What do a newborn infant, a shortstop, and a television have in common?
The answer: we can accurately predict how each will perform, because we have supplemented our own intuition with something more concrete: consistent, objective data. The Apgar Test is far better at identifying babies in distress than doctors with years of experience. The book Moneyball famously captured how a baseball player's stats could out-predict a seasoned scout. And we all gather data on electronics before making a major purchase.
My colleagues and I at the Laura and John Arnold Foundation are now working to add another group to this list: pretrial defendants.
In courtrooms across America, judges are left to rely on experience and gut instinct when deciding whether to release or detain a defendant before his or her trial. Despite the profusion of data and analytics in countless industries today, our criminal justice system has not similarly modernized. It continues to function largely the same way it did 50 years ago.
We've found that, despite judges' best efforts, we too frequently make the wrong pretrial decisions. Data show that we are detaining many defendants who pose little risk to public safety, and releasing almost half of the defendants who pose the greatest threat to public safety.
The consequences are severe. Research suggests that low-risk defendants—the majority of arrestees—who are incarcerated prior to trial are more likely to become repeat offenders and more likely to be sentenced to jail or prison terms than similar defendants who are not incarcerated before trial. Of course, the risks of releasing a defendant who falls into the small but critical group of those highly likely to commit another crime or, worse, a violent crime, are immense.
Meanwhile we spend $9 billion annually to keep on any given day more than 500,000 people in jail who are awaiting trial.
With the stakes so high—for individual lives, for public safety, and for our taxpayer dollars—we should be doing everything we can to accurately identify which defendants pose a risk to public safety if released, and which do not. Until recently, however, a consistent, evidenced-based risk assessment tool, that would help judges objectively answer precisely that question, seemed out of reach.
The Laura and John Arnold Foundation saw an opportunity and began collaborating with researchers to develop a data-driven tool that could be implemented with little cost and scaled easily to accommodate the 12 million people arrested each year.
Our researchers worked with the largest, most diverse dataset of pretrial criminal records ever assembled—1.5 million cases from 300 jurisdictions across the country—to see what factors best predicted whether a defendant was going to commit a new crime, commit a new violent crime, or fail to reappear for a court hearing.
After considering hundreds of options, we found that we could get consistent, accurate results from an assessment that consists of just nine questions. This includes factors like prior failures to appear for trial and past criminal convictions—all things that can be answered by looking at a standard case file. None of the questions asks for any demographic data other than the defendant's current age, and none requires costly interviews to answer.
The tool also does not usurp judicial decision-making in any way; it simply provides objective information about the risk a given defendant poses.
We call it the Public Safety Assessment, or PSA, and it is one of the most significant criminal justice reform initiatives underway in the United States today. It tackles the three problems that plague the modern pretrial system: too much crime, too much incarceration, and too much cost.
Pilot programs suggest that the PSA helps judges reduce all three. Jurisdictions using the PSA have been able to reduce their pretrial jail populations with no negative impact on the crime rate. In fact, in Kentucky, the pretrial crime rate has fallen since the tool was implemented in July 2013. In Charlotte, North Carolina, the jail population has declined by 20 percent since the city began working toward implementation of the tool in the spring of 2014.
Further, officials in pilot jurisdictions estimate that use of the tool has freed up millions of dollars that can now be used on other priorities.
The Laura and John Arnold Foundation is now rolling out the PSA to 21 new jurisdictions around the country. The tool will be available to judges in 29 jurisdictions, including three entire states—Arizona, Kentucky, and New Jersey—as well as three of the largest cities and two of the largest jail systems.
Our goal is to make the PSA available, at no cost, to judges in every jurisdiction in the U.S. within the next few years.
Data has transformed our ability to predict the future in many areas. It is high time that we start using it to reduce crime and incarceration in America.
Anne Milgram, a former New Jersey Attorney General (2007-2010), is vice president of criminal justice at the Laura and John Arnold Foundation. She welcomes comments from readers.