| Re: Weighting for the Pakistan Special 2019 DHS on Stata [message #25091 is a reply to message #25074] |
Tue, 30 August 2022 11:20   |
Bridgette-DHS
Messages: 3117
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
For your first item, the answer is "yes".
For the second item, there is a bias in the national estimate for 2019 if you include GB and AJK in a national estimate because they were over-sampled. You can get an unbiased estimate for Pakistan MINUS those areas, and an unbiased estimate for those areas, but you cannot combine them for a national estimate. (That is the position of the implementing agency in Pakistan and our samplers; I cannot explain it further.) If comparing successive surveys in Pakistan, you should work with the same geographic areas. However, as I understand it, the population of those areas is a very small part of the national population. There have been other countries, such as Mali, in which small areas (in terms of population, not necessarily in geographic size) have dropped or included in successive surveys based on the security situation at the time of the survey. It's unfortunate but unavoidable.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
