What Should Our Justice System Do with Lower-Level Offenders?

I f you catch a domestic terrorist or a deranged serial killer, the next steps are obvious enough. Lock him up for life. Maybe even consider the death penalty.

But as Jennifer Doleac — a prominent economist whose work focuses on crime — writes in The Science of Second Chances, “Most people who cycle through our criminal justice system are far more sad than scary.” Most criminal cases arise from minor offenses; even the felonies, meaning crimes punishable by a year or more in prison, are mostly nonviolent.

Much of the American justice system therefore revolves around lower-level offenders who might spend a little time behind bars now and then. We need to figure out what to do with people who break the law, often repeatedly, but will inevitably, and probably soon, reenter society. Doleac’s book serves as an excellent introduction to a growing body of research focused on exactly that.

Scientifically, the key questions — such as what deters criminals from reoffending, when leniency improves outcomes, which rehabilitation programs actually work, and what policies backfire despite good intentions — are extremely difficult to study. The obvious ways of looking for answers can mislead us.

Imagine that we compare criminals who are sent to jail with those set free, and we find that members of the second group commit fewer crimes in the future. Unfortunately, that doesn’t prove that leniency reduces crime: Prosecutors and judges have reasons for giving some offenders, but not others, a break, and those let off the hook presumably tend to be lower-risk to begin with. Similarly, if prisoners who finish a rehabilitation program have better outcomes, that’s not particularly strong evidence that the program works. People who sign up for and dutifully complete such programs tend to be more motivated to change.

This is where Doleac’s expertise as an economist comes in handy. She focuses on research that addresses these problems, much of which she conducted herself. Two of her better-known studies, for instance, focused on the expansion of DNA databases: When a state or country requires more offenders to give DNA, it runs a sort of experiment around the date the new policy goes into effect. Very similar criminals either have to give DNA or not, depending on exactly when they were charged. It turns out that adding criminals to these databases reduces future convictions, presumably by deterring further crimes.

In general, increasing the chance that offenders will get caught is a good way to keep their behavior in check. Another example is the Hawaii Opportunity Probation with Enforcement (HOPE) program, which requires probationers to submit to frequent (but random) drug tests and, if they fail, to spend a few days in jail. The principle is to make punishment “swift, certain, fair”: Those who break the rules are quickly caught, but punishment takes the form of a short jail stay rather than the loss of probation entirely. In a randomized experiment, folks assigned to HOPE were less likely to be subject to future drug charges, though attempts to replicate this policy elsewhere have seen more mixed results.

What about reforms that head in a more lenient direction: giving a break to low-risk offenders to avoid disrupting their jobs, introducing them to bad influences behind bars, and saddling them with criminal records? Used judiciously, these can be effective, too. To study this form of leniency, Doleac and other researchers exploit the fact that some decision-makers (such as judges and prosecutors) are more punitive than others. In Suffolk County, Mass., for example, Doleac and some colleagues analyzed nonviolent misdemeanor cases and found that some prosecutors were particularly likely (or unlikely) to voluntarily dismiss cases — and that when a defendant lucked out thanks to a lenient prosecutor, he or she became less likely to face new charges later.

It’s easy to overinterpret studies like this one. Generally, they focus on less serious offenses and/or first-time offenders, as well as on borderline cases about which reasonable decision-makers can be more or less harsh. (Even many “progressive” prosecutors don’t change their offices’ standard practices as much as you might expect.) In short, these findings don’t imply that leniency is always a good course of action but that authorities might, to use Doleac’s word, “err” on the side of leniency when in doubt.

Doleac emphasizes that even when a case is dismissed, that doesn’t mean there were no consequences at all for the defendant: There’s still an arrest, a court hearing, and the harrowing threat of serious punishment. Defendants often leave such situations feeling as though they dodged a bullet and were given, well, a second chance.

The rehabilitation of criminal offenders is another dicey subject in the context of second chances. Prison rehabilitation programs should be easy to study: You have a (literally) captive audience, so randomly assign some prisoners to take a program and then measure whether they do better than those who weren’t assigned to participate in it. But less of this work has been done than one might think, and the results are dismayingly inconsistent. Nonetheless, Doleac reviews some programs, including some that provide mental health treatment, that are shown to have benefits.

Just as important as the question of what works, of course, is the question of what doesn’t. For example, a recurring theme in The Science of Second Chances is that many attempts to supervise at-risk offenders backfire. Some parole, probation, and halfway-house programs, for example, seem to incarcerate people over technical violations without reducing the incidence of crime. Also on the list of ineffective reforms is “banning the box” — the “box” being the part of a job application that asks employment-seekers to indicate if they have a criminal record. When employers can’t see which applicants have such records, it turns out, they often respond by discriminating against demographic groups, such as African Americans, whose members are statistically more likely to have criminal records. Addressing employers’ concerns, through publicly or privately funded insurance, for example, can be more effective if also more expensive.

Doleac closes the book with a clarion call for incrementalism. She urges us to grope our way to marginally better policy through careful study rather than big bets on untested ideas. Sometimes big ideas pan out, but more often they fail or even backfire. And sometimes little decisions — as little as reminding someone to show up for a hearing or putting him or her in touch with people who can help, as Doleac points out — can change things for the better.