Multi‐study examination of criminal‐legal professionals’ use of risk assessments in pretrial decision‐making
Abstract
Introduction
Risk assessments represent a contemporary solution to reform bail practices. Current research suggests such tools can promote nonfinancial release. However, fewer studies have examined why judges may not regularly adhere to pretrial risk assessments and how other pretrial practices may affect risk assessment-guided decision-making.
Aim
This study examined how long-standing and newer pretrial practices—the presence of counsel at first appearance, continued reliance on traditional bail setting practices and the use of structured guidelines—informs pretrial release decision.
Methods
We conducted a national, cross-sectional survey of 146 US criminal-legal professionals representing defence attorneys (37.2%), pretrial services officers (26.4%), prosecutors (14.0%), judges (8.3%), and other roles (14.0%). The survey involved three randomized vignette scenarios and questions on respondents’ background and perceptions of risk assessments. In each vignette, participants rated their probability of a release on recognizance, detention, bail and supervision decision. Analyses incorporated a combination of t-tests, ANOVAs and multiple regressions.
Results
Findings indicated the absence of counsel at first appearance and presence of a traditional bail system led to a lower likelihood of releasing on recognizance, while the provision of structured guidelines did not alter participants’ likelihood of releasing on recognizance, assigning bail or assigning significantly higher bail amounts. Participants’ race, political affiliation and confidence in daily decision-making were associated with certain release decisions.
Conclusion
Our findings suggest other pretrial practices limit the consideration of risk assessment information in decision-making. To promote greater adherence to risk assessment tools, we recommend strategies to curtail discretionary use of bail when pretrial risk assessments are present.