Problem
The Problem Addressed:
The study aimed to evaluate and revise the Oregon Public Safety Checklist (PSC), an actuarial risk assessment tool originally developed in 2008, to address its declining predictive accuracy and criticisms such as limited demographic consideration and lack of transparency. The revision aligns the tool with updated legislative definitions of recidivism.
General Impact:
The PSC is vital for allocating funding to community corrections agencies and guiding interventions for probationers and post-prison supervisees. Improving its accuracy impacts public safety, resource distribution, and equitable offender management.
Research Questions:
- Does the current PSC accurately predict new recidivism metrics (arrest, conviction, incarceration)?
- How can the PSC be revised to address predictive accuracy and fairness across demographics?
- What new risk factors or methods improve the tool’s performance?
Method and Analysis
Program Evaluated:
The PSC and its revised version (PSC-R) aimed to enhance predictive accuracy and utility in offender risk assessment.
Data and Sample Size:
- Over 70,000 cases spanning baseline (2008/2009), construction (2013), and validation (2014) samples.
- Included data from Oregon DOC, statewide law enforcement systems, and judicial networks.
Analysis Used:
- Logistic regression and ROC (Receiver Operating Characteristic) analyses evaluated predictive accuracy.
- Subgroup analyses ensured fairness across demographics (e.g., gender, race, age, region).
- Bivariate correlations identified new risk factors, while multivariate models assessed their predictive validity.
Outcome
Key Findings:
- The PSC is moderately accurate in predicting recidivism (AUC = 0.640 to 0.708). Predictive validity improved for arrests and convictions but was stable for incarcerations.
- Subgroup analyses revealed variability in predictive strength across demographics, with disparities noted for gender, age, and region.
- Nine original and 21 new risk factors were identified for inclusion in the PSC-R, improving predictive accuracy and equity.
Implications or Recommendations:
- Continued use of the PSC is supported, but concurrent refinement is critical.
- Incorporating validated new risk factors and ensuring demographic fairness enhances the PSC’s utility.
- Policymakers and practitioners are advised to adopt the PSC-R for improved public safety outcomes and informed decision-making.
This structured update of the PSC ensures its relevance in advancing criminological practices and addressing systemic challenges in offender management.