Propensity Score Analysis

Propensity score analysis is a technique for estimating the causal effect of a treatment in an observational study. Although randomized experiments are the ideal method for estimating the causal effect of a treatment—because randomization ensures that, on average, the distribution of both observed and unobserved characteristics are the same for treated and untreated units—there are many cases where randomized experiments are unethical or impractical. In an observational study, unlike in a randomized experiment, the researcher has no control over treatment assignment. As a result, due to self-selection into treatment or other nonrandom aspects of the treatment assignment, there may be systematic differences between the treated and control groups that can bias the estimate of treatment effects. Using propensity scores to match treated units to similar ...