey there was subsampling of men. As is typical for such subsampling, the selection was done at the household level. In half of the households, any eligible men (de facto residents age 15-59) would be interviewed (a selected household might include NO eligible men) and in half the households no men would be interviewed. The HR and PR files include a variable hv027 which is 1 if the household was selected for the men's interview and 0 if it was not. Unfortunately, hv027 was not copied into the IR files, so we would have to merge with the PR file to find the value of hv027 for a specific woman. If you look at the women's questionnaire in the appendix to the final report on the 2005 survey, you will see that 910A is a filter, where the interviewer is instructed to skip several of the HIV-related questions, including the one on testing, for v781, if the woman is not in a household selected for the men's sample. Such women would have been assigned the Not Applicable (NA) code for v781. That code should not be interpreted as No Response (NR) or "missing". The question about testing was not asked of those women.
The justification for just asking the question about testing in the subsample was that HIV tests were subsequently administered (conditional on consent, of course) to the women and men in the subsample. The results of the tests are in the AR file. Some other questions in the women's questionnaire related to HIV testing during pregnancy were not restricted to the households with hv027=1.
The main reason for subsampling is to balance the costs of different parts of the data collection. Given the purposes of the surveys, the data coming from women need a larger sample than the data coming from men. Also the HIV estimates are needed mainly at a national level and don't require the same sample size as other indicators that need sub-national estimates.
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
