The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Rwanda 2014 HIV weights
Rwanda 2014 HIV weights [message #13344] Wed, 25 October 2017 20:34 Go to next message
amage is currently offline  amage
Messages: 6
Registered: September 2017

I merged the women's, men's, household member, and HIV testing datasets from the 2014-2015 Rwanda survey and then subset the HIV+ population > 15 years of age. Among this group, there are 3 men and 1 woman who are not included in the men's or women's datasets, respectively. For all 4 of them, the HIV05 weight=0. Is this correct, or did I somehow merge the files incorrectly? Why would the weight=0?

I also noticed that the three men are the only HIV+ respondents with height and weight data. Was there very little overlap between the subset of participants who were tested for HIV and those who were measured for anthropometrics? I noticed there was much greater overlap between these groups in the Burundi 2010 survey.

Thank you!
Re: Rwanda 2014 HIV weights [message #13461 is a reply to message #13344] Mon, 06 November 2017 10:59 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3062
Registered: February 2013
Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Sorry for the delay with this reply. We have been discussing the nature of those cases. We find that several other AR files, not just this one, include a handful of cases with weight hiv05 equal to 0. In general, those cases are women who are not in the IR file or men who are not in the MR file, because they were not eligible for the individual interview under the criterion that hv103=1 (de facto residents who "slept here last night"). However, there may be exceptions to this generalization. They should not have been tested. We always recommend that when using hiv03, you should use hiv05 as the weight. If you do that, those cases will of course be omitted. There is no weight for them, so there would not appear to be a way to include them in any analysis. We do not see why they are in the data files. Please just ignore them.
Previous Topic: Weights for survey-specific tabulations with pooled data
Next Topic: Weighting the husband/partner
Goto Forum:

Current Time: Mon May 20 20:47:08 Coordinated Universal Time 2024