Less than a month ago, London endured another terrorist attack. Wielding two knives and wearing a fake suicide vest, Usman Khan killed two people and injured three others. Khan had been imprisoned in 2012 for plotting attacks on the London Stock Exchange and Parliament, but was released “on license” (the British version of parole) in December 2018, subject to security measures including a government-run disengagement and desistance program and a GPS tag. He was considered, by some, a “poster boy” for rehabilitation.

While prison can rehabilitate some terrorists, it can further radicalize others. Unfortunately, experts are not very adept at assessing the risk — predicting who will return to terrorism, either after time in prison or returning from a conflict zone.

That’s a problem for the United States, which is set to release scores of terrorist offenders from federal prisons. And it will be a problem in the future, as the number of people convicted of terrorism steadily increases.

What factors are associated with a return to terrorism? Can you rehabilitate a terrorist?

The answer is yes. But it’s not easy.

What’s the risk?

The challenge in assessing who’s most likely to return to terrorism is that terrorists, by definition, want to remain undetected. That means we often don’t know whether they become involved again unless they are caught in the act, like Usman Khan. Further, adequately assessing risk requires following up for years, not just months. As Khan’s attack makes clear, just because someone hasn’t returned to terrorism today doesn’t mean he or she won’t tomorrow.

To sidestep these problems, we read autobiographies covering the life course of 85 terrorists from over 70 different groups. These 85 terrorists represent 185 cases of “engagement” in terrorism — as individuals may participate in and disengage from terrorism several times in their lives. In the 82 cases where the engagement ended in prison, we found that 64 percent returned to terrorism post-release. In the instances where the individual voluntarily walked away from terrorism, 58 percent went back. Of the 12 who disengaged after their group was “defeated” — as has happened with the Islamic State — 66 percent became involved again. Most of our sample hadn’t been involved in programming to help them disengage; nor had they been closely watched by parole officers or other security measures.

These statistics may seem alarming. But they are comparable to criminal recidivism rates in the United States and elsewhere, including for violent offenders.

Of course governments should continue to improve programming and security measures. But they’re doing a relatively good job at keeping the risk low. Among released prisoners who were convicted of jihadi terrorism in the United States and United Kingdom, research shows that recidivism rates are less than 7 percent to date. The low recidivism rate is probably due to strict security measures on terrorism offenders, long prison sentences in the United States and the relatively short time post-release that most have had to re-engage.

Who’s a risk?

There’s no single explanation for why individuals return to terrorism. However, using statistical analysis, we’ve identified a few general risk factors — and Khan had most.

First, as with criminal behavior, individuals generally “age out” of terrorism. Our data show the probability of re-engaging peaks in the mid- to late 20s, starts to decline around one’s early 30s, and drops below 20 percent by age 45. But Khan was relatively young. He’d first been convicted at 20; last month, when he attacked pedestrians on the London Bridge, he was only 28.

Second, Khan’s family and friends disowned him after his initial conviction. While this is understandable on their part, research shows that individuals who maintain ties with nonradical family and friends are more likely to walk away from terrorism. We also find that, as with other crimes, those who maintain ties to individuals still involved are much more likely to re-engage.

Third, it’s likely Khan remained deeply committed to a radical ideology — despite his claim that he wanted to change. That’s another risk factor. Before his arrest, surveillance caught Khan referring to non-Muslims as “dogs”; such dehumanizing language is often tied to violent ideologies. And even after release, Khan was preoccupied with the Hindu-nationalist Indian government’s takeover of predominantly Muslim Kashmir, where his family came from and where he originally sought to establish a terrorist training camp and institute sharia law. The grievances that may have motivated Khan’s initial involvement in terrorism were still present and probably exacerbated by recent events.

Fourth, leaders and violent operators may have a harder time disengaging. Khan had been both, at least within his al-Qaeda cell.

Finally, and beyond what we know about Khan, our research shows those from an upper- or middle-class family are less likely to re-engage. We also find that although marriage and employment deter criminal recidivism, they don’t generally prevent a return to terrorism in the short term.

Do terrorists need to be “de-radicalized?”

Not all terrorists need to be de-radicalized — because not all of them were “radicalized” to begin with. Some clearly believe in the ideology. Others only feign belief because they want the friendship, notoriety, money, or excitement, our interviews find.

Moreover, “deradicalization” is not a magic solution or even a leading driver of terrorist disengagement. Our autobiographies reveal that only 16 percent of decisions to walk away from terrorism were because an individual lost faith in the radical ideology (i.e., “de-radicalized”). Much more often, they were disillusioned with the strategy or actions of the terrorist group or its leaders, got tired of their daily tasks, or simply burned out.

Deradicalization or the absence of a strong commitment to the ideology can make a difference. We find that terrorists who are not or are no longer committed to the radical ideology are more open to other reasons to get out, like marriage, family, employment and financial gain. And we find former terrorists who still hold radical beliefs are significantly more likely to return to terror.

Who, then, is the biggest risk post-release — whether from prison or a Syrian camp? Lessons from the research will help policymakers and practitioners make the most successful decisions.

Mary Beth Altier (@marybethaltier) is an associate professor at New York University’s Center for Global Affairs.

Emma Leonard Boyle (@EmmaJLeonard) is an assistant professor of politics at LaSalle University.