The Government’s Early Release of Prisoners Is a Big Social-Science Experiment

Photo: Saul Loeb/AFP/Getty Images

Starting at the end of this month, a group of 6,000 federal drug prisoners will be released early as part of the U.S. government’s ongoing effort to soften certain drug-sentencing guidelines. Many of them are already in halfway houses or under home supervision, and by November 2 they’ll all have transitioned into the probation system, where they’ll have a chance to rebuild their lives. According to the Department of Justice, the average offender is being released eight and a half years into a ten-and-a-half-year sentence, and about 4,000 of them will remain in the country, with the remaining 2,000 or so being sent straight to Immigration & Customs Enforcement for deportation back to their home countries. The DOJ says those being released will have access to drug rehab, job training, and other programs to help them reintegrate.

For advocates of criminal-justice reform, this is an important milestone in the journey toward fairer sentencing and the unwinding of America’s disastrous system of mass incarceration. But it’s also an important real-world social-science experiment — one that could have a long-lasting impact on the country’s ongoing decarceration efforts. The U.S. is still trying to figure out how to release so many people from jail without also causing the crime rate to go up or inviting other unforeseen consequences, after all, so it will be important for researchers to keep a close eye on what happens here — and to put the outcomes in the proper context given the heated nature of the U.S.’s crime debate.

To be clear, the DOJ’s prisoner release shouldn’t be seen as a “traditional” real-world experiment. For such an experiment, there would need to be at least two groups: an experimental group that received the “treatment” in question — in this case, an early release from prison — and a control group that didn’t. That wasn’t possible here. As a Justice Department official who didn’t want her name used explained, “We pair inmates with the reentry plan that works best for their situation” — assigning them randomly to different sentences wouldn’t be ethical.

But there are still many ways researchers can capitalize on this. One possibility, as Mark Kleiman, a criminology expert and visiting professor at NYU’s Wagner School of Public Service, explained, would be to effectively create a control group: that is, find a group of prisoners similar to the ones being released, but who didn’t get anything chopped off of their sentences (perhaps because they’re in state prisons rather than federal), and then see if they do appreciably better or worse once they’re released.

Or researchers could take a cost-benefit approach, asking, as Kleiman put it, “Compared to the X thousand a year it would have cost to lock them up, how much damage are they doing?” In other words, simply keep an eye on the released prisoners’ recidivism rates, and stack up the crimes they commit during what would have been their final years in jail versus the dollars saved by not having to lock them up. Researchers could also monitor the benefits to society accrued by releasing the offenders early, whether through tax revenue or whatever else.

Matt Osterrieder, a spokesman for the U.S. Sentencing Commission, said he couldn’t give a definitive answer about whether and how the commission would be tracking the 6,000 inmates’ outcomes, but he did highlight that the group has done such research in the past and views recidivism research as a priority. “There’s lots to learn, sure,” Kleiman said of recidivism and reintegration. “There’d be more to learn if they did it systematically.” He said he hoped the government is going to be tracking the data associated with the released prisoners very closely: “I would be very pleased if that were true.”

It’s worth pointing out that the science behind the release is pretty sound, according to Ann Jacobs, director of the Prisoner Reentry Institute at the John Jay School of Criminal Justice. She said there’s little evidence that chopping the last two years off a decadelong sentence — again, the average situation with the released inmates — would have any negative effect on the deterrent value of jail time. Rather, “What you really want to be doing is mainstreaming them so they are able to make more decisions for themselves and feel the consequences of those decisions, and be able to act on their ambition about being a different person or having a better life,” she said. “And if you lock ‘em up for too long sometimes that spark, that energy, that motivation, dissipates, and you end up with someone who’s depressing, someone who doesn’t have any hope or any energy. That last two years is not going to make the difference for most people” in terms of their risk of re-offending.

But whatever the literature says, as the 4,000 released inmates are tracked over the months and years to come and, inevitably, some re-offend, there will be many opportunities for people to score political points off of whatever happens. Jacobs says she hopes that when the recidivism numbers start trickling in, people understand that not all instances of recidivism are created equal.

All too often, she said, people lump together minor crimes and technical parole and probation violations — which can occur very easily and which, in the case of parole violations, accounted for a third of all prison admissions in 2009 — in the same category as serious re-offenses. “What do we mean by recidivism, and what was the actual crime, and were they actually convicted of it or just arrested for it — making those kind of distinctions is really important,” said Jacobs. Important or not, she said she’s not optimistic about what hard-core law-and-order types will say when some of the prisoners being released end up back where they started. “There are people who will use anything that happens as evidence that in fact we should be locking people up and throwing away the key, despite all the evidence that that doesn’t produce the results that we want,” she said. “These are the instances in which those people aren’t very interested in being evidence-based.”