Matching with Multiple Control Groups and Adjusting for Differences between the Groups

• Activity Number: 80
• Type: Invited
• Date/Time: Monday, August 8, 2005 : 8:30 AM to 10:20 AM
• Sponsor: Section on Health Policy Statistics
• Author(s): Elizabeth A. Stuart*+ and Donald B. Rubin
• Companies: Mathematica Policy Research, Inc. and Harvard University
• Keywords: causal inference ; propensity scores ; matching methods ; observational study

Abstract

When estimating causal effects using observational data, it is desirable to replicate a randomized experiment as closely as possible by obtaining treated and control groups with similar distributions of observed covariates. This goal often can be achieved by choosing well-matched samples of the original treated and control groups, thus reducing bias due to these covariates. However, sometimes the originally selected control units cannot provide adequate matches for the treated units. In these cases, it may be desirable to obtain matched controls from multiple control groups. Multiple control groups have been used to test for hidden biases in causal inference (Rosenbaum 2002); however, little work has been done on their use in matching or adjustment for these biases. In addition, there may be concern regarding systematic differences between the control groups. Here, we present a method that uses matches from multiple control groups and adjusts for potentially unobserved differences between the groups in the analysis of the outcome.