PS Match and Mahalanobis Distance

SG1993 — Wed, 06 Mar 2024 20:28:36 GMT

Hi,

I am running the following code and am getting the error that "The variable Gender in the STAT=MAH option is in the class list".

proc psmatch data=drugs region=cs;

class Drug Gender;

psmodel Drug(Treated='Drug_X')= Gender Age BMI;

match method=greedy(k=1)

distance=mah(lps var=(Age BMI Gender)) caliper=0.2

stat = lps

weight=none;

assess lps var=(Gender Age BMI);

output out(obs=match)=OutEx7 matchid=_MatchID;

run;

I wanted to see of there was a way to include categorical variables in the Mahalanobis distance calculation in the above procedure. Any suggestion would be really appreciated.

Thanks!

Re: PS Match and Mahalanobis Distance

mkeintz — Thu, 07 Mar 2024 02:07:34 GMT

If GENDER were taken out of the CLASS statement, and if gender were coded as, say 0 and 1, wouldn't it then serve your purpose? I believe that's all you would need to account for the confounding of GENDER with AGE & BMI in assignment of DRUG.

topic Re: PS Match and Mahalanobis Distance in New SAS User

PS Match and Mahalanobis Distance

Re: PS Match and Mahalanobis Distance