Survey Weights for 2010 Kenya SPA Health Worker Interview [message #9310] |
Tue, 08 March 2016 11:16 |
cudjoe@gwmail.gwu.edu
Messages: 1 Registered: August 2014
|
Member |
|
|
Hello,
I am hoping to do an analysis of the 2010 Kenya SPA - Heath Worker Interview only (not the complete dataset). I will be looking only at providers in Nairobi. Do these data need to be weighted? If so, which variable do I use for weighting and what are the STATA commands to do this?
Thanks,
|
|
|
Re: Survey Weights for 2010 Kenya SPA Health Worker Interview [message #9325 is a reply to message #9310] |
Thu, 10 March 2016 14:17 |
Bridgette-DHS
Messages: 3203 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Stata Specialist, Rebecca Winter:
Thanks for your question! Yes, the 2010 Kenya SPA is a sample of all health facilities in Kenya. Likewise, it is a sample of health providers. In order to tabulate results that are representative of all health facilities (or in this case, all health providers) in the country, you will need to use the data file's survey weight. This weight will adjust for any over- or under- sampling in certain areas, and will also adjust for any nonresponse.
If you are using the recoded provider file, KESR10PV.DTA, the survey weight variable is called u005. To prepare this weight variable for use, you need to divide the variable by 1,000,000, as shown below.
generate wgt = u005/1000000
Once the weight is prepared, you just need to tell STATA to use the weight when running your tabulations. For example, to tabulate the male/female distribution of health providers in Nairobi, you could use the code below:
tabulate u030 if u001==1 [iw=wgt]
We hope this helps!
|
|
|