| Re: Svy, subpop [message #3952 is a reply to message #3942] |
Tue, 10 March 2015 08:34  |
Bridgette-DHS
Messages: 3230
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Specialist, Tom Pullum:
Here are the lines you need for a logit regression of stunting on the wealth index for children age 0-23 months.
* logit regression with stunting in months 0-23 as outcome
* use the PR file; KR file is limited to children living with mother
* usually limit to de facto residents, i.e. hv103=1
keep if hv103==1
* hc70 is the WHO haz score, already edited
gen stunting=0
replace stunting=1 if hc70<-2
replace stunting=. if hc70>600
* the cluster id is hv001=hv021
* Stata will normalize the weights; no need to divide by 1000000
* hv022 may be the strata variable but always check
* always a good idea to include one of the singleunit options
svyset hv001 [pweight=hv005], strata(hv022) singleunit(centered)
* it is normal to use hc1 (=hv008-hc18) as age
svy: logit stunting i.hv270 if hc1<24
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
