Racial fairness in violence risk instruments: a review of the literature

Abstract

Violence risk instruments are used in numerous countries to estimate an individual's likelihood 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.

Predictive accuracy of Static-99R across different racial/ethnic groups: A meta-analysis.

“…There have been attempts to reconcile and summarize varying definitions of test bias or racial/ethnic fairness, both specific to risk assessment (e.g., Ashford et al, 2022) and more broadly (Warne et al, 2014). Given that mean group differences in scores (also referred to as statistical parity ) are more a reflection of systemic inequality rather than test bias (Warne et al, 2014) and that risk scales are fundamentally intended to predict recidivism, what is of primary importance for evaluating cross-cultural validity is examining differential predictive accuracy of the scale (and ideally the individual items as well).…”

“…Given that mean group differences in scores (also referred to as statistical parity ) are more a reflection of systemic inequality rather than test bias (Warne et al, 2014) and that risk scales are fundamentally intended to predict recidivism, what is of primary importance for evaluating cross-cultural validity is examining differential predictive accuracy of the scale (and ideally the individual items as well). Differential predictive accuracy can be defined and assessed in numerous ways (e.g., Ashford et al, 2022), but we follow the primary analyses of discrimination and calibration recommended by Hanson (2022) and Helmus and Babchishin (2017).…”

The Internet is a scary place: How does evidence source and examinee race or ethnicity influence determinations of threat?

“…Both jurors and evaluators would also have opportunities to clearly view the examinee, making physical attributes and mannerisms more salient. Research that more closely resembles the assessment and/or trial process is needed, especially considering research suggesting individuals of color may be found to be at higher risk because of their race (Ashford et al., 2022; Harcourt, 2015).…”

Aboriginal and Torres Strait Islander perspectives on forensic risk assessment

“…Barnes et al., 2016 ; Chenane et al., 2015 ; Gutierrez et al., 2013 ; Lowder et al., 2019 ; Meyers & Schmidt, 2008 ; Olver et al., 2014 ; Onifade et al., 2009 ), several studies suggest that this prediction is marginally less accurate (Shepherd, Adams, et al., 2014 ; Shepherd, Luebbers, Ferguson, et al., 2014 ; Wilson & Gutierrez, 2014 ; Wormith et al., 2012 ). Furthermore, studies have found that BIPOC offenders are misclassified more regularly than White offenders on actuarial instruments (Ashford et al., 2021 ; Campbell et al., 2018 ; Fass et al., 2008 ; Whiteacre, 2006 ; Wilson & Gutierrez, 2014 ).…”