On Tue, Jan 26, 2016 at 4:36 AM, 서연수 <yssshinhwa(a)naver.com> wrote:
Hello,
I had two questions regarding using the CEM package in Stata that I would
be very greatful if someone helped me with.
1) In the 2009 Stata journal (cem: Coarsened exact matching in Stata),
specifically the part of 4.7 Using cem to improve other matching methods,
it says that after CEM, PSM(propensity score matching) can be used to
obtain better inferences.
I was wondering if this applied to propensity score weighting as well as
propensity score weighting has the disadvantage of producing "bad" weights
for outliers I believe. (In short, I want to know if I can conduct CEM
and then using the data, conduct propensity score weighting-in order to get
rid of much bias as possible)
Moreover, if this works, would I add the same covariates used in CEM and
propensity score weighting in the final analysis model?
Yes, you can use propensity score weighting after using CEM to adjust for
any residual confounding due to the covariates. Using CEM before this
weighting will also help to remove outliers that would cause large weights.
In the weighting model, you should include any covariates that are not
exactly matched in CEM and you think might require additional control.
2) Also, when using CEM, the weights are incorporated in following
analysis using the "iweights" option in Stata,
However, in propensity score weigthing, "pweights" seem to be the norm.
Is there a reason why the two methods use different weights?
The results of iweights and pweights differ for both methods and I was
confused.
I am not familiar with weights, so I might be asking a silly question but
I was curious.
Each of the weighting commands in Stata varies in implementation from
command to command. Typically there is very little difference between them.
In cases where pweights are allowed, this is usually a fine way to go. You
can usually combine PS weights and CEM weights by multiplying them together
to create a new set of weights and pass those to pweight.
Hope that helps!
Cheers,
Matt
~~~~~~~~~~~
Matthew Blackwell
Assistant Professor of Government
Harvard University
url:
http://www.mattblackwell.org
I know it is a lot of questions, but would really appreciate it if someone
could help me out.
Thank you in advance.
-
--
cem Mailing List, served by HUIT
Send messages: cem(a)lists.gking.harvard.edu
[un]subscribe Options:
http://lists.gking.harvard.edu/?info=cem
More information on cem:
http://gking.harvard.edu/cem
Cem mailing list
Cem(a)lists.gking.harvard.edu
To unsubscribe from this list or get other information:
https://lists.gking.harvard.edu/mailman/listinfo/cem