Causal Inference in Criminal Justice Research: The Past, Present, and Future

• In Event: Machine Learning and Causal Inference

Abstract

Social science research has benefited greatly over the past few decades from the methodological advances in causal inference techniques. Specifically, new approaches in the potential outcomes framework—including fixed effects panel designs, synthetic control, regression discontinuity designs, and multiple forms of statistical matching—have provided researchers with new, relatively easy-to-use methods for estimating causal relationships. Despite these advancements, however, criminal justice research as a whole lags behind other similar fields (e.g., economics, sociology, psychology) in terms of high-quality causal research. The need for methodologically rigorous causal research—including experimental research—is especially high in the current “evidence-based” era because these methods can produce better policy (Sampson, 2010). As such, this paper discusses the history, current trends, and potential future directions of causal inference in criminal justice and criminology research. In addition, some of the methodological advances in the potential outcomes framework are presented, as well as a discussion of how these advances can improve criminal justice research.

Author

• Shaun M. Gann, Boise State University