Oregon Pretrial Risk Scale Demonstration Project

Problem

The Problem Addressed: 

The study aimed to improve the accuracy, equity, and efficiency of pretrial risk assessments in Oregon by designing an actuarial risk assessment scale using data from existing statewide sources.

General Impact: 

Enhancing pretrial decision-making through standardized risk assessment methods.
Reducing reliance on informal judgments, which may vary widely and lead to inconsistencies.
Informing stakeholders about the predictive validity of such tools for reducing pretrial misconduct.
 

Research Questions Answered:

1. Can pretrial misconduct in Oregon be accurately predicted using existing criminal justice data?
2. How do predictive models perform across different types of misconduct and demographic groups?
   
    

Method and Analysis

Program Evaluated: 

• Evaluation of pretrial misconduct prediction models.
• Addressed the lack of standardization in pretrial decision-making, particularly as mandated by Oregon’s SB 48.
   

Data and Sample Size: 

• Data: Statewide criminal justice data systems (LEDS, DOC, Odyssey).
• Sample Size: 26,013 criminal cases from Oregon's Circuit Courts in 2018.
   

Analysis Used:

• Bivariate and multivariate regression analyses to identify optimal risk factors.
• Assessed predictive validity across three forms of misconduct:
  1. Arrest for a non-property, non-controlled substance (non-PCS) offense.
  2. Arrest for a person crime.
  3. Failure to appear (FTA) for a court hearing.
      

Outcome

Key Findings:

• One-third of defendants failed to appear in court or were arrested for non-PCS offenses; 11% were arrested for person crimes.
• Predictive validity of the risk scale was comparable or superior to existing tools like the VPRAI and PSA, achieving AUCs of:
  • 0.71 for non-PCS offenses.
  • 0.73 for FTAs.
  • 0.65 for person crimes.
• The risk scale was valid across different racial and ethnic groups.
   

Implications/Recommendations: 

• The findings support the use of automated, standardized risk assessment tools based on existing statewide data.
• Implementing such tools could improve consistency and accuracy in pretrial release decisions, aligning with Oregon’s legislative goals for reform.
• Broader application of these methods could streamline the judicial process and mitigate systemic disparities.

This study represents a significant step in advancing evidence-based practices in the pretrial decision-making process, with potential applications in other jurisdictions facing similar challenges.