Both PROC CAUSALTRT and PROC PSMATCH use the logistic regression model to conduct propensity score analyses. If you specify the same propensity score model for these two procedures, you will get the same set of model fitting results and hence the same set of estimated propensity scores. Both procedures enable you to estimate the average treatment effect (ATE) and the average treatment effect for the treated (ATT). In addition, both procedures provide numerical and graphical tools for examining covariate balancing. Nonetheless, the two procedures have several distinct features that would suit different modeling situations.
You should consider using PROC CAUSALTRT in the following situations:
You have a reliable outcome model to use for estimating causal effects by either an outcome regression method or a doubly robust method (for more information, see the METHOD= option and the MODEL statement in Chapter 39, The CAUSALTRT Procedure).
You are confident about the appropriateness of the specified propensity score model and want to estimate the standard errors of causal effects by taking the estimation of propensity scores into account.
You should consider using PROC PSMATCH in the following situations:
You need to use the propensity score matching or stratification method (for more information, see the MATCH statement and the STRATA statement in Chapter 101, The PSMATCH Procedure).
You need to explore different propensity score models to achieve adequate covariate balance before conducting an outcome analysis (for more information, see the section Using PROC PSMATCH Repeatedly to Ensure Proper Covariate Balancing).