Read our COVID-19 research and news.

Protesters look for jailed friends during a protest against detention conditions in St. Louis, Missouri.


Does jailing people before trial make cities safer? Not always, new research suggests

In February 2017, Jonathan Barnes was arrested by Philadelphia, Pennsylvania, police and charged with marijuana possession and intent to distribute. In the 2-minute hearing after his arrest, a magistrate told him he could go free if he paid $5000 in cash bail, money he'd get back if he showed up at his next hearing. Or he could pay just $500, with the rest secured by a bail bondsman who could send out a bounty hunter if Barnes didn't show. Barnes didn't have $500, let alone $5000, so he spent more than a week in the city's House of Correction awaiting trial. Because he couldn't work, he lost his job. More than 2 years later, he's paid $2000 in fines and is still searching for a steady paycheck.

To make sure other defendants don't end up like Barnes, cities and states across the United States are moving to ease cash bail and other pretrial detention policies that critics say are unfair, counterproductive, and contribute little to public safety. But the reforms—backed by liberal and libertarian groups alike—have drawn stiff opposition from some law enforcement organizations and the bail bond industry. Relatively little hard evidence informs the battle. "People are kind of flying by the seat of their pants," says Megan Stevenson, an economist and legal scholar at George Mason University in Arlington, Virginia.

Now, she and other social scientists are trying to fill the gap. Backed by funding from major foundations, they are launching studies to find out whether pretrial practices such as cash bail really do result in higher appearance rates and safer communities. They're also exploring how such practices affect case outcomes. Results published last month, for example, bolstered reformers' case that cash bail is ineffective, at least in Philadelphia. But such research is hard to apply widely, and scientists face other obstacles, including police, prosecutors, and prisons reluctant to share data.

For decades, U.S. rates of pretrial detention have soared as the effects of the "tough on crime" policies of the 1980s have trickled down. At the federal level, nearly 75% of all arrestees were confined prior to court hearings last year, up from just 30% in 1988, according to the U.S. Bureau of Justice Statistics. (The number drops to about 50% if those detained on immigration charges aren't counted.) At the local level, where most people are charged, about 60% of defendants are held in pretrial detention.

Recently, however, a backlash has set in. In Philadelphia, prosecutors have stopped seeking bail payments for some two dozen offenses, including drug possession with intent to distribute. And over the past 3 years, several states—including New Jersey and California—have passed laws dropping cash bail for most offenses. Related legislation is on the table in at least 24 other states. As the debate over such changes has intensified, so has the call for data. And with good reason: A 2016 meta-analysis found that just 32 of 811 pretrial justice studies published since 1970 contained enough original data to allow rigorous analysis.

Many of those studies look at risk assessment, the art of predicting whether a defendant will skip their court date or reoffend if set free. Judges are strikingly inconsistent in their pretrial rulings, so many states use computerized tools that aim for more objective assessments. But the tools, which generate risk scores based on characteristics such as a defendant's age and their number of prior arrests, can also produce inconsistent results. And some analyses have suggested the data used to train and feed the systems can reinforce racial and economic biases, resulting in more onerous pretrial conditions for poor, minority defendants.

Studies on the effectiveness of cash bail have also exploded. In Philadelphia, Stevenson and economist Aurélie Ouss of the University of Pennsylvania have examined how the recent reforms affected defendants in 21,023 cases—roughly half from before the changes were announced in February 2018 and half after. Postreform, they found a 12 percentage point increase, to 59%, in defendants released on their own recognizance (an additional 1750 releases per year), and a five percentage point drop, to 19%, in defendants who spent at least one night in jail. But eliminating cash bail appears to have had little effect on court appearance rates or public safety: Neither failure-to-appear rates nor new arrests went up after the shift, they write in a 17 February preprint.

The findings contradict several earlier studies, including one by George Mason University economist Alex Tabarrok. In 2004, he looked at a sample of defendants charged with felony crimes in 75 of the most populous U.S. counties and found that those with cash bonds backed by bail bondsmen had significantly higher appearance rates than those with other pretrial conditions.

Tabarrok says one likely reason for the disparity is that the study populations were different: The defendants in the Philadelphia analysis may have been at lower risk of jumping bail to begin with. Such caveats point to a larger challenge facing pretrial detention studies, says Sue Ferrere of the Pretrial Justice Institute, a nonprofit based in Rockville, Maryland. Confounding variables are rife, she says, and the data collected by different police forces, courts, and jails vary widely, making comparisons difficult.

To avoid such pitfalls, researchers are now trying to run controlled experiments with willing courts and law enforcement agencies. For example, one pilot study will provide public defenders to a subset of defendants at bail hearings in Allegheny County in Pennsylvania. Researchers from the State University of New York in Albany and the Rand Corporation in Pittsburgh, Pennsylvania, hope to see whether quickly providing an attorney will result in fewer detentions, fewer rearrests, and more court appearances.

That project and others are also exploring how cash bail correlates with other outcomes, including future employment and trial results. One such study, published last year in the American Economic Review, for example, suggests defendants who can't make bail are more likely to plead guilty. "You can be a public defender for about a day, and you realize very quickly the role bail plays in forcing guilty pleas," says Robin Steinberg, a former public defender and CEO of the Bail Project in Marina del Rey, California, which helps defendants make bail. When a judge sets an unaffordable payment, "almost everybody will plead guilty, whether they did it or not."

Barnes says that's what happened to him. After his week in jail, he decided to plead guilty so he could get back to work. Now, he says it's hard not to be bitter: Had he been arrested just 1 year later, he would have qualified for relief under the new Philadelphia reforms. "I paid for it," he says, "and I'm still paying for it."