topic PROC PSMATCH for three or more groups in Statistical Procedures

PROC PSMATCH for three or more groups

dac_js — Thu, 02 Nov 2023 13:09:22 GMT

Hello SAS Community,
I am currently working on a project where I have more than two groups I would like to compare using propensity score matching. I am familiar with using PROC PSMATCH for a typical treatment vs. control setup, but I am unsure how to approach it for multiple group comparisons. Is there a way to directly compare more than two groups using PROC PSMATCH? If not, what is the best practice? Should I perform pairwise comparisons and then adjust for multiple comparisons?
Are there alternative techniques or procedures in SAS that might be better suited for this type of analysis? Any guidance, sample code, or references would be greatly appreciated!
Thank you in advance!

Re: PROC PSMATCH for three or more groups

sbxkoenk — Sat, 04 Nov 2023 13:58:44 GMT

@dac_js wrote:

... , but I am unsure how to approach it for multiple group comparisons. Is there a way to directly compare more than two groups using PROC PSMATCH?

Hello,

The short answer is: NO.

PSMATCH only supports (matching) two groups, treatment and control. There is no provision for matching three (or more) groups. It's kind of baked into the code to match treatment and controls (two groups). So, ... binary instead of multinomial.

I'm sorry I am unable to give you better news at this time.

I am not sure either how you need to proceed with your analysis.

BR, Koen

Re: PROC PSMATCH for three or more groups

dac_js — Wed, 08 Nov 2023 02:39:51 GMT

Thank you for the response!