Propensity score analysis is a statistical method used in nonrandomized studies to compare treatments after removing selection bias due to observed baseline covariables. To the extent that all true confounders are observed and selection bias removed, unbiased estimation of the average treatment effect can be made.

Propensity score analysis is a two-stage process. First, a logistic regression model predicting treatment assignment from available baseline potential confounders is used to assign each patient a propensity score representing the probability that he or she would receive treatment (vs. control, no treatment), regardless of whether treatment is actually received. Second, treated and nontreated patients are compared on the outcome of interest after conditioning on the propensity scores— typically through stratification or matching.

Propensity score analysis is often preferred over traditional ...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles