Omitted Variables, R-Square, and Bias Reduction in Matching: A Monte Carlo Study

• In Event: Advances in Propensity Score Analyses and Causal Inference

Abstract

Based upon a two-level structural equation model,
this simulation study examines how omitted variables affect estimation bias in matching. Six simulated cases of omitted variables are examined by manipulating level-1 and/or level-2 residual variances and $R^2$. Results show 1) Mahalanobis distance matching is less effective than propensity score matching; 2)level-1 matching is less sensitive than level-2 matching to omitted variables; 3) dual-matching (level-1 plus level-2 matching) is robust to omitted variable problem. This study can help researchers use appropriate matching strategy to reduce selection bias for program evaluation in math education.

Authors

• Qiu Wang, Syracuse University

• Kimberly S. Maier, Michigan State University

• Richard T. Houang, Michigan State University