The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Clustered Standard Errors
Re: Clustered Standard Errors [message #22130 is a reply to message #22115] Fri, 05 February 2021 09:32 Go to previous messageGo to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3203
Registered: February 2013
Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Your Stata code, using "egen group", looks fine. The only way to include the stratum adjustment is with svyset.

I don't quite understand the role you want for "district". The purpose of svyset is to adjust for the design effect in multi-stage samples. For DHS data, the clusters are the Primary Sampling Units (PSUs). The strata are essentially subpopulations within which separate samples are drawn. The adjustments for clusters and strata work in opposite directions, but only affect the standard errors of the estimates. Districts are administrative units that usually have no direct role in the design of the sample.

However, you can conceptualize a multi-level analysis in which, say, respondents are level 1, clusters are level 2, and districts are level 3. The justification would be that individuals within the same district are more similar than individuals in different districts, and you have some district-level covariates. We encourage the use of spatial covariates, but at the finest level of aggregation, which would be the cluster rather than the district. Please clarify.

 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: Confirming correct survey design
Next Topic: What is the difference between v005 and sweight DHS India?
Goto Forum:
  


Current Time: Sat Dec 7 21:39:14 Coordinated Universal Time 2024