The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Weight about pooling 2003,2008,2013,2018 BR dataset in Nigeria (How to weight pooling dataset)
Re: Weight about pooling 2003,2008,2013,2018 BR dataset in Nigeria [message #18980 is a reply to message #18938] Mon, 30 March 2020 15:54 Go to previous messageGo to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

I don't really see what is gained by pooling these surveys into one file. As a conceptual matter, rather than a technical one, a consolidation of four surveys from the same country but at different times does not seem to me to be a meaningful population. It might make sense if nothing was changing, but because most characteristics ARE changing over time in Nigeria, wouldn't you want to analyze each survey separately?

Sometimes I do pool surveys into one file, but only in order to make it easier to make comparisons between the surveys and to graph trajectories over time. For that purpose, no re-weighting is needed. It helps to number the surveys 1, 2, 3, 4, although v007 can be used to distinguish the surveys. Again, this would be done to make data processing easier; I would not run models that ignored the identifiers for the four specific surveys.

For what I think you want to do, I believe the re-weighting is correct. To check it, you want to see whether the total weight in each round is proportional to the population in each round.

The number of women age 15-49 in Nigeria at the time of the survey would be appropriate for re-weighting the IR file, but I see that you are using the BR file. The BR file has one record for every child that appears in a birth history, whether living or dead. It would be better to re-weight using the IR files and then reshape (wide) the b variable to get one record for each child.
 
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: Sample weights and stratification - Nigeria 2008 and 2018
Next Topic: Weighted result's total gives different total value with unweighted one.
Goto Forum:
  


Current Time: Thu Mar 28 20:52:18 Coordinated Universal Time 2024