The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Using iweights for districts
Using iweights for districts [message #10986] Fri, 14 October 2016 23:58 Go to next message
shayankhan is currently offline  shayankhan
Messages: 16
Registered: September 2016
Location: Karachi, Pakistan
Member
Hi,

I want to run the xtlogit model with fixed effects to control for district level heterogeneity. I was told to use iweights for this as the xtlogit command does not accept pweights. My question is two-fold:

1-How do I calculate these iweights?
2-Would using only the weights and not accounting for the survey design (using the svy estimation) affect the validity of my estimation?

Note:- It was recommended to me by Dr.Kit Baum, a highly respectable figure and well known econometrician to use the xtlogit model with iweights rather than using the svy:logistic estimation with district dummies as this approach might not give consistent estimates due to the incidental parameter problem.

Thanks.
Re: Using iweights for districts [message #11001 is a reply to message #10986] Mon, 17 October 2016 14:11 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 2537
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Quote:
Usually, to use iweights, you just insert "[iweight=v005/1000000]" into the estimation command, before the comma (options come after the comma).

Omitting the other adjustments in the svyset command, for clusters and stratification, will affect the standard errors and confidence intervals, not the estimates. I believe that xtset and svyset are not compatible in the current version (14) of Stata. I'm not 100% sure about this--I hope other users can contribute their experience.
Re: Using iweights for districts [message #11002 is a reply to message #11001] Mon, 17 October 2016 14:49 Go to previous message
shayankhan is currently offline  shayankhan
Messages: 16
Registered: September 2016
Location: Karachi, Pakistan
Member
Thanks Tom. I actually decided to go along with PSU fixed effects as each PSU within a district would capture more heterogeneity. Appreciate your input throughout my queries.
Previous Topic: sample weights when using a subsample
Next Topic: Sample weights disappear after merge household with childs recode
Goto Forum:
  


Current Time: Thu Jun 30 09:01:56 Coordinated Universal Time 2022