Strengthening Causal Inference in Extension Evaluations: Practical Application of Propensity Score Matching with Secondary Data

This proposal introduces a practical method for evaluators to assess program outcomes when randomized controlled trials are not feasible. We present a step-by-step application of Propensity Score Matching to create balanced treatment and control groups using secondary data. Though an example comparing physical and mental health satisfaction as an outcome variable between youth development program participants and the general population, the presentation demonstrates how secondary datasets can substitute control groups while maintaining analytical rigor. Using MatchIt R package, the authors detail the process from selecting datasets to analyzing matched samples. Our presentation will focus on secondary sample selection, covariate identification, handling missing data using imputation, propensity score models, caliper determination, and balance diagnostic methods. This approach enhances the credibility of non-experimental designs and offers a replicable model for evaluators working with limited resources. The presentation will serve as a valuable guide for those seeking evidence-based outcome comparisons in community-based programs.

Anne Iaccopucci; Nicole Marshall-Wheeler; Steven Worker