Determining Racial Equity in Pretrial Risk Assessment

This study investigates the equity of pretrial risk assessments in predicting outcomes for defendants, particularly focusing on racial disparities. It explores whether pretrial risk assessments fairly classify Black and White defendants in terms of their risk for failing pretrial supervision, addressing concerns that these tools may contribute to racial biases in the justice system.

BACKGROUND

The study responds to criticisms following the George Floyd Movement, where advocacy groups argued that pretrial risk assessments unfairly labeled minority defendants as high-risk, overshadowing their potential benefits. There is a need to understand the extent of racial bias in these assessments, as the findings of racial disparities remain ambiguous and contentious.

KEY FINDINGS

  1. Racial Disparities in Risk Scores: Black defendants generally scored higher on the Pre-Trial Release Risk Assessment (PTRA) compared to White defendants, which supports earlier research indicating similar racial discrepancies.
  2. Predictive Accuracy: The PTRA was found to be a good predictor of pretrial outcomes, with higher accuracy among Black defendants compared to White defendants, though racial disparities in scoring and removal were observed.
  3. Classification Errors: The study found that false positives were more common than false negatives, with no significant difference between Black and White defendants in the probability of these errors.
  4. Bias in Predictive Validity: The study did not conclusively find racial bias in the classification errors of the PTRA, but racial disparities were noted in removal rates and total risk scores.
  5. Need for Further Research: The study highlighted the need for additional research to understand the impact of racial disparities in pretrial risk assessments and recommended adjustments to improve the predictive accuracy of the PTRA, especially in relation to specific charges.

RECOMMENDATIONS

– Validation on Local Populations: Pretrial risk assessments should be regularly validated on local populations to ensure they accurately reflect the specific characteristics and risks of those communities.

– Refinement of Risk Assessment Tools: Adjust the assessment tools to minimize racial disparities in risk classification by incorporating additional factors that account for structural and systemic inequalities affecting minority groups.

– Focus on Specific Offenses:Modify risk assessment items to better predict specific offenses committed under pretrial supervision, ensuring that assessments are tailored to the nuances of different charges and defendant profiles.

CONCLUSION

The study finds that while racial disparities exist in risk scores and supervision outcomes, they do not necessarily stem from racial bias in the assessment tools themselves. However, the observed differences in removal rates suggest a need for ongoing evaluation and refinement of pretrial risk assessments to promote racial equity. Future research should further explore the underlying causes of these disparities to enhance the fairness and effectiveness of pretrial justice practices.

Racial Disparities in Risk Scores

Black defendants are often scored higher on risk assessments than White defendants, a disparity partially attributed to their higher likelihood of prior arrests and involvement in the criminal justice system.

False Positives and Negatives

The study found that false positives (high-risk classification despite successful pretrial completion) were more common than false negatives (low-risk classification despite pretrial failure), with racial distribution showing no significant differences between Black and White defendants.

Predictive Accuracy

The Pre-Trial Release Risk Assessment (PTRA) was generally accurate in predicting pretrial outcomes, with AUC (Area Under the Curve) statistics indicating good predictive ability, but with slight variations between racial groups.

Supervision Outcomes

Black defendants had a higher probability of removal from pretrial supervision compared to White defendants, highlighting a potential racial disparity in pretrial supervision outcomes.

Equity in Predictive Errors

Despite scoring disparities, the study found no significant racial differences in the likelihood of false positive or false negative classifications, suggesting that the PTRA's predictive errors do not disproportionately impact one racial group over another.