Causal Forest and Doubly Robust Machine Learning for Political Science

Machine learning (ML) methods in general and tree-based methods in particular are onthe rise in the social sciences. While ML is widely valued for predictive accuracy, newcausally-oriented machine learning methods actually leverage this power for statisticalinference and treatment effect estimation. In this article, we synthesize this causalML literature into a new paradigm we coin “Doubly Robust Machine Learning,” orDRML. Our conceptual framework makes clear the benefits of these new methods, likecausal forest, and explains how researchers can fruitfully employ these approaches toimprove on the power, precision, and coverage of standard estimators. We demonstratethese benefits for average and conditional effect estimation through a massive array ofover one million Monte Carlo simulations spanning a wide range of possible researchsettings. Taken together, our results provide the conceptual and empirical basis forthe use of DRML in social science applications.