| Weighting data after merging survey rounds with different levels of representation [message #11566] |
Wed, 11 January 2017 01:11  |
jswindle
Messages: 5
Registered: September 2016
|
Member |
|
|
Hello DHS experts and other forum users,
My problem relates to weighting after merging survey rounds.
I have merged the IR for the Malawi 2000, 2004, and 2010 surveys. When I weight this data, I am unsure how to best create the strata given different levels of representation across the surveys. The 2000 and 2004 surveys were representative across the 10 largest districts and across the three regions. The 2010 survey was representative across all 27 districts, as well as across the three regions.
I am combining these survey data with district level, time-varying data for foreign aid.
Currently I am doing the following, but am unsure about the commands. I am especially looking for guidance about the fifth through seventh lines below, which begin with "egen mw_00_strata..."
Thank you.
* Weight the dataset
generate weight = v005/10000000
recode survey (2000=1) (2004=2) (2010=3)
egen clusters=group(survey v021), label
egen mw_00_strata = group(survey region urban), label
egen mw_04_strata = group(survey region urban), label
egen mw_10_strata = group(survey region district urban), label
gen strata = .
replace strata = mw_00_strata if year==2000
replace strata = mw_04_strata if year==2004
replace strata = mw_10_strata if year==2010
svyset clusters [pweight=weight], strata(strata) singleunit(centered)
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
