By Anne Milgram, JD, Vice President of Criminal Justice, Laura and John Arnold Foundation
Over the past two decades, crime rates in the United States—and, in particular, violent crime rates—have declined sharply, and leading experts suggest that innovative policing strategies deserve a large part of the credit for this decrease. From CompStat to hot-spotting, law enforcement has led the way in leveraging data and technology to determine how best to deploy its resources to enhance public safety.
And yet significant challenges remain. Violent crime may be down overall, but it continues to plague many urban neighborhoods. Recidivism is disturbingly high, with nearly two-thirds of people who leave jail or prison winding up back behind bars within three years. And this is despite a dramatic increase in spending: The states and the federal government now spend $70 billion a year on corrections, an increase of 660 percent since 1982.
So how can we change this? While serving as Attorney General of New Jersey from 2007 to 2010, I spent a great deal of time considering ways we could make our criminal justice system more effective at reducing crime and producing just outcomes. For the past three years, I have continued to tackle those same issues from a different angle—as the Vice President of Criminal Justice at the Laura and John Arnold Foundation (LJAF).
What I have come to believe based on my experience in both posts is that the key to improving public safety, fairness, and cost-effectiveness is the use of data and research to guide criminal justice policies and decisions. In everything from education to health care, this is the approach that has been successful in addressing many societal challenges. Clearly, we can—and must—do the same in the field of criminal justice.
My colleagues and I at LJAF have spent much of the past three years gathering data, conducting research, and developing tools to help police, prosecutors, judges, and others address some of the key challenges they face every day.
One of the first questions we focused on is how to determine—at the earliest stages of the criminal justice process—what risks a given defendant poses to public safety. This, of course, is a question police must grapple with as they make decisions about who to arrest, who to cite, and who simply to give a warning. Likewise, prosecutors face this issue when making decisions about charge, diversion, deferred prosecution, and recommendations to the court about how to deal with a defendant while his case is ongoing. And judges, too, must assess risks to public safety when they decide whether—and under what conditions— to release a defendant before trial.
Yet, despite the importance of these decisions, very few jurisdictions use research-based instruments to assist those making these critical determinations about risk. Unfortunately, this means that dangerous defendants are often back out on the street quickly, while significant numbers of low-risk defendants remain incarcerated for long periods.
To address this problem, we are building data-driven tools that provide police, prosecutors, and judges with objective information about the risk that a defendant will commit a new crime, commit a new violent crime, or fail to appear for court.
The first of these instruments, the Public Safety Assessment -Court (PSA-Court), is already in use in pilot sites across the country. The instrument was developed using the largest-ever dataset of 750,000 cases from across the country, which were rigorously analyzed to determine what information about a defendant is most predictive of risks to public safety. Based on this research, we created the PSA-Court, which relies on just nine objective risk factors —all of which can be gathered from the administrative record without interviewing the defendant—to determine the likelihood that a defendant will be rearrested, engage in violence, or skip court.
Kentucky, whose judges have been using the PSA-Court since July 2013, has seen striking results: In the first six months of use, judges have been able to reduce pretrial crime by 15 percent, while at the same time increasing the proportion of defendants released before trial. And the PSA-Court is providing them with critical information about violence. In Kentucky, defendants flagged by the assessment as posing an increased risk of violence are, in fact, rearrested for violent acts at a rate 17 times that of defendants who are not flagged.
Importantly, because it does not rely on factors like neighborhood or income, the PSA-Court is helping deliver these results without discriminating on the basis of race or gender.
Soon, we will begin piloting a similar tool for prosecutors, the PSA-Prosecution, which will help inform their decisions regarding charge, diversion, and deferred prosecution.
And on the policing front, we are partnered with the International Association of Chiefs of Police to conduct research on how police departments currently use citations in lieu of arrest. This work will provide the foundation for developing instruments to help law enforcement make data-driven decisions about which individuals pose a risk of committing a new crime or failing to come back to court, and therefore should be arrested and booked rather than cited and released.
But risk assessment is not the only area in which data and technology can help improve public safety. LJAF has also developed an app for tablets that will allow police to conduct eyewitness identifications quickly and effectively in the field. It will also provide law enforcement agencies with a simple way to adopt research-based best practices that improve the accuracy of identifications. Given the important role that eyewitness IDs play in investigations and trials, enhancing their accuracy can help get dangerous people off the streets.
From policing to prosecution to courts, data and research can generate powerful insights and help develop tools that make our neighborhoods safer, promote fair outcomes, and to help use our resources as effectively and efficiently as possible.