Causal Effects of Judicial Sentiment: Methods and Application to U.S. Circuit Courts

Abstract

This paper provides a general method for analyzing the causal effects of sentiments expressed in the language of judicial rulings, with an application to the effect on social attitudes. We apply natural language processing tools to the text of U.S. appellate court opinions to extrapolate judges’ sentiments toward a number of specific target groups. Exogenous variation in those sentiments comes from an instrumental variables approach, which exploits the random assignment of judges to cases (and the fact that judge characteristics provide good cross-validated predictors of expressed sentiments). Our estimates are consistent with a backlash effect from judge sentiments to social attitudes. This effect does not persist over time and is heterogeneous depending on the target group considered.