2014/07/23 by Daniel Katz, Katz, Daniel Martin, Michael James Bommarito +3 · 3 citations
Economics, Econometrics and Finance · Social Sciences · #Artificial Intelligence in Law #FOS: Computer and information sciences #FOS: Physical sciences #Judicial and Constitutional Studies #Law, Economics, and Judicial Systems #Physics and Society (physics.soc-ph) #Social and Information Networks (cs.SI)
paper · pdf · doi:10.48550/arxiv.1407.6333
openalex publication_date 2014/07/23 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28
Building upon developments in theoretical and applied machine learning, as\nwell as the efforts of various scholars including Guimera and Sales-Pardo\n(2011), Ruger et al. (2004), and Martin et al. (2004), we construct a model\ndesigned to predict the voting behavior of the Supreme Court of the United\nStates. Using the extremely randomized tree method first proposed in Geurts, et\nal. (2006), a method similar to the random forest approach developed in Breiman\n(2001), as well as novel feature engineering, we predict more than sixty years\nof decisions by the Supreme Court of the United States (1953-2013). Using only\ndata available prior to the date of decision, our model correctly identifies\n69.7% of the Court's overall affirm and reverse decisions and correctly\nforecasts 70.9% of the votes of individual justices across 7,700 cases and more\nthan 68,000 justice votes. Our performance is consistent with the general level\nof prediction offered by prior scholars. However, our model is distinctive as\nit is the first robust, generalized, and fully predictive model of Supreme\nCourt voting behavior offered to date. Our model predicts six decades of\nbehavior of thirty Justices appointed by thirteen Presidents. With a more sound\nmethodological foundation, our results represent a major advance for the\nscience of quantitative legal prediction and portend a range of other potential\napplications, such as those described in Katz (2013).\n