The Big Picture
What We Did
Researchers analyzed data from 34,238 female probationers in a large southwestern state, including Black, Hispanic, and White individuals. The study assessed the influence of risk scores, location (two counties), and offense severity on probation outcomes. Logistic regression models were used to explore individual predictors and interaction effects, examining whether the Wisconsin Risk Needs Assessment Instrument maintained its predictive validity across racial/ethnic groups.
What We Found
1
Predictive Validity Across Groups:
– The Wisconsin Instrument showed varying predictive accuracy across racial/ethnic groups, with higher error rates for Black probationers compared to White and Hispanic probationers.
– Overclassification (false positives) was similar between Black and White probationers but significantly lower among Hispanics. Underclassification (false negatives) was notably higher for Hispanic probationers.
2
Impact of Location, Offense Severity, and Race:
– Location (County A vs. County B) significantly influenced probation outcomes, with County A showing higher rates of probation failure.
– Being Black and having higher risk scores were significant predictors of unsuccessful probation completion.
– The interaction between race and location was significant, with minority status having a more pronounced impact in one of the counties.
3
Predictive Bias in Risk Assessment:
– The study found that risk scores were less predictive for Black probationers compared to others, raising concerns about racial biases in risk assessment tools.
– The overall predictive accuracy of the Wisconsin Instrument was around 20% better than chance, but significantly less for Black probationers.
Why This Matters
The findings highlight critical inequities in the predictive validity of risk assessment tools used in probation settings. These tools may contribute to disproportionate supervision and treatment among female minority probationers, influencing probation outcomes and perpetuating systemic biases. Addressing these biases is essential for fairer and more effective correctional practices, ensuring that risk assessments do not unfairly disadvantage any group.
What Can We Do?
1
Refine Risk Assessment Tools:
2
Targeted Training and Oversight:
3
Further Research:
About the Study
This study provides new insights into the intersection of race, gender, and probation outcomes, challenging the assumption that risk assessment tools are universally equitable. By examining the predictive accuracy and biases of the Wisconsin Risk Needs Assessment Instrument, it underscores the need for ongoing evaluation and improvement of correctional risk assessment practices. The full APA citation is: Threadcraft-Walker, W., & Henderson, H. (2018). Reflections on race, personality, and crime. Journal of Criminal Justice. https://doi.org/10.1016/j.jcrimjus.2018.05.005