Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
No women appear in more than one round. The samples in successive surveys are statistically independent. Even the sample clusters are different.
As shown below, there are no duplicates in the IR files for any of the India surveys. Duplicates would be indicated by any values greater than 1 when the number of cases with each ID are added up. Earlier posts have stated that v024 (region) must be included in the unique identifiers. I collapsed over "v024 v001 v002 v003". In another run I collapsed over "caseid" and got exactly the same result. There are no duplicates in the India IR files.
. local lpvs 23 42 52 74 7B
.
. foreach lpv of local lpvs {
2.
. use caseid v024 v001 v002 v003 using "C:\Users\26216\ICF\Analysis - Shared Resources\Data\DHSdata\IAIR`lpv'FL.DTA", clear
3.
. gen dups=1
4.
. collapse (sum) dups,by(v024 v001 v002 v003)
5.
. tab dups
6.
. }
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
