The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Other countries » Zambia hiv dataset 2001-02
Zambia hiv dataset 2001-02 [message #3148] Thu, 23 October 2014 03:44 Go to next message
hannekeyserhegdahl is currently offline  hannekeyserhegdahl
Messages: 9
Registered: October 2014
Member
Hi!

I would like to use the HIV dataset from the 2001-02 Zambia DHS for analysis, but I donĀ“t understand how to survey set this dataset in STATA. Do I have to use a psu and which variable is it? I assume that the weight to be used is hiv_wgt.

Thanks for any answers.
Re: Zambia hiv dataset 2001-02 [message #3154 is a reply to message #3148] Fri, 24 October 2014 11:34 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
It is not possible to use this dataset in the conventional way with the svyset command as we have no PSU identifier for this dataset. The best you can do is the following:

egen strata=group(hivprov hivresid)
svyset hivid [pweight=hiv_wgt], strata(strata)

Here we are using hivid in place of the psu, which will effectively give one case per "psu".
Re: Zambia hiv dataset 2001-02 [message #3156 is a reply to message #3154] Fri, 24 October 2014 15:31 Go to previous messageGo to next message
Reduced-For(u)m
Messages: 292
Registered: March 2013
Senior Member


Trevor,

Given that, on its own, stratification should (weakly) reduce the size of standard errors, where the clustering on PSU (weakly) inflates them - isn't this likely to produce standard error estimates (and thus p-values and CIs) that are all too small?

I would think that clustering on some geographic level greater than PSU would be the more conservative way to do this. The AIS tabulation plan* seems to indicate that they keep region of residence. Clustering on region-X-urban/rural might provide enough clusters to use the cluster robust estimator, but there are small-cluster-number analogs that can be used as well.

*Maybe I'm looking at the wrong documentation: http://dhsprogram.com/pubs/pdf/AISM9/HIV_Testing_Tabplan.pdf
Re: Zambia hiv dataset 2001-02 [message #3157 is a reply to message #3148] Fri, 24 October 2014 16:43 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
I asked our sampling specialist Ruilin Ren for his thoughts on this. Here is his reply:

"It is true that without declaring the cluster, the standard error of the estimators will be under estimated, especially for indicators with strong design effect. But taking the stratum (province x residence) as cluster will over estimate the standard error equally. So there is no perfect solution. For HIV prevalence, it may not be a major problem without declaring the cluster because HIV prevalence usually has weak design effect compared to other indicators."

Re: Zambia hiv dataset 2001-02 [message #3166 is a reply to message #3154] Tue, 28 October 2014 11:07 Go to previous messageGo to next message
hannekeyserhegdahl is currently offline  hannekeyserhegdahl
Messages: 9
Registered: October 2014
Member
Thank you!
Re: Zambia hiv dataset 2001-02 [message #3545 is a reply to message #3154] Mon, 05 January 2015 03:47 Go to previous messageGo to next message
hannekeyserhegdahl is currently offline  hannekeyserhegdahl
Messages: 9
Registered: October 2014
Member
Hi again!

I have the same problem with the Mali 2001 hiv dataset as with the Zambia 2001-02 hiv dataset. Can I use the same method on the dataset from Mali?
Re: Zambia hiv dataset 2001-02 [message #3547 is a reply to message #3148] Mon, 05 January 2015 09:56 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
Yes, the Mali 2001 HIV dataset can be treated the same way as the Zambia 2001-02 HIV dataset.
Re: Zambia hiv dataset 2001-02 [message #10652 is a reply to message #3154] Thu, 25 August 2016 15:19 Go to previous messageGo to next message
CKAllen is currently offline  CKAllen
Messages: 7
Registered: April 2016
Location: London, UK
Member
Because there is no PSU identifier, does this mean this dataset is impossible to merge with the IR dataset?
Re: Zambia hiv dataset 2001-02 [message #10654 is a reply to message #10652] Thu, 25 August 2016 20:55 Go to previous message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
Yes, it is not possible to merge the HIV test results for Zambia 2001-02, Mali 2001, and Dominican Republic 2002 with the Individual Recode datasets. These surveys were the first three to collect blood samples for HIV testing and at the time the protocol called for completely de-linked testing. However, these datasets do include a few characteristics of respondents such as age, sex, marital status, type of place of residence.
Previous Topic: Senegal- STI in last 12 months
Next Topic: DRC - outpatient care and hospitalization data
Goto Forum:
  


Current Time: Thu Mar 28 11:30:28 Coordinated Universal Time 2024