The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Using normalized weight in analysis (What if normalized weight in all cluster is 1)
Using normalized weight in analysis [message #18434] Sun, 01 December 2019 09:40 Go to next message
kamal.chaulagain123@gmail is currently offline  kamal.chaulagain123@gmail
Messages: 4
Registered: January 2019
Hello everyone!!
I am doing an analysis with the cluster survey data. The normalized weight in all cluster is calculated to be 1. There is no difference in weighted and unweighted calculation. In this context, how can be analysis done? Is it a self -weighted data? I will be thankful for your help.
Re: Using normalized weight in analysis [message #18516 is a reply to message #18434] Fri, 13 December 2019 07:17 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3114
Registered: February 2013
Senior Member

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

If you are using pweights in Stata, there is no need to normalize. Stata will automatically adjust the weights to have a mean value of 1.

All cases (households or women or men) in the same cluster always have the same weight. It will not be 1, but it will be the same value. Have you divided the household (etc.) weight by the cluster weight? That's not a correct way to normalize.

If you want to normalize the weights, i.e. to have a MEAN weight of 1, then calculate the mean weight for the entire sample (e.g. witn "summarize hv005") and divide hv005 by that value.

However, to repeat, there is no good reason to go through that step. You will get the same estimates whether or not you do that.
Previous Topic: PSU and Strata identifiers - DV module Nepal
Next Topic: Deriving district population size from DHS weights?
Goto Forum:

Current Time: Thu Jul 18 13:25:57 Coordinated Universal Time 2024