The Supreme Court’s Crucial Mistake About Sex Crime Statistics

Proponents of criminal justice reform never talk about sex offenders. They’re political untouchables subject to lifelong restrictions that continue long past their confinement, restrictions justified as necessary to protect the public from their propensity to re-offend. Two Supreme Court decisions established that justification. But they rely on a scientific study that doesn’t exist.

"Frightening and High"

McKune v. Lile, 536 U.S. 24, 33 (2002) rejected, 5-4, Robert Lile’s claim that Kansas violated his 5th Amendment rights by punishing him for refusing to complete a form detailing prior sexual activities that might constitute an uncharged criminal offense for which he could then be prosecuted. The form was required for participants in a prison therapy program; refusing to join the program meant permanent transfer to a higher security unit where he would live among the most dangerous inmates and lose significant privileges, including the right to earn the minimum wage for his prison work and send his earnings to his family. Justice Kennedy explained the treatment program helped identify the traits that caused “such a frightening and high risk of recidivism” among sex offenders—a rate he said “has been estimated to be as high as 80%.” The following year in Smith v. Doe, 538 U.S. 84 (2003) the Court upheld Alaska’s application, to those convicted before its enactment, of a law identifying all sex offenders on a public registry. It reasoned that the ex post facto clause was not violated because registration is not punishment, but merely a civil measure justified because the “risk of recidivism posed by sex offenders” is “frightening and high”, 536 U. S. at 34.

The idea that sex offenders repeat their crimes at high rates has fed legislation imposing increasingly harsh post-release burdens on them, nearly all triggered by being on a sex offender registry. Registrants may face residency restrictions sometimes severe enough to exclude them from entire cities and prevent them from living with their families, “presence restrictions” barring them from using public libraries or parks with their families, formal exclusion from many jobs, and informal exclusion from many more. The registration requirement typically extends for decades, and in some states, such as California, for life, with no path off the registry for most registrants. Courts have usually turned back challenges to registration and the consequences that flow from it; a Lexis search finds that in 91 cases the court’s opinion quotes Justice Kennedy’s dramatic statement that the sex offender recidivism rate is “frightening and high”. But is it? Do those convicted of sex offenses really re-offend 80% of the time, or anything close to that?

A "Statistic" With No Support

McKuneprovides just one citation for its much-quoted statement: a 1988 Justice Department “Practitioner’s Manual”. That reference likely came from the amicus brief supporting Kansas filed by the Solicitor General, then Ted Olson, which also cites it. This Practitioner’s Guide itself provides but one source for the claim, but it’s no scientific study. It’s a 1986 article from Psychology Today, a mass market magazine aimed at a lay audience, which had this sentence: “Most untreated sex offenders released from prison go on to commit more offenses–indeed, as many as 80% do.” Freeman-Longo, R., & Wall, R, Changing a lifetime of sexual crime, Psychology Today (1986). That sentence is a bare assertion with no supporting reference. Nor did its author have the scientific credentials needed to qualify at trial as an expert on recidivism. He was a counselor, not a scholar, and the article containing the sentence isn’t about recidivism statistics. It’s about a counseling program for sex offenders he then ran in an Oregon prison. His unsupported assertion about the recidivism rate for untreated sex offenders was offered to contrast with his equally unsupported assertion about the lower recidivism rate for those who complete his program.

So what is the re-offense rate for those convicted of a sex offense? One cannot calculate it without first defining “re-offense,” without specifying the time period to employ, and without considering whether the categories of people these laws label “sex offender” all present the same risk level. Consider first what counts as a re-offense. If the purpose of the sex offender registry is to aid the police in investigating sex offenses, or warn the public about persons thought likely to commit them, then we want to know the rate at which those convicted of a sex offense commit another one. That’s different than the rate at which they commit any offense that returns them to prison. When the California Corrections Department recently examined cases of sex offender registrants returned to prison for a new offense, they found that in 88% of the cases, the new offense was a parole violation. Parole violations are generally acts that aren’t crimes for anyone not on parole—things like going to a bar or visiting a friend who’s also an ex-felon. Only 1.8% of those re-incarcerated had committed a new sex offense.

The time period also matters. The most cautious measure would ask whether an offender ever commits another sex offense, but there’s the disadvantage that a sample limited to deceased offenders would necessarily exclude most released in the past ten or twenty years. There are studies that track people for long periods, however, and a recent meta-analysis by leading scholar Karl Hanson combined the data from 21 studies which followed nearly 8,000 offenders for an average of 8.2 years, and as long as 31. Sixteen of the 21 studies tracked offenders in other western countries (most often, Canada), allowing us to measure the re-offense rate we get without the distinctly harsh American system of long sentences and post-release restrictions. The studies examined different populations of offenders; some might be expected to present a higher risk of re-offense than others. Hanson used a well-established risk measure, the Static 99-R, to sort the offenders into three risk categories. Nearly 20% of the high-risk offenders committed a new sex offense within five years of release, and an additional 12% did so during the next 10 years. But the 68% who hadn’t committed a new sex offense fifteen years after release rarely did later. Indeed, none of the high-risk offenders who were offense-free after 16 years committed a sex offense thereafter.

This point is important because most people are typically put on registries for decades, and often for life; being offense-free for fifteen years or more won’t get them removed even though that history tells us they’re very unlikely to commit a new offense. Indeed, it’s mistaken to think of anyone offense-free for fifteen years as high-risk. At the time of their release we cannot tell which high-risk offenders will be among the two-thirds who won’t re-offend, but that is revealed over time. Those who haven’t re-offended after fifteen years are not high-risk for doing so.

