May 8, 2021 06:26

Pretrial Risk Assessment Tools More Accurate Than ‘Human Judgments Alone’: Experts

A group of criminologists and law professors has challenged a new report by the Pretrial Justice Institute that says pretrial risk assessment instruments [PRAIs] used by many judges  to help determine which defendants should be released while their cases are pending are dangerous and have unintended consequences.

Risk assessment tools “can support the effectiveness of other reform efforts,” contends the group, led by James Austin of the JFA Institute, Sarah Desmarais of North Carolina State University and University of Virginia law Prof. John Monahan.

They say the process of assessing crime suspects scientifically provides “empirical evidence to inform a presumption of release, elimination of money bail, improved representation at pretrial proceedings, provision of pretrial services, diversion to community-based programs and, ultimately, reductions in rates of pretrial detention.”

The group of critics says the Pretrial Justice Institute, a nonprofit advocacy group, “has concluded, in contrast with scientific evidence, that use of a reliable and valid PRAI precludes reductions in pretrial detention and increases racial and ethnic disparities.”

Austin’s group argues that a “large body of social science evidence” shows that “objective, reliable and valid risk assessment instruments are more accurate in assessing risk” than  human judgments alone.

Because judges often must make decisions on pretrial releases within a few minutes, PRAIs are a more efficient, transparent, and fairer basis for making that assessment than a judge haphazardly and quickly scanning a myriad of documents, the group says.

“The benchmark here is not perfection but rather improving upon unaided human judgment, which is universally acknowledged to introduce racial and other biases,” say the criminologists and law professors in an “open letter.”

The group says the argument that PRAIs are racially biased is based on a single study by the news organization ProPublica of a single risk assessment instrument called COMPAS, in one jurisdiction, Broward County, Fl.  Other studies have shown that the ProPublica analysis was “flawed and misleading,” say the critics.

The group takes issue with the Pretrial Justice Institute’s assertion that PRAIs ͞cannot reliably predict violent crime. It says studies show that several PRAIs predict new violent crime during the pretrial release period at levels much better than does chance.

Because it is well known that there racial biases in the deployment of police, arrest policies, charging decisions, pretrial decisions, and sentencing practices, “any assessment that relies on the data used by criminal justice agencies will have some level of bias, whether conducted using PRAIs or completed by judges in the absence of PRAIs,” the group says.

The critics argue that risk assessment tools should not be abolished but rather that “any form of bias in the assessment process [should be] reduced as much as possible.”

After suggesting various ways of reducing bias in risk assessments, the group concludes that “abolishing PRAIs and allowing judges to return to, or, more accurately, to continue the practice of making subjective judgments of what constitutes risk would be a major step backwards.

“While the use of PRAIs will not eliminate pretrial detention and racial bias, they are a step in the right direction. They also are easier to fix than biased human decision-making,” concludes the group.

Other signers of the letter are Richard Berk of the University of Pennsylvania; Mary Ann Campbell of the University of New Brunswick; Todd Clear of Rutgers University; Matthew DeMichele of RTI International; Alex Holsinger of the University of Missouri-Kansas City; Michael Jacobson of the CUNY Institute for State and Local Governance; Brian Lovins of Justice System Partners; Evan M. Lowder of George Mason University; Christopher Lowenkamp of the University of Missouri-Kansas City.

Also: Sandra G. Mayson of the University of Georgia School of Law; Cynthia Rudin of Duke University; Jennifer L. Skeem of the University of California, Berkeley; Christopher Slobogin of Vanderbilt University; Victoria Terranova of the University of Northern Colorado; Jodi Viljoen of Simon Fraser University; Gina M. Vincent of the University of Massachusetts Medical School; Kyle C. Ward of the University of Northern Colorado; and Kevin Wolff of John Jay College of Criminal Justice.

Ted Gest is president of Criminal Justice Journalists and Washington bureau chief of The Crime Report.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top