Propensity Score Matching
Propensity score matching is a method that estimates the probability of taking a treatment given a vector of observed variables. It is a method to draw causal inference on observational data which many social science studies use. Through matching and balancing samples, propensity scores help adjust for selection bias and estimate counterfactual effects, and therefore obtain average treatment effect, average treatment effect for the treated (ATT), and average treatment effect for the untreated.
Essential Reading:
- Books -
Morgan, Stephen L. and Christopher Winship. 2007. Counterfactuals and Causal Inference: Methods and Principles for Social Research. New York: Cambridge University Press.
- Articles -
Harding, David J. 2003. "Counterfactual Models of Neighborhood Effects: The Effect of Neighborhood Poverty on Dropping Out and Teenage Pregnancy." The American Journal of Sociology 109(3): 676-719.
Heckman, James J. 2005. "The Scientific Model of Causality." Socialogical Methodology 35:1-97.
Morgan, Stephen L. and David J. Harding. 2006. "Matching Estimators of Causal Effects: Prospects and Pitfalls in Theory and Practice." Sociological Methods and Research 35:3-60.
Rosenbaum, Paul R. and Donald B. Rubin. 1983. "The Central Role of the Propensity Score in Observational Studies for Causal Effects." Biometrika 70: 41-55.
- 1984. "Reducing Bias in Observational Studies Using Subclassfication on the Propensity Score." Journal of the American Statistical Association 79:516-24.
Rubin, Donald B. 1973. "Matching to Remove Bias in Observational Studies." Biometrics 29:159-83.
Implementation note:
Sianesi, B. 2001. "Implementing Propensity Score Matching Estimators with STATA."
Becker and A. Ichino. 2002. "Estimation of average treatment effects based on propensity scores." Stata Journal 2: 358–377.
Zhao, Zhong. 2005. "Sensitivity of Propensity Score Methods to the Specifications."
