<< Occupational Hazard | Main | Justices' Speaking Engagements >>

Predicting Recidivism

Leon Neyfakh has this long article in the Boston Globe on predicting recidivism and the use of such predictions in sentencing and parole decisions.

Part of the controversy over the use of risk assessment tools goes to the dual purpose of punishment.  We punish for both utilitarian and retributive purposes.  The practical reasons to punish are to (1) deter other potential wrongdoers with the example, (2) incapacitate the individual wrongdoer, at least for a time, and (3) hopefully rehabilitate the individual.  Risk assessment informs us on the latter two points.

The retributive purpose is to punish evildoers because they deserve it, period.

The two philosophies of punishment are in harmony when it comes to punishing repeaters more severely than first-timers.  Recidivism is the single strongest predictor of repeating again, and the repeaters deserve more punishment as a matter of simple justice.  For other factors, though, things get more complicated.  Is it moral to punish someone more severely due to a "risk factor" beyond his control?
The story notes the notorious case of "Dominic Cinelli, a career criminal who had been paroled in 2008 while serving three concurrent life sentences for armed robbery."  He shot and killed a police officer last Christmas.

Cinelli's record on [risk assessment factors] was almost cartoonishly alarming. He was a heroin user starting at the age of 15 and had committed multiple armed robberies by the time of his arrest in 1976. He also tried to escape from prison, not once but twice, robbing a jewelry store and shooting a security guard the first time and stealing a gun from a sheriff the second. According to the state's review of the case, COMPAS, the tool now used by the Massachusetts parole board, would have assigned him a risk level of nine out of ten.

But the business of risk assessment isn't all common sense. Sometimes there are surprises. The 1993 study that formed the basis of a widely used tool known as the VRAG, for instance, is where researchers determined that in a sample of offenders who'd been locked up for serious violent crimes, those whose victims had been women were less likely to reoffend than those whose victims were men. The study also found that murderers were less likely to commit more crime upon release than people who'd merely injured their victims.

The factors in Cinelli's case we can and should use without hesitation.  The surprising factors in the second paragraph illustrate the hazards of defining the factors.  A closer look reveals that both are correlated with other factors.

"It is not a theory of crime," said VRAG creator Vernon Quinsey, a psychologist at Queens University in Canada. The factors on the questionnaire, he explained, aren't necessarily causes of criminal behavior: They just match certain patterns. "Some of the offenders we followed had killed their wives. Most of these guys are not career criminals and they have relatively low recidivism rates," he said. "Some of the offenders were homosexual pedophiles -- these guys have relatively high recidivism rates. These observations likely explain why murder is good and male victims bad in terms of recidivism."

In the case of the male victim factor, it is probably not correct to state it in terms of "those whose victims were men."  The pedophiles' victims were boys.

Applying risk assessment requires caution both from a validity standpoint and a morality standpoint.  A factor should not be used if its effect disappears when you control for another more salient factor.  The retributive aspect of punishment should also be kept in mind.  Even if murderers as a group did have lower recidivism than batterers after correcting for the prior criminal record factor, it would still be wrong to let them out earlier.


I draw a different conclusion, that being that we have gone too far in trying to quantify risk. I doubt that scales and measured "factors" are anything more than, at best, false precision, and, at worst, guessing impersonating science.

The unfortunate truth is that having a shrewd, balanced adult sit down with the offender and talk with him is, in my view, more likely to yield a sensible assessment than these "measurements" and scores.

It may be considered scandalous to say it out loud in a system as devoted to rationality as ours (quite properly) is (thus my support for sentencing guidelines, for example), but sometimes intuition is better than scoring, and assessing future risk is one of those times.

This is not to say that we should ignore a criminal's record. Of course we shouldn't. But thinking that we can decide future dangerousness based on his record, however finely analysed, strikes me more as social science bluster than as anything in which we can have confidence.

I understand that the comeback will be, "But you think we should have confidence in INTUITION instead???" To which my answer is: Think about how you make the most difficult and important decisions in your own life. Sure, you gather as much objective information as possible. You talk with smart and experienced people you trust. You hire experts. You think about how similar decisions have turned out in the past. But then what?

You assess how it feels to you.

Did a person of good sense and discernment ever once sit down with Mr. Cinelli and look him in the eye? And no, I don't mean did Cinelli get lots of read-the-list-of-questions interviews with this or that parole officer. I mean, did someone in authority ever actually have a conversation with Cinelli, and then, in a quiet moment, ask himself, "In his heart, has Cinelli become a person of normal good faith?"

Maybe. I don't know. But I doubt it.

There are zillions of clever and manipulative inmates. This makes risk assessment, well, risky (which in turn means we must be more circumspect about release than the defense bar would have us be). Even smart, experienced and intuitive people can be fooled. There is no way to be sure, to say the obvious.

This means two things. The first is that we should remind ourselves that doubtful cases must be resolved in favor of public safety, not gauzy "compassion." The second is that we have to approach risk assessment using all our tools, including intuition, rather than simply citation to "scoring" results.

I conducted hundreds of parole hearings in the federal system in the 80's. We used a risk assessment device and a parole guideline system that was used as the precursor to the current system of sentencing guidelines. This system was developed in the 70's after indeterminate sentencing and the rehabilitative model had fallen into disfavor.

As shrewd and discerning as I thought I was, there was simply no way to accurately determine when an offender had reached his "magic moment" and would go forth and sin no more. Hence the default to public safety oriented parole/sentencing guidelines.

Leave a comment