The PSMATCH Procedure

Examples: PSMATCH Procedure

In practice, the outcome data for an observational study might or might not be available at the time that a propensity score analysis is done. You can handle these situations as follows:

When the outcome data are not yet available, you might not need to retain the covariate data for all individuals in the study in order to carry out the outcome analysis. For example, if you use the matching method for propensity score analysis, only the matched units are needed for follow-up. Retaining only the matched units reduces the cost of the study (Stuart 2010, p. 2). The clinical trial described in Getting Started: PSMATCH Procedure is an example in which outcome data are not yet available.
When the outcome data are available at the time of the propensity score analysis, they should not be used in the analysis (Stuart 2010, p. 2). In Example 95.4, the question is whether taking a music class improves grade point averages, and the grades together with other measures are available when the propensity score analysis is done at the completion of the school year.

The examples in this section illustrate the main methods for propensity score analysis that are available in the PSMATCH procedure. For simplicity, the examples use only a few variables. In practice, propensity score analysis often involves many more variables.