Genetically Adjusted Propensity Score Matching: Closing the Gap Between Non-experimental Designs and True Experiments in the Social Sciences

• In Event: Refining the Analysis of Individual Differences in Criminology: Biosocial Contributions

Abstract

The fundamental purpose of advanced statistical techniques is to estimate causal relationships independent of confounding influences. Generally, these statistical techniques used to approach causality have been categorized into two clusters: standard social science methodologies (SSSMs) and quantitative genetic methodologies. Both methodologies approach the problem of causality by establishing a counterfactual condition where observed or unobserved predispositions are held constant at varying exposure to a treatment condition. The counterfactual condition established by SSSMs generally hold environmental predispositions constant, while quantitative genetic methodologies generally hold genetic predispositions constant. Consequently, scholars using these techniques are still limited in their ability to establish a counterfactual condition emulating that of a true experiment. In an effort to address these limitations, the current study proposes an innovative methodology for combining polygenic risk scores and propensity scores into a single estimate – genetically adjusted propensity score (GAPS) – representing the variation in the treatment condition predicted by observed environmental and observed genetic factors. Methodological strategies of estimating and using GAPS to establish a counterfactual condition approximating that of a true experiment will be provided. Furthermore, the importance of genetically adjusted propensity score matching (GAPSM) and theoretical and empirical implications of the methodology will be discussed.

Authors

• Ian A. Silver, University of Cincinnati

• Joseph L. Nedelec, University of Cincinnati

• J.C. Barnes, University of Cincinnati