Assessment and Research Methods
Generating Post-Hoc Comparison Groups for Persistence Research
Persistence research is often limited by the lack of a comparison group for testing the effect of participation in a program or receipt of a particular service. Predicted values or propensity scores provide a post-hoc mechanism to generate a comparison group from basic enrollment data by matching non-participants to the program participants. The benefits and limits of generating a post-hoc comparison groups with predicted values or propensity scores to evaluate progress and persistence outcomes are reviewed in the following slides. Included are instructions for generating predicted values in SPSS or STATA.
This information was presented with some modifications by Jennifer Lowman, Ph.D. at the following conferences and organizations:
- Association for Institutional Research, May 2014
- NASPA: Assessment and Persistence Conference, June 2014
- National Council of Juvenile and Family Court Judges, June 2014