Sentencing and the Limits of Actuarial Risk Assessment

As child molesters go, Cory Reibel seems a relatively low-risk proposition.  He is a first-time offender, was not sexually abused himself as a child, and victimized a girl instead of a boy — studies indicate that all of these factors point to a reduced risk of recidivism.  Yet, he was sentenced to the statutory maximum of 30 years in prison by a judge who wanted to prevent him from offending again.

The judge’s sentence seems to fly in the face of the science of risk assessment.  Actuarial risk assessment (that is, the determination of an offender’s risk based on a statistically sound analysis of recidivism data involving other offenders with similar characteristics) seems to be playing an increasingly prominent role in both pretrial release and post-conviction sentencing decisions.  Scientifically speaking, this is pretty clearly an advance on pure intuition as a basis for predicting risk.  However, actuarial risk assessment does present some important ethical difficulties when it is used as a basis for determining how severe a punishment should be.

These difficulties were on display earlier today when the Seventh Circuit turned aside Reibel’s challenge to the reasonableness of his sentence.  

Reibel relied on an earlier Seventh Circuit decision, United States v. Miller, 601 F.3d 734 (7th Cir. 2010), in which the court overturned an above-guidelines sentence that had been based on an unsupported (and probably incorrect) belief that sex offenders have a higher than normal recidivism rate and cannot be deterred from reoffending.  Read for all it’s worth, Miller might be understood as pointing in the direction of a general requirement that sentences based on recidivism prevention must be justified by reference to good actuarial data.

So why did Reibel lose?  The court cited four reasons.  First, the sentencing judge in Reibel did not make statements that were as strong or as dubious as the judge in Miller.  For instance, the Reibel judge characterized it as merely an “open question” whether sex offenders could be effectively deterred.  Second, while Miller received an above-guidelines sentence, Reibel’s was within the range, which meant that it could be presumed reasonable.  Third, the scientific studies on which Reibel relied in the court of appeals had not actually been presented to the district court, so there was a flavor of forfeiture in the case.

Finally — and what I find most intriguing — the Seventh Circuit raised the ethical objections:

[T]o tie sex offenders’ sentences to the statistics Reibel presents in his brief would be repugnant: offenders would be able to secure a shorter sentence by molesting girls rather than boys; offenders who were once victims would receive longer sentences than those who were not; and abusers of young children would receive shorter sentences than those whose victims were older.  (7-8)

The court did not elaborate further on this point, so I am left wondering how far the court would push it.

Given the high level of discretion sentencing judges have in this post-Booker world and the presumption of reasonableness that attaches to within-guidelines sentences, Reibel was really a pretty easy case.  Imagine, though, a not-so-unrealistic variation on the facts.  Assume that Reibel presents his studies to the district court and the judge is persuaded; determining that Reibel is unlikely to reoffend, the judge gives him a sentence well below the guidelines minimum of 30 years.  What if the government now appeals the sentence?  Would the Seventh Circuit really be willing to say that the ethical objections are so strong that a sentencing judge is not permitted to use the best available risk-assessment techniques in selecting a sentence?

Considering everything the Supreme Court has said about judicial sentencing discretion post-Booker, I doubt that the Seventh Circuit would go so far as to rule out this type of use of risk assessment.  (The use in risk assessment of certain well-recognized suspect categories, though, like race and national origin, might be another matter.)  Whether, as a matter of public policy, courts should take into account the sorts of factors Reibel urged is a more difficult question.

Reibel also had an interesting discussion about the use of a statutory maximum as a sentencing benchmark.  Reibel, who was convicted of producing child pornography, argued that many other defendants convicted of this crime did things that were much worse than he did.  Although it’s hard to say anything even indirectly positive about Reibel’s offense, he’s probably actually correct on this point: he had only one victim, produced only four images (as far as I can tell from the Seventh Circuit opinion), and seems not to have caused any serious physical injuries.  Other have produced far more images of many more victims and used much more violence along the way.  Although he’s certainly not anywhere near the most benign of kiddie porn producers, he’s also far from the most egregious.

Given that he’s not among the worst of the worst, Reibel argued that it was unreasonable to give him the statutory maximum.

The argument is premised on a principle that is hard to argue with: the worst punishments should go to the worst offenders.  This principle is, of course, an aspect of the proportionality ideal, and it has animated much the Supreme Court’s recent Eighth Amendment jurisprudence.

The problem with Reibel’s argument was not that proportionality doesn’t matter, but that the “worst of the worst” among kiddie porn producers likely do get longer sentences than Reibel.  How?  Through multiple counts and consecutive sentences.  Here’s what the Seventh Circuit said:

[O]ffenders worse than Reibel can be given consecutive sentences or prosecuted separately for child molestation (or another crime). Reibel correctly points out that the consecutive-sentence option is available only for defendants facing multiple charges, but we have difficulty imagining an offender worse than Reibel who could neither be charged with more than one child-pornography count nor prosecuted separately for a related crime. (6, citations omitted)

In this context, the court’s reasoning seems commonsensical, but, pushed hard enough, it might undermine the usefulness of the statutory maximum as a baseline for sentencing reasonableness.  Why?  In light of the expansive and redundant nature of the federal criminal code and the weakness of Double Jeopardy Clause limitations on count-multiplication, it will almost always be possible to imagine that the “worst of the worst” of those convicted of any given crime would surely be convicted on other counts, too, and sentenced consecutively.

The full opinion in United States v. Reibel (No. 11-3416) (per curiam) can be found here.

Cross posted at Seventh Circuit Cases.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.