The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Dataset use in Stata » Stata weights vs CSPro weights (CSPro weighted value is different with stata value)
Stata weights vs CSPro weights [message #17895] Mon, 08 July 2019 02:06 Go to next message
coach1327 is currently offline  coach1327
Messages: 1
Registered: April 2019
Hi, I am currently doing a study on domestic violence using the Philippines 2017 DHS datasets. When I do frequency for CSPro using the following:
Universe = v502 in 1:2
weight = d005/1000000

I get the published value of 11,558.

But doing the same variable in Stata using [aw=d005]
I get a value of 13,215, which is also the UNWEIGHTED value in CSPro.

What am I missing in Stata?

Thank you guys in advance!

Re: Stata weights vs CSPro weights [message #17923 is a reply to message #17895] Thu, 18 July 2019 08:48 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 1965
Registered: February 2013
Senior Member
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

There is not really an issue here. When you use d005 in Stata as a pweight, you should get exactly the same results that CSPro gets. The difference you describe has to do with whether the weight has been normalized in such a way that the weighted total equals the unweighted total. Stata always normalizes the weights (with pweight), without you asking it to. I encourage you to check this by using Stata with d005 to match something in the DV chapter in the main report. Let us know if you have trouble getting a match.
Previous Topic: Additional versions in Zip File
Next Topic: Duplicated label categories
Goto Forum:

Current Time: Sat Oct 24 11:04:51 Eastern Daylight Time 2020