ALBUQUERQUE, N.M. (AP) — A Texas-based public policy group is pushing to expand the use of a risk assessment tool to help judges determine the fate of defendants pending trial as it looks to boost research on how effective such algorithmic systems are in predicting whether someone will commit another crime while on bail.
The Laura and John Arnold Foundation announced Wednesday that it's seeking proposals as part of a five-year, multimillion-dollar project to determine the effectiveness of its assessment, which relies on several factors.
The push comes as New Mexico and other states grapple with bail reform.
Foundation officials say there's a growing demand for the ability of judges to better assess a defendant and more research will help in understanding the limitations of such tools.
Jeremy Travis, the foundation's executive vice president of criminal justice and former president of the John Jay College of Criminal Justice, said the idea is to look deeper at the predictive power of the assessments through field work at a local level.
"The science question is not something that we're looking at on the macro level," he said. "We're really suggesting and encouraging local jurisdictions — where they have the ability to do it — to test the risk assessment tool they have against their local population."
Travis said there are differences when it comes to a community's level of poverty, the types of crimes occurring, demographics and even the public's trust in the judicial system. So one size won't fit all, he said.
"It can and should evolve and the evolution will be some mix of what does the research show and what is the reality of implementation," he said.
New Mexico voters in 2016 overwhelmingly supported a constitutional amendment aimed at keeping dangerous defendants in custody pending trial, while allowing for the release of nonviolent suspects who can't afford bail. The change was spurred by criticisms that the system was routinely allowing violent offenders out on the streets.
Rules related to the change have spurred much debate over the past year among judges, prosecutors and others, and the New Mexico Supreme Court has been working to streamline the process.
The model developed by the Arnold Foundation is used in about 40 jurisdictions, including in New Jersey, Kentucky and in New Mexico's busiest judicial district — Bernalillo County.
Supporters say the assessments limit prejudicial judgments that could be influenced by race, gender or even clothing. Critics worry that such algorithms might end up supplanting judges' own judgment.
There's also disagreement among prosecutors and public defenders about which risk factors should be considered. Factors often used to evaluate a defendant include age, past criminal convictions and whether the person has previously failed to appear in court cases.
James Hallinan, a spokesman for the New Mexico Attorney General's Office, said Wednesday that assessment tools can be useful, but there's no one substitute for sharing information among law enforcement and the courts as the specific facts of each case are weighed.
"As we work on bail reform in New Mexico, our court rules must prioritize public safety in all cases," Hallinan said.
State and local court officials did not immediately comment on whether they have interest in the research project. The foundation hopes to have its partners in place by the fall.
Noting that the issue of risk is always present, Travis said the findings so far indicate that pretrial assessments are helping judges on a daily basis.
"The ultimate goal — this is what we'll measure and what we measure all the time — is a reduction in crime and an increase in return to court to meet one's legal obligations," he said.