The Volokh Conspiracy

Mostly law professors | Sometimes contrarian | Often libertarian | Always independent

Sex Crimes

Denying Bail

The Arizona Supreme Court got it right: categorical denials of bail to persons charged with sexual assault violates the Constitution.


In a posting here on the Volokh Conspiracy last week ("Is a Categorical Denial of Bail for Sex Offenders Constitutional?"), Paul Cassell summarized an amicus brief he co-authored urging the US Supreme Court to grant certiorari in Arizona v. Goodman, a case involving a provision of Arizona law under which pre-trial bail must be denied to persons charged with sexual assault where "the proof is evident or the presumption great that the person is guilty." [The Scotusblog page for the case is here.]

Arizona's general bail statute provides (as, I believe, do the statutes of all other States) that courts may deny bail to persons charged with a felony, but only after an individualized determination that the defendant "poses a substantial danger to another person or to the community."*

*The Arizona general bail statute, AZ Rev. Stat. 13-3961, provides that a person charged with a felony offense "may not be admitted to bail" where the court finds, after a pre-trial bail hearing: (a) that there is "clear and convincing evidence that the person charged poses a substantial danger to another person or the community"; (b) that "no condition or combination of conditions of release may be imposed that will reasonably assure the safety of the other person or the community"; and (c) that "the proof is evident or the presumption great that the person committed the offense."

The provision at issue in Goodman replaces this individualized determination with a categorical one: all persons charged with sexual assault are to be denied bail, notwithstanding the constitutionally-mandated presumption of innocence, without the need to produce any evidence showing that they are a danger to the community; they will, in effect, be presumed dangerous (at least, if "the proof is evident or the presumption great that the person committed the offense").

The Arizona Supreme Court struck this provision down as an unconstitutional deprivation of due process. Due process, the court held, requires an individualized determination of dangerousness unless the crime charged "inherently demonstrates that the accused will pose an unmanageable risk of danger if released pending trial." The court, by a 4-3 majority, found that it did not:

"[The law] is facially unconstitutional because it categorically prohibits bail without regard for individual circumstances. [It] does not provide any procedures to determine whether a person charged with sexual assault would pose a danger if granted pre-trial release. A court's finding that the proof is evident or the presumption great only shows a likelihood that an accused committed the charged sexual assault. It does not address the likelihood that an accused would commit a new sexual assault or other dangerous crime if released pending trial….

To be clear, courts can [and must] deny bail to a person charged with sexual assault when the proof is evident or the presumption great as to the charge and that person 'poses a substantial danger to another person or the community.' Before doing so, however, courts must engage in an individualized determination …"

Paul's brief argues they got it wrong: Arizona can, consistent with the Constitution, "categorically deny bail based on the charged offense if it can show that defendants charged with that offense categorically present a continuing danger to the community," and "a well-supported charge of sexual assault is a sufficient proxy for future dangerousness":

As this Court recognized in Smith v. Doe, releasing sex offenders pending trial presents a substantial danger to the community because the "risk of recidivism posed by sex offenders is 'frightening and high.'" 538 U.S. 84, 103 (2003) ("high rate of recidivism among convicted sex offenders" means they pose risk of future "dangerousness as a class")…. A substantial body of academic literature … confirms that sex offenders reoffend at extremely high rates — regardless of how reoffending is defined and regardless of subsequent offense that counts as reoffending, be it another sex crime, a different violent crime, or any other type of subsequent crime"

He is, in my opinion, wrong about that—wrong on the facts, and wrong on the law.

The Facts. The recidivism statistics** for sex offenders, as Paul's brief illustrates, have been much discussed in connection with this case (and many others involving sex offenders). Although the Supreme Court did indeed declare back in 2003 that the recidivism risk for sex offenders is "frightening and high"—"estimated to be as high as 80%," in Justice Kennedy's words—that turns out (as Ira and Tara Ellman have convincingly demonstrated here) to have been based upon no actual evidence whatsoever, having been derived from a single, unsupported, and entirely uncorroborated sentence in a 1986 article in Psychology Today. [Jacob Sullum, here at, discusses the remarkable and damaging persistence of this "frightening and high factoid" in his article here; see also Adam Liptak's NY Times article "Did the Supreme Court Base a Ruling on a Myth?"]. As Sullum puts it, "even if you think Arizona's bail ban is good policy (or at least constitutional), you should be troubled by the continued judicial reliance on repeatedly refuted claims about sex offenders that were erroneously endorsed by the Supreme Court 16 years ago."

**Note: Technically speaking, of course, the recidivism data for persons who have been convicted of sex offences is not strictly relevant to evaluation of this statute, insofar as the persons covered here have only been charged, and not convicted, of any crimes.

The actual data on sex offender recidivism are complicated, often contradictory, nuanced, and do not support the notion that rates are uniformly or significantly higher for sex offenders than for other categories of crimes. [For a summary of the many studies in this area, see the Scholars' Amicus Brief in Vasquez v. Illinois, here]. The Arizona Supreme Court got this right:

The State points to recidivism rates among sex offenders as evidence of the likelihood that sexual assault arrestees would commit a new sexual assault pending trial if released on bail. The cited empirical studies are not illuminating, however, as they concern a wide variety of sex crimes besides sexual assault, arrive at disparate conclusions, and for the most part do not focus on the relatively short time period between arrest and trial. Regardless, none of the studies cited reflects that most convicted rapists re-offend, the highest number being 5.6% reoffending within five years of release from prison. And the only cited study concerning accused rapists released on bail reflects that 3% committed another unspecified felony pending trial. [emphasis added]

The Law But interpretation of the recidivism statistics is not, in my opinion, at the heart of this case. The presumption of innocence—as fundamental a constitutional principle as we have—is at the heart of this case. To deprive an individual of his liberty because he is a member of a class that, statistically speaking, is more likely than others to commit a crime is an odious principle that is not consistent with the due process of law. Many categories of individuals, surely, have higher-than-average recidivism rates: unemployed males between the ages of 18 and 25; persons earning less than $25,000 a year; high-school dropouts; drug addicts; …

They are—all of them—presumed innocent until the State has persuaded a jury of their peers, beyond a reasonable doubt, that they are guilty as charged.

So, too, for persons charged with sex crimes, no matter how heinous. They can, as the Arizona court was at pains to acknowledge, be detained pre-trial in appropriate circumstances; but those circumstances must be such as to demonstrate that they—not just the members of the class to which they belong, but they themselves—are a continuing danger to others. It is a high bar. It is supposed to be, because it protects us all against the arbitrary imposition of punishment. The recidivism statistics might be relevant as evidence in a particular case; but they cannot be dispositive, lest we punish people, in effect, for the behavior of others whom they resemble in one way or another.