Problem
The Problem Addressed:
The study aimed to evaluate the predictive accuracy of the Public Safety Checklist (PSC) in determining pretrial misconduct outcomes such as failure to appear (FTA), new arrests for any crime, or person crimes. It addresses concerns regarding the applicability and equity of risk assessment tools in Jackson County.
General Impact:
The lack of standardized and validated pretrial risk assessment tools leads to inconsistent decision-making across jurisdictions, potentially jeopardizing community safety and fairness.
Research Questions:
The study explored nine questions, including the demographic and criminal characteristics of defendants granted pretrial release, types of misconduct during release, and the PSC's predictive accuracy across different demographic groups.
Method and Analysis
Program Evaluated:
The Public Safety Checklist (PSC), initially developed to predict recidivism among adjudicated populations, was evaluated for its suitability in pretrial decision-making.
Data and Sample Size:
The sample included 2,028 cases from Jackson County in 2018, extracted from a larger dataset of 24,438 cases across Oregon. Data sources included court records, state arrest records, and pretrial release information.
Analysis Used:
Predictive accuracy was assessed using Area Under the Curve (AUC) statistics derived from Receiver Operating Characteristic (ROC) analyses. The study also examined variations in accuracy by demographic factors (e.g., race/ethnicity and sex) and explored potential modifications to the PSC's scoring system.
Outcome
Key Findings:
- Predictive Accuracy:
- The PSCFelony and PSCProperty scales moderately predicted general misconduct (FTA and new arrests).
- All PSC scales were weak or unreliable in predicting arrests for person crimes.
- Demographic Variations:
- The scales showed consistent predictive accuracy across sex groups.
- Lower predictive accuracy was noted for Black/African American defendants, raising equity concerns.
- Scoring Adjustments:
- Regression analyses suggested the original PSC scoring could be optimized for pretrial purposes by modifying item weights and risk factors.
Implications or Recommendations:
- Use of PSC: Continued cautious use of the PSCFelony and PSCProperty scales is recommended for predicting FTA and general arrests, with acknowledgment of their limitations.
- Validation and Alternatives: Local validation of the PSC for pretrial decisions is critical. Exploring alternative risk assessment tools may enhance predictive accuracy and equity.
- Equity Concerns: Address disparities in the predictive performance of the PSC for racial/ethnic groups, particularly Black/African American defendants.
This report highlights both the strengths and limitations of the PSC in the pretrial context, emphasizing the need for localized validation and consideration of alternative tools to improve fairness and predictive reliability.