Laura and John Arnold Foundation Launches PSA to Improve Risk-Based Decision-Making

The Laura and John Arnold Foundation (LJAF) recently launched the Public Safety Assessment (PSA) in dozens of jurisdictions across the country, including three entire states. To build the PSA, researchers identified nine core factors that predict the likelihood that a pretrial defendant will commit a new crime, will commit a new a violent crime, or will fail to reappear in court if released before trial. The preliminary results have been very promising: in Kentucky, the PSA has assisted judges in reducing both pretrial crime rates and jail populations, and in Charlotte, North Carolina, the jail population has declined 20 percent since the city began working towards implementation of the tool in the spring of 2014. The PSA, which does not use any factors related to race or gender, has shown that we can use evidence-based data to make our streets safer while at the same time detaining fewer people in jail.

The next challenge is to understand how judges can best use the PSA and other pretrial risk assessment tools to make better decisions about who should be released and who should be detained. New research funded by the Department of Justice and LJAF strongly suggests that using a pre-determined decision making framework (also known as a ​“structured decision making” tool) with a risk assessment tool reduces crime, enhances jail release rates, improves consistency, and makes the system fairer and more cost-effective overall.

The research, reported here, is based on a controlled experiment in Virginia that examined the impact of a predetermined decision making framework on decisions to release a defendant pretrial and on decisions about what kind of pretrial supervision is warranted in individual cases. Virginia uses its own pretrial risk assessment tool, the Virginia Pretrial Risk Assessment Instrument (VPRAI). Prior to the study, however, there was no structured guidance on how to use it. Pretrial officers made pretrial release and supervision recommendations to judges based on their subjective assessment of the situation.

In the study, approximately half (14) of Virginia’s 29 jurisdictions were given a decision making framework, called a ​“Praxis,” that made release and supervision recommendations based on a defendant’s VPRAI risk assessment and charge category. Court officers in these jurisdictions were given training on the decision making framework and ongoing technical assistance.

The results are powerful and clear: in the jurisdictions that were trained on the decision making framework, pretrial officers followed the recommendation 80% of the time and were 2.3 times more likely to recommend release at first appearance compared to non-framework jurisdictions (where officers based recommendations on their subjective judgment).

The use of the decision making framework also had an impact on how the pretrial officers’ recommendations were viewed by judges. In jurisdictions that used the framework, judges released defendants at first appearance nearly twice as often and were 8.8 times more likely to release a defendant at first appearance when release was recommended by the pretrial officer.

Critically, the study also showed that using risk assessment tools with a structured decision making framework yielded better results than using the risk assessment alone. Defendants were 20 percent less likely to experience any pretrial failure, such as re-arrest, and 30 percent less likely to fail to appear or to experience a new arrest pending trial in the jurisdictions that used the framework to make pretrial recommendations.

Adopting evidence-based approaches to criminal justice is partially about creating better data and partially about learning to use that data. With this new study, we are closer to reaching both goals.