ABSTRACT
There are concerns that risk assessment instruments may exacerbate racial/ethnic disparities in the justice system. Risk factors measured by Western-normed tools may not accurately reflect the life experiences of non-White offenders. Systemic biases may also be baked into tools if certain risk items are unfairly associated with race/class. Despite these concerns, few studies have rigorously examined how common risk factors for violence compare across racial/ethnic groups in the U.S. This cross-sectional study addressed these gaps by examining cross-cultural differences in the performance of 10 common historical risk factors. White, Hispanic, and Black adults (N = 270) sampled from Amazon Mechanical Turk completed risk ratings and self-reported lifetime violence/offending. Overall, risk scores performed relatively well in the strength and form of their associations with offending across groups. However, rates of general offending were consistently lower for Hispanics and Blacks compared to similar-scoring Whites. In addition, several risk domains overclassified high-scoring Hispanics and Blacks but were better calibrated with low-scoring minority groups. Although our results reveal modest evidence of racial bias, definite conclusions about the universality of similar risk factors may be premature given mixed findings of overestimation/over-classification. Results have implications for fairness in risk assessment and underscore the need for further prospective investigations.
Acknowledgments
This work was supported by funding from the AP-LS Grant in Aid for Graduate Students awarded to Sanam Monjazeb, the principal investigator. Additionally, this research was supported by funding from the Psychology Department at Simon Fraser University and research grant funds provided by Dr. Kevin Douglas, the investigator’s primary supervisor.
Data availability statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to their containing information that could compromise the privacy of research participants.
Disclosure statement
The authors report there are no competing interests to declare. This article is based on the thesis completed by Monjazeb (Citation2019). A subset of the specific ideas and analyses contained in this article were presented at the 2020 annual meeting of the American Psychology-Law Society (AP-LS) in New Orleans, Louisiana (Monjazeb, Citation2020).