This document provides an overview of topics from roundtable discussions at the 2021 National Institute of Justice Recidivism Forecasting Challenge, highlighting discussions on fairness in criminal justice risk assessment, with themes such as bias reduction, gender-specific needs and their implications for programming, and more.
The National Institute of Justice Recidivism Forecasting Challenge was designed to enhance knowledge about predicting recidivism. In December 2021, the winning contestants attended a symposium. This article reviews points about fairness in criminal justice risk assessment discussed by symposium participants. The transcripts were thematically coded to reveal discussion points. This article reports the topics from roundtable discussions which centered on developing algorithms to achieve racially equitable outcomes and account for gender-specific needs. The following specific themes regarding bias in risk assessment were identified: reducing bias, improving fairness in risk assessments, gender-specific needs, and implications of gender-specific needs for programming. These identified themes point to the complexity of addressing bias in risk assessment. Conflicting definitions of fairness mean that stakeholders must prioritize what is most important to them. Stakeholders should pay close attention to variables that predict female recidivism and develop risk assessments and programming with gender-specific needs in mind. (Published Abstract Provided)