I thought about this once and didn't clearly come out one way or the other.
It is my impression that the literature does not use weights. Because the
propensity scores are intermediate calculations , it shouldn't matter much
one way or the other.
Matt
-----Original Message-----
From: Andres Vork [mailto:[email protected]]
Sent: Friday, January 17, 2003 1:03 AM
To: [email protected]
Subject: st: psmatch and sampling weights
Hello,
How should I take into account sampling weights when using -psmatch- command
(or one-to-one matching based on a propensity score, in
general)?
I have a survey data where both treated and non-treated observations come
from a larger population of unemployed people with
different sampling weights.
1) Should I use the sampling weights when estimating a propensity score
(with a logit model, for example)? Should I use -svylogit-?
2) And how should I use the sampling weights when trying to generalise my
results to the population?
Any hints or references to literature?
Thanks,
Andres Vork
PhD student in Economics
Estonia
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
