Creating a pooled NFHS dataset [message #30230] |
Sat, 19 October 2024 06:07 |
shivani
Messages: 12 Registered: March 2021
|
Member |
|
|
Hi,
I am trying to do a district level analysis for child undernourishment by pooling NFHS 4 and 5. Can I simply keep the variable list short and append the 'Kr" files for the two rounds. In doing so, I believe I'd need to create a unique district level id and create another "time variable' so as to get a pooled response with "time" variable denoting NFHS round 4 or 5.
Additionally, I needed to ask whether the same weights can be applied in this pooled dataset, to get the separate national level estimates for individual rounds of NFHS 4 and 5?
Awaiting your reply
Thanks
|
|
|
Re: Creating a pooled NFHS dataset [message #30248 is a reply to message #30230] |
Tue, 22 October 2024 08:20 |
Bridgette-DHS
Messages: 3202 Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
If you are looking at changes and differences between the two surveys, you can stay with the original weights. No need to modify them. Your "time" variable cold be the month and year of the interview, or just v008, the century month code for the date of interview. The interviews in both surveys were spread out over long intervals.
You need a variable for district that is the same in both surveys. Looking at the KR files, I see that in the NFHS-4, the variable is sdistri and the label is SDISTRI. In the NFHS-5, the variable is sdist and the label is SDIST. If you list those labels you will see that they are not the same. You will have to recode sdist into sdistri, or the reverse, or recode the two of them into a blended version. Good luck doing that.
|
|
|