Critical Issues in Propensity Score Matching

• Activity Number: 632
• Type: Topic Contributed
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Health Policy Statistics Section
• Author(s): Haiyan Bai* and Wei Pan
• Companies: University of Central Florida and Duke University
• Keywords: propensity score ; propensity score matching ; observational studies ; causal inference ; selection bias

Abstract

Existing research in the literature has vigorously studied the effectiveness of bias reduction for observational studies using various propensity score matching methods. However, some crucial issues still exist without conclusive answers, such as whether propensity score matching should be done with or without replacement, how much overlapping (or common support) between treatment and control groups is sufficient, and how influential the sample ratio between treatment and control groups is. Regarding these concerns, a systematic investigation on the above issues is conducted using 60 trials applying to commonly used propensity score matching methods. This study contributes to the literature by suggesting critical points for crucial unsolved issues of propensity score matching. Researchers can use it as a guideline for selecting propensity score matching methods considering their sample situations to increase the validity of causal inference.