Home » Data » Weighting data » Weighting Data for Pooled Indonesian DHS Dataset
| Re: Weighting Data for Pooled Indonesian DHS Dataset [message #10081 is a reply to message #10056] |
Thu, 23 June 2016 10:55  |
Bridgette-DHS
Messages: 3043
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
Normally I would not recommend that you apply the wealth index from one survey to another survey. However, that may be best here. That is, I suggest that you look up (on www.dhsprogram.com) the formula for the wealth index in the 2007 survey, and then apply it to the 2012 survey, with the same cut points for the quintiles (of equal size in 2007 but not in 2012). You can then do a decomposition of the change in mortality, with components for change in the wealth-specific mortality rates and change in the distribution of wealth (and, depending on how you do it, a component for the interaction between the two).
For that purpose I am pretty sure that there is no need to re-normalize the weights, but that may depend on what decomposition procedure you use. I would first do the analysis with the original weights. I would then repeat it with the weights in one survey changed by some arbitrary factor, such as 2, just to see whether that affects your decomposition. If it does, then re-weighting should be considered. The options are somewhat arbitrary. The total weighted number of births in each survey, rather than the number of women, could be what you want to match with the population, but then you get into the potential role of changes in wealth-specific fertility as a source of change in mortality. This is really your decision and I am not comfortable making a recommendation.
|
|
|
|
Current Time: Mon Apr 29 02:53:40 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
