In late 2013, the Laura and John Arnold Foundation (LJAF) released findings from a large-scale, two-year research project focused on the role that data and analytics can play in helping judges determine what risk defendants who have been arrested pose to public safety and whether they should be detained in jail or released prior to trial.
The objectives of this study were:
Develop a pretrial risk assessment that can be completed without a defendant interview;
Determine if the non-interview-based pretrial risk assessment is predictive of failure to appear and commission of new crimes during the pretrial stage; and
Validate the non-interview-based pretrial risk assessment on a secondary dataset.
Assessing Risk Without a Defendant Interview
The research revealed that scientific, data-driven, risk assessments can assist judges in making decisions about what risks defendants pose and allow jurisdictions to spend less money on pretrial incarceration while better protecting the public. Although the use of pretrial risk assessment had increased in recent years, only about 10 percent of jurisdictions actually employed these instruments. As of 2012, there were eight multi-jurisdictional pretrial risk-assessment instruments, and all relied to some degree on information garnered from defendant interviews. The low adoption rate was primarily attributed to the time-consuming and resource-intensive data collection process.
Researchers studied 750,000 cases from more than 300 jurisdictions across the nation to determine now what factors could be used to accurately predict this across all jurisdictions. A meta-analysis of pretrial risk assessments showed that the strongest predictors of court appearances and the commission of new crimes were actually static factors largely based on the defendant's prior criminal history such as prior convictions, prior misdemeanors, prior felonies, and prior failures to appear, while more dynamic factors, such as residence and employment, were less predictive indicators of risk, if at all. (See the LJAF report, Assessing Pretrial Risk without a Defendant Interview)
Public Safety Assessment-Court (PSA-Court)
Given these findings, LJAF developed a comprehensive, universal risk assessment—the Public Safety Assessment-Court (PSA-Court)—to accurately, quickly, and efficiently assess the risk that a defendant will engage in violence, commit a new crime, or fail to come back to court.
The PSA-Court was test in Kentucky, which had used a 12-point risk assessment tool called the Kentucky Pretrial Risk Assessment (KPRA), and relied on a mix of interview and non-interview-based questions. In 2013, all of the state’s 120 counties began using the PSA-Court. The Foundation reported that, in the first six months, the state was able to reduce crime by close to 15 percent among defendants on pretrial release while, at the same time, increasing the percentage of defendants who are released before trial. See the LJAF report, Results from the First Six Months of the Public Safety Assessment - Court in Kentucky.
Plans to expand the PSA-Court to other jurisdictions were announced in early 2014, and included counties in Arizona, California, and North Carolina.