Practical Issues to Consider Before Using Propensity Score Analysis

Abstract

Research methodology involving a comparison group and an experimental group is commonly used in many academic disciplines, including education, psychology, business, and medicine where random assignment of participants to groups is not possible. Generally, quasi-experimental or regression discontinuity designs have been used when true experimental designs are not possible. More recently, propensity score analysis has been suggested to address issues associated with the analysis of covariance approach used in quasi-experimental designs, especially selection bias. Key issues not addressed in the use of propensity score analysis are discussed.