Consider Propensity Scores to Compare Treatments

May 08, 2006

Overview

The underlying question when comparing treatments is usually whether an individual would do better with treatment X than they would with treatment Y. There are often practical and theoretical problems, however, in giving people both treatments and then comparing data. This paper presents the use of propensity score matching as a methodology that can be used by programs with large amounts of data to compare the effectiveness of different treatments. The method is applied to answer two questions: 1) “Should examinees take a college admissions test near or a few years after graduation?” and 2) “Do accommodated students receive an unfair advantage?” Data from a large admission testing program is used.