Treatment-Comparison Group Sample Size Ratio and Accuracy of Treatment Effect Estimates After Propensity Score Matching

• In Event: 28.030 - Analysis in Experimental Designs and Quasi-Experimental Designs With Propensity Score Matching

Abstract

Given that random assignment to treatment and comparison groups is typically not feasible in educational research, educational researchers may employ propensity score matching when evaluating the effectiveness of programs. There are currently few guidelines on the ideal treatment to control group sample sizes when propensity score matching. Therefore, data were simulated with a known treatment effect for purposes of evaluating the accuracy of treatment effect estimates following propensity score matching (Nearest Neighbor and Nearest Neighbor with 0.2 caliper) across seven treatment-comparison sample size ratios. Nearest Neighbor with 0.2 caliper and a 1:4 treatment to comparison group ratio resulted in the most accurate recovery of the treatment effect. Implications for researchers will be provided.

Authors

• Madison A. Holzman, James Madison University

• S. Jeanne Horst, James Madison University