Could you elaborate on what you mean by matching on three groups?  Why couldn't you compute the propensity scores based on all the relevent variables and then match on those scores using your macro or the FUZZY extension command?

Jon Peck
Senior Software Engineer, IBM
peck@us.ibm.com
new phone: 720-342-5621




From:        Dale Glaser <glaserconsult@sbcglobal.net>
To:        SPSSX-L@LISTSERV.UGA.EDU
Date:        07/07/2011 01:04 PM
Subject:        [SPSSX-L] Matching/propensity scores for 3 groups
Sent by:        "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>




Greetings.......I have used the SPSS macro (that I believe is discussed in an article in Behavior Research and Methods) that uses propensity scores (via binary logisitic regression) for matching two groups; however, I am currently on a study that necessitates matching on three groups.  So I was curious what any of you have done for matching three groups using SPSS (I will be matching on sex, Length of stay, and # of comorbidities)?  Do you think it would be problematic to use propensity scores from a multinomial logistic regression (and insert such in the macro?).  Or will that just give me two sets of propensity scores (e.g., grp 1 vs grp 2; grp 1 vs grp 3, etc.)?
 
Otherwise, I do have Stata but have not used it for matching purposes.......if SPSS is not equipped for 3 group matching, do you know if Stata has a package that does such?
 
thank you very much in advance for any suggestions.....dale
 


Dale Glaser, Ph.D.
Principal--Glaser Consulting
Lecturer/Adjunct Faculty--SDSU/USD/Alliant
3115 4th Avenue
San Diego, CA 92103
phone: 619-220-0602
fax: 619-220-0412
email: glaserconsult@sbcglobal.net
website:
www.glaserconsult.com