Mental Illness as a Sentencing Determinant: A Comparative Case Law Analysis Based on a Machine Learning Approach

Abstract

This study identifies factors that contribute to sentencing outcomes for criminally sentenced individuals experiencing mental disorders, in two U.S. states with divergent sociopolitical ideologies. Recent case law ( n = 130) from appellate courts in New York and Kansas (from 2020 to 2021) was analyzed using regression and machine learning to predict sentence severity for individuals experiencing mental disorders. Across both states, trauma-related and personality disorders led to the most severe sentences, while paraphilia, addiction, and mood disorders had the lowest probability of imprisonment. Sentencing outcomes in Kansas were significantly more severe as compared with New York. A classification analysis identified important patterns of sentencing determinants that predicted which mental disorders were more likely to lead to incarceration. Findings and implications are discussed in relation to punishment disparities as well as the potentials and pitfalls regarding the use of machine learning approaches in criminal justice research and policy.