The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Weighting appended samples using different countries' KR data files.
Weighting appended samples using different countries' KR data files. [message #23440] Thu, 16 September 2021 05:02 Go to next message
Melese is currently offline  Melese
Messages: 17
Registered: August 2019
Member
Dear Experts, greetings.
I appended KR data files from different countries for further analysis.
I created a sample weight, wgt (gen wgt=v005/1000000)for each country's data file before the append.
I then appended the files and started the further analysis. Am I wrong, please?
Should I de-normalize the women's individual sample weights (v005) before the append? If so, how can I do it, please?
thank you in advance.
Regards
Re: Weighting appended samples using different countries' KR data files. [message #23441 is a reply to message #23440] Thu, 16 September 2021 11:38 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3203
Registered: February 2013
Senior Member
There have been many similar postings on this topic - see two such posts below:

https://userforum.dhsprogram.com/index.php?t=msg&goto=36 2&&srch=pooling#msg_362
https://userforum.dhsprogram.com/index.php?t=msg&goto=35 70&&srch=pooling#msg_3570

You can also access the FAQs page to read about normalization:

https://dhsprogram.com/data/analysis-faqs.cfm #4. What is the normalization of weights?

If these do not provide enough information, please search the user forum for additional posts, using strings like: "pooling", "appending", "normalize'.

[Updated on: Thu, 16 September 2021 11:43]

Report message to a moderator

Re: Weighting appended samples using different countries' KR data files. [message #23443 is a reply to message #23441] Thu, 16 September 2021 14:21 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3203
Registered: February 2013
Senior Member

Following is a response from DHS Senior Sampling Specialist, Mahmoud Elkasabi:

There have been many postings about the weights in appended files. We have an FAQ about normalized weights (see #4 under weights section): https://dhsprogram.com/data/analysis-faqs.cfm . Equations and Stata code for de-normalization have been presented in MR27: https://www.dhsprogram.com/pubs/pdf/MR27/MR27.pdf

Note that if you do a full denormalization, then the pooled results will be dominated by the largest country. An alternative is to give equal weight to each survey. The options are discussed in previous postings.
Re: Weighting appended samples using different countries' KR data files. [message #23445 is a reply to message #23443] Thu, 16 September 2021 22:30 Go to previous messageGo to next message
Melese is currently offline  Melese
Messages: 17
Registered: August 2019
Member
Thank you Bridgette and Mahmoud for the timely responses. I always appreciate.
So, given the alternatives and risks of majority dominance, I chose the "equal weight for each survey" alternative.
That is I generate individual sample weight var, wgt for each survey before appending.
gen wgt= v005/1000000 and then append. I hope I am not wrong.
Regards.
Re: Weighting appended samples using different countries' KR data files. [message #23448 is a reply to message #23445] Fri, 17 September 2021 11:02 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3203
Registered: February 2013
Senior Member

Following is a response from DHS Senior Sampling Specialist, Mahmoud Elkasabi:

I'm afraid that appending the surveys as you described is not correct because surveys with large sample size will dominate the analysis regardless of the country size. I would advise you to take another look at the posted discussions on this issue as we mentioned earlier.

Previous Topic: How to use weights when using different samples of women
Next Topic: PSU and Uganda IR data
Goto Forum:
  


Current Time: Sat Dec 7 10:23:17 Coordinated Universal Time 2024