Propensity score matching

Typing up an observation:

I had one old data and I updated two categorical variables (black and asian variables) in the new data.  The change was only in one group in the data (there were 10 groups all together).  Theoretically I would expect the matching results (in terms of means of the two variables) to change only in that group.  The changes also happened in two other groups.

proc psmatch data=psm region=cs;
where &outcome ne .;
class CP_FLAG districtname SCHOOLNAME ;
psmodel CP_FLAG(Treated="Y")= &exactvar &predictors;
match method=greedy(k=1)/*(order=random)*/ exact=districtname stat=lps caliper=&caliper;
output out(obs=match)=outgs lps=_Lps matchid=_matchID;
proc sort data=outgs;by _matchID;run;

Leave a Reply