Motivated Reasoning in the Field: Polarization in Precedent, Prose, Vote, and Retirement in U.S. Circuit Courts, 1800-2013

Abstract

We document motivated reasoning among U.S. Circuit Court judges. We employ a supervised learning approach to measure partisan influences on prose (writing style), precedent (citations to previous cases), and policy (dissenting votes). We find persistent but low partisanship of language overall, with the notable exception of Civil Rights, Labor Relations and First Amendment jurisprudence. Citations display a significant level of partisanship and increase over time. Voting along party lines (dissenting against judges from opposing party) and strategic retirement (retiring while one’s own party controls the presidency) have also increased. Motivated reasoning is predictive of decisions in the Supreme Court and grows with judicial experience.