Health Services Research (HSR) Methods
Print Page

Matched Propensity Score Analysis
Related Terms: Propensity Score

A statistical technique designed to construct a comparison group similar to the treatment group with respect to the observable characteristics of subjects. The propensity score expresses the probability of being in one group as opposed to another, based on observable characteristics prior to an intervention. Assuming that all variables associated with assignment to treatment are observed and measured, the propensity score can be used to match individual observations with the match(es) that are closest, while eliminating those that are the worst. This approach weeds out the irrelevant observations that are poor matches for members of the treatment group more efficiently than attempting to construct a relevant comparison group by matching on one or two observable characteristics.

Further Reading
Smith, J. ''Evaluating Active Labor Market Policies: Lessons from North America, MittAB-Schwerpunktheft2000: Evaluation Aktiver Arbeitsmarktpolitik, Nuremberg:.'' Vol. IAB, 2000, pp.345-356.

Dehejia, R.H. and S. Wahba. ''Causal Effects in Nonexperimental Studies: Reevaluating the Evaluation of Training Programs.'' Journal of the American Statisitical Association Vol. 94, No. 448, 1999, pp.1053-1062.

Rosenbaum, P.R. and D.B. Rubin. ''The Central Role of the Propensity Score in Observational Studies for Causal Effects.'' Biometrika Vol. 70, 1983, pp.41-55.

Rubin, D.B. and N. Thomas. ''Characterizing the Effects of Matching Using Linear Propensity Score Methods with Normal Distributions.'' Vol. 79, 1992, pp.797-809.

Rubin, D.B. and N. Thomas. ''Matching Using Estimated Propensity Scores: Relating Theory to Practice.'' Biometrics Vol. 52, 1996, pp.249-264.