Polarization of U.S. Circuit Court Judges: A Machine Learning Approach

Abstract

We employ a supervised learning approach to measure the polarization of U.S. Circuit Court judges using text data of court opinions from 1890s to 2010s. Our results show persistent but low partisanship of court opinions in the past century, but whether the partisanship has been increasing over the years is yet to be discussed. We also visualize voting networks of Circuit Court judges, and we find that judges are not polarized, although the graph exhibits clustering within the courts. We proceed to study the behavior of Circuit Court judges during Supreme Court vacancies. Our findings indicate that judges who were candidates of nomination actually chose to write less separate opinions if the senate was controlled by the opposing party.