Betas from PSMATCH

[ptimusk]

I have not really been lectured on pairs analysis and am trying to use PROC PSMATCH but would now like to use an output from the PROC to get the Betas for my written report. Is this possible or am I not approaching this with the right concept? I thought there were Betas in a logistic regression. I have my diagnostic p values from before and after matching but would like a table of Betas too.

[cyrus · Accepted solution]

PSMATCH doesn't create a parameterEstimates table.  An alternative is to run the equivalent model in PROC LOGISTIC, which provides the betas, and then use the PSDATA Statement in PSMATCH to import those propensity scores.  See  Example 100.8 Matching with Precomputed Propensity Scores.