Propensity Score Matching In R

See previous post on propensity score analysis for further details.

Propensity score matching in r.

Rosenbaum and rubin 1983 is the most commonly used matching method possibly even the most developed and popular strat egy for causal analysis in observational studies pearl 2010. Propensity score matching in spss provides spss custom dialog to perform propensity score matching. Here i use a loess smoother to estimate the mean of each covariate by treatment status at each value of the propensity score. Proper citations of these r packages is provided in the program.

Propensity score matching psm paul r. Matching is based on propensity scores estimated with logistic regression. If matching is done well the treatment and control groups will have near identical means of each covariate at each value of the propensity score. According to wikipedia propensity score matching psm is a statistical matching technique that attempts to estimate the effect of a treatment policy or r bloggers r news and tutorials contributed by hundreds of r bloggers.

The concept of propensity score matching psm was first introduced by rosenbaum and rubin 1983 in a paper entitled the central role of the propensity score in observational studies for casual effects statistically it means. It is used or referenced in over 127 000 scholarly articles 1. Once we implement matching in r the output provides comparisons between the balance in covariates for the treatment and control groups before and after matching. Psm attempts to reduce the bias due to confounding variables that could be found in an estimate of the treatment effect obtained from simply comparing outcomes among units that received the treatment versus those that did not.

The package provides functions for multivariate and propensity score matching and for finding optimal balance based on a genetic search algorithm. The output below indicates that the propensity score matching creates balance among covariates controls as if we were explicitly trying to match on the controls themselves. Below is an example using the four covariates in our model.