Propensity score analysis methods aim to reduce bias in treatment effect estimates obtained from observational studies, which are studies estimating treatment effects with research designs that do not have random assignment of participants to conditions. The use of propensity scores to reduce selection bias in nonexperimental studies was proposed by Rosenbaum and Rubin and was connected to earlier work by D. Rubin on matching methods for selecting an untreated group that was similar to the treated group with respect to covariates. The chapter details the effects of confounding on the effect of intervention. It outlines the issue of propensity score matching as an effective way to deal with confounding. The propensity score reduces all the information in the predictors to one number, which greatly simplifies analysis. For example, matching based on multiple covariates to reduce selection bias can be simplified to matching based on the propensity score.