 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: propensity score matching
At 03:51 PM 5/8/2008, Susan wrote:
I would welcome some advice on the use of psmatch2 - I seem to have
become confused somewhere along the line.
I have estimated a propensity score using logistic regression. I now
want to match on this and then use conditional logistic regression
to assess the effect of treatment on outcome. I have approximately
11000 subjects on treatment A and 2300 on treatment B. I was hoping
to create a matched data set with A:B in the ratio n:1 where 1<n<4.
I had hoped that psmatch2 would do this with nearest neighbour
matching without replacement. However I seem to be having some
problems or am misunderstanding the help.
1. I was under the impression that "standard" nearest neighbour
matching is done without replacement (matches removed) and yet
psmatch2 creates the _weight for which I get very high values ( e.g
n=45) . this suggests that a subject is used 45 times as a match (
there are other subjects available with the same propensity score
but not used).
2. I would like to do 4:1 matching but cannot seem to find a way of
identifying the id's of all the matches - is this possible. If not
can I get the outcomes for all the matched pairs so that I can compute OR's.
[...]
I can't address all the issues, but may I suggest mahapick or
mahascores in my mahapick package. This will give you as many nearest
neighbors as you desire. BUT it does matching with replacement. You
can, however, take the results and do you own juggling to obtain a
matching-without-replacement result.
HTH
--David
*
* 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/