Hi Casey,
Matching on the outcome is certainly problematic....imagine if you got perfect balance
(which you would hope to do), then that would also imply a 0 treatment effect!
What wouldn't be problematic (and in fact would be good) is if they were actually
matching on a pre-treatment measure of the outcome. E.g., in education studies there is
often pre-treatment test scores available, where those same tests form the outcome value
as well (but then measured post-treatment). This kind of matching on the outcome
variables (but measured before the treatment) can be very effective.
Thanks
Liz
On 7/27/09 2:56 PM, "Casey Klofstad" <klofstad(a)gmail.com> wrote:
Hey all-
In a paper I'm reviewing one of the covariates used in a 1-to-1
matching procedure ends up also being used as the dependent variable
in the final analysis. From a theoretical point of view this seems
odd, since the variable is being treated as both a pre- and
post-treatment variable. But, is there any empirical reason to be
worried about this?
Thanks much,
-c
--
Casey A. Klofstad
University of Miami
Department of Political Science
Coral Gables, FL
klofstad(a)gmail.com
http://www.as.miami.edu/personal/cklofstad/
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe:
http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation:
http://gking.harvard.edu/matchit/
Browse/Search List Archive:
http://lists.hmdc.harvard.edu/lists/matchit/