Interval Matching: Propensity Score Matching Using Case-Specific Bootstrap Confidence Intervals

• Activity Number: 415
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Wei Pan*+ and Haiyan Bai
• Companies: University of Cincinnati and University of Central Florida
• Keywords: propensity score matching ; propensity score analysis ; caliper matching ; bootstrap

Abstract

Propensity score matching is an essential procedure in propensity score analysis. The current procedure in propensity score matching is to match each of the treated cases with one or more control cases based on closest propensity scores which are the point estimates of the likelihood of the cases to be assigned into the treatment group. The problem with this procedure is that it is difficult to establish a criterion to evaluate the closeness of the matched cases without knowing the standard error of the estimate of each case's propensity score. Cochran and Rubin (1973) suggested using a caliper band to avoid "bad" matches. However, this fixed or case-invariant caliper band still cannot address the standard errors of the estimates of propensity scores. The present study proposes interval matching to capture the standard error of the estimate of the propensity score for each case using case-specific bootstrap confidence intervals. In this proposed interval matching, if the confidence interval of a treated case overlaps with that of one or more control cases, they will be taken as "good" matches. The implementation of interval matching is illustrated with an empirical example.