And what about those who were not classified high-risk in the first place? About 97.5% % of the low-risk offenders were offense-free after five years; about 95% were still offense-free after 15 years. Some context can help here. About 3% of felons with no known history of sex offenses commit one within 4.5 years of their release. Of course, they’re not on the sex offender registry after release even though the chance of their committing a sex offense is the same or higher than the chance of a new sex offense by a either a low-risk offender, or a high-risk sex offender who has been offense free for fifteen years. What about the chance of a sex offender committing some other serious crime? Released sex offenders are actually less likely to commit a new felony of any kind, after release, than are other released felons.

Sex offender registries include a lot of people who are low-risk from the outset: a teenager who had consensual sex with another teenager, people who possessed erotic images of anyone under 18 but never even attempted to commit any contact offense, and even, depending on the state, someone convicted of public urination. A Justice Department study found that more than a quarter of all sex offenders were minors at the time of their offense. People may assume the registry’s purpose is to warn people about those who committed violent, coercive, or exploitative contact sex offenses, but they’re in fact filled up with people who never did any of those things.

Or, people who once did but are very unlikely to do so again because it’s been so long since they committed any crime. The Smith respondents who challenged the Alaska registry were classified as “aggravated” sex offenders, required under Alaska law to register four times a year for life, because they had been pled nolo contendere in 1984 to sexual contact with minors. Doe v. Otte, 259 F.3d 979 (9th Cir. 2001). They were released in 1990. One had completed a two-year post-release treatment program. The other had remarried and been granted custody of his daughter after psychiatric evaluations found he had "a very low risk of re-offending" and was "not a pedophile". Neither had re-offended in the twelve years since release, a fact that alone predicts a re-offense rate below 5%. Alaska posts the address and place of employment of all registrants “for public viewing in print or electronic form, so that it can be used by “any person” and “for any purpose.”Alaska Admin. Code tit. 13, § 09.050(a) (2000) as described in Doe v. Otte, 259 F.3d 979 (9th Cir. 2001).

Why The Re-offense Statistic Matters to Courts

There’s a constitutional problem with rules that justify substantial burdens on large groups of low-risk offenders by mistakenly classifying them high-risk. InIn the Interest of J.B., 107 A.3d 1 (Pa. 2014), the Pennsylvania Supreme Court struck down state rules required by federal law that placed juveniles over 14 on the state registry for 25 years if they committed a rape or “aggravated indecent assault”. The rules’ rationale was the legislative finding that such offenders “pose a high risk of committing additional sexual offenses” but the court objected that this finding just wasn’t true of juveniles, and therefore unconstitutionally burdened their ability “to obtain housing, schooling, and employment, which in turn hinders their ability to rehabilitate.” The California Supreme Court used different labels but a similar logic when it held this year, in In re Taylor, 60 Cal. 4th 1019, that it was unconstitutionally irrational to automatically subject every sex offender parolee in San Diego County to residency restrictions that impeded their rehabilitation and left many of them with no place to live. Once again, the problem with the statute was its application to every sex offender, without regard to their individual circumstances including an individualized assessment of each offender’s risk of re-offense.

The logic of these decisions offers hope for a wider judicial rationalization of the rules on sex offender registries, but to realize that hope, one must apply the principle adopted by both the Pennsylvania and California supreme courts to a correct understanding of the facts. The principle is that public safety policies that restrict and burden individuals cannot be based on sweeping generalizations about the risk posed by anyone who commits an act that puts him on a sex offender registry, given the fact that the risk varies across individual registrants in ways we can easily assess, and also declines over time for any individual who remains offense-free. The burdens of registration must be targeted on those who are in fact high-risk. But while these recent decisions offer hope, the Pennsylvania opinion also illustrates the difficulty of getting courts to understand the facts well enough to apply them properly. It rejected the law’s application to juveniles because of their low re-offense rate of (“between 2-7%”), but it failed to understand that the re-offense rate for many if not most adults on the registry is within the same 2-7% range, especially if one includes adults who have been on the registry fifteen years without a new offense.

Writing on a different subject entirely, Eula Biss recently observed:

Risk perception may not be about quantifiable risk so much as it is about immeasurable fear. Our fears are informed by history and economics, by social power and stigma, by myth and nightmares. And as with other strongly held beliefs, our fears are dear to us. When we encounter information that contradicts our beliefs, we tend to doubt the information, not ourselves.

The label “sex offender” triggers fear, and disgust as well. Both responses breed beliefs that do not yield easily to facts. That’s why even those politicians now urging criminal justice reforms conspicuously omit mentioning sex offenses when they argue for less punitive policies that would facilitate the offenders’ reintegration into civil society. Unfortunately, the Supreme Court has fed the fear. It’s become the “go to” source that courts and politicians rely upon for “facts” about sex offender recidivism rates that aren’t true. Its endorsement has transformed random opinions by self-interested non-experts into definitive studies offered to justify law and policy, while real studies by real scientists go unnoticed. The Court’s casual approach to the facts of sex offender re-offense rates is far more frightening than the rates themselves, and it’s high time for correction. Perhaps there’s now hope it may soon happen.

This is a condensation of an article that will appear in late 2015 in Constitutional Commentary. A more complete version with fuller documentation is available now at SSRN.