The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Tanzania 2010
Tanzania 2010 [message #3101] Wed, 15 October 2014 13:56 Go to next message
azucena2014 is currently offline  azucena2014
Messages: 5
Registered: August 2014
Member
I am using the Tanzania 2010 DHS and am trying to use the survey commands in Stata. So, I use the commands I usually use when I want to apply the survey settings in Stata with DHS data:

gen weight=hv005/1000000
svyset hv021 [pweight=weight], strata(hv022)

However, when I try to anything that require a test statistic, I get the following message:

"Note: Missing test statistics because of stratum with single sampling unit."

Any help is greatly appreciated


EAS
Re: Tanzania 2010 [message #3112 is a reply to message #3101] Thu, 16 October 2014 09:06 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
You do not have to create a new weight variable - the following command should be sufficient:

svyset hv021 [pw=hv005], strata(hv022)

For a similar post, please reference message # 2784
Re: Tanzania 2010 [message #3113 is a reply to message #3112] Thu, 16 October 2014 10:56 Go to previous messageGo to next message
azucena2014 is currently offline  azucena2014
Messages: 5
Registered: August 2014
Member
Thanks, but I just tried this and got the same message (Note: Missing test statistics because of stratum with single sampling unit) when I tried the following:

svy, subpop(age059): tab stunting depndgtr1, col obs

I am running a similar analysis on various countries and this is the first time I have encountered this problem


EAS
Re: Tanzania 2010 [message #3115 is a reply to message #3113] Thu, 16 October 2014 22:24 Go to previous messageGo to next message
Reduced-For(u)m
Messages: 292
Registered: March 2013
Senior Member

What happens if you tabulate (list) the strata? Is there one strata value that is a typo or "missing" or something? It sounds like you might have a rogue strata in there, and if you find it you could just drop that observation. Let me know if that works.
Re: Tanzania 2010 [message #3117 is a reply to message #3113] Thu, 16 October 2014 22:47 Go to previous messageGo to next message
azucena2014 is currently offline  azucena2014
Messages: 5
Registered: August 2014
Member
Yes, I had already tried that and everything looks normal. I also did a cross-tab of hv021 and hv022 to see if there was something there that was odd but that also all looked normal, too. I downloaded this data 3 days ago. Any other ideas? I need to be able to report the test statistics

EAS
Re: Tanzania 2010 [message #3125 is a reply to message #3101] Sat, 18 October 2014 14:39 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
The problem is that in certain strata there is only one cluster, and for the calculation of sampling errors and confidence intervals at least two clusters are required in each cluster. In this case the best thing is to regroup the strata to permit the calculation of the sampling errors. In the case of this Tanzania dataset, the following have only one cluster:
18: Kagera, urban
22: Kusini Pemba, urban
23: Kaskazini Pemba, urban
I suggest grouping these with the rural strata for the same regions, e.g. with
44: Kagera, rural
48: Kusini Pemba, rural
49: Kaskazini Pemba, rural

Thus I suggest using the following:
recode hv022 (18=44) (22=48) (23=49), gen(strata_new)
svyset hv021 [pw=hv005], strata(strata_new)

As you are using a subpop with your tabulation, you may hit this issue with other strata, in which case you will need to make a decision to regroup further strata together.
Re: Tanzania 2010 [message #3126 is a reply to message #3125] Sun, 19 October 2014 09:26 Go to previous message
azucena2014 is currently offline  azucena2014
Messages: 5
Registered: August 2014
Member
ok, thanks! I will try this

EAS
Previous Topic: Colombia 2010 weights
Next Topic: Phase I surveys
Goto Forum:
  


Current Time: Thu Mar 28 04:45:16 Coordinated Universal Time 2024