PSMatching: A SAS Macro for Treatment Effect Estimation and Missing Data Imputation Based on Propensity Score Matching

Matching estimators based on the propensity score are widely used in the field of treatment effect evaluation and a viable technique also for missing data imputation. This paper describes an implementation of the technique in SAS®, a statistical software where only limited implementations of it are currently available. The user can choose among the most common variants of the matching algorithms (nearest neighbour-, stratification- and kernel matching), and the main pre- and post estimation analyses proposed in the literature (reduction of standardised bias, Sianesi test, Ichino-Becker test on the balancing hypotheses, Lechner bounds). To these, two additional diagnostics are proposed in order to better monitor some aspects of the matching process. A validation of the procedure is made using the available STATA tools as a benchmark, with both artificial data and data already used in the literature.