sure, in the cem output there's a weight variable. just use it in your
zelig call. it will then weight some observations to zero; those where
pruned. it will also weight observations so that the number of treated and
controls are equalized within each stratum.
Gary
--
*Gary King* - Albert J. Weatherhead III University Professor - Director,
IQSS - Harvard University
GKing.Harvard.edu <http://gking.harvard.edu/> - King(a)Harvard.edu -
@kinggary<http://twitter.com/kinggary>- 617-500-7570 - Asst 495-9271 -
Fax 812-8581
On Sun, Apr 10, 2011 at 2:45 PM, Brandon Stewart
<brandonmstewart(a)gmail.com>wrote;wrote:
Noah,
There should be a way to do both. The one that I found the quickest was to
fix the matchit command. You can give it the argument
k2k=FALSE
(You can also pass it any arguments you pass CEM)
To fix the problem with Zelig taking CEM objects....I'm less sure. I bet
CEM has an object in the list that it returns which Zelig would take. Try
taking a look at the returned values and see if there is something you can
grab there...
Brandon
On Sun, Apr 10, 2011 at 2:33 PM, Noah Nathan <nlnathan(a)fas.harvard.edu>wrote;wrote:
Hi everyone,
For our replication paper we're trying to do matching with CEM and running
into some coding issues:
We want to do two things with CEM, one of which we can do with the cem()
command and one with the matchit() command, but we're trying to figure out
how to do both with only one or the other. We want both: (a) to be able to
use our matched data in zelig so we can do all the cool things with zelig,
like setx(), etc, after we run our final model with the matched data. We can
get information to pull into zelig from matchit() but not from cem(). (b)
When we use cem() it, by default, matches each treated case to ALL controls
in each treated case's stratum. In matchit() it by default only does a 1:1
match with the nearest control case within the stratum. (we have many more
C's than T's in the area of common support. going to 1:1 means throwing out
hundreds of observations)
Is there a way to pull cem() output into a zelig model? OR Is there a way
to change matchit()'s default settings to get the 1:(whole stratum) matching
when method = "cem"?
Thanks,
- Noah and Julie
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l