Search

Violence risk instruments are used in numerous countries to estimate an individual's risk of reoffending. The racial fairness of violence risk instruments has received increasing attention due to ostensible differences among Anglo populations and racial minorities (e.g. African Americans and Indigenous populations). Fairness, which has numerous definitions (sensitivity fairness, error rate balance, calibration, predictive parity, statistical parity), can affect a risk instrument's utility in varying ways. This literature review explored how notions of fairness are discussed in the risk instrument literature with a specific focus on racial fairness. It also examined and critiqued the varying proposed resolutions to increase fairness. Many of these forms of fairness were found to be rarely satisfied in the literature. Further, the complications in achieving multiple forms of fairness simultaneously and the challenges of optimising both fairness and accuracy are discussed. Last, proposed solutions to increase racial fairness were often found to encompass significant limitations. Future directions for racial fairness in risk instruments are discussed, with a focus on exploring the trade-offs among varying fairness definitions and among fairness and accuracy. [ABSTRACT FROM AUTHOR]

The use of statistical learning methods has recently increased within the risk assessment literature. They have primarily been used to increase accuracy and the area under the curve (AUC, i.e., discrimination). Processing approaches applied to statistical learning methods have also emerged to increase cross-cultural fairness. However, these approaches are rarely trialed in the forensic psychology discipline nor have they been trialed as an approach to increase fairness in Australia. The study included 380 Aboriginal and Torres Strait Islander and non-Aboriginal and Torres Strait Islander males assessed with the Level of Service/Risk Needs Responsivity (LS/RNR). Discrimination was assessed through the AUC, and fairness was assessed through the cross area under the curve (xAUC), error rate balance, calibration, predictive parity, and statistical parity. Logistic regression, penalized logistic regression, random forest, stochastic gradient boosting, and support vector machine algorithms using the LS/RNR risk factors were used to compare performance against the LS/RNR total risk score. The algorithms were then subjected to pre- and postprocessing approaches to see if fairness could be improved. Statistical learning methods were found to produce comparable or marginally improved AUC values. Processing approaches increased several fairness definitions (namely xAUC, error rate balance, and statistical parity) between Aboriginal and Torres Strait Islanders and non-Aboriginal and Torres Strait Islanders. The findings demonstrate that statistical learning methods may be a useful approach to increasing the discrimination and cross-cultural fairness of risk assessment instruments. However, both fairness and the use of statistical learning methods encompass significant trade-offs that need to be considered. Public Significance Statement: The purpose of this study is to investigate the use of algorithms in increasing fairness between cultural groups by utilizing a commonly used instrument that assesses the risk of reoffending. Although this preliminary study demonstrated positive findings in increasing cross-cultural fairness, various trade-offs emerged, including among different forms of fairness, that require thoughtful deliberation. [ABSTRACT FROM AUTHOR]