Emory Law Journal

Briefing the Supreme Court: Promoting Science or Myth?
Melissa Hamilton  *Senior Lecturer of Law & Criminal Justice, University of Surrey School of Law; J.D., The University of Texas School of Law; Ph.D (criminology), The University of Texas at Austin.

Introduction

The Supreme Court recently decided, in Packingham v. North Carolina, whether North Carolina’s ban on the use of social networking websites by registered sex offenders is constitutional.  1Packingham v. North Carolina, 137 S. Ct. 1730 (2017). The principal legal issue in the case was whether the ban violates the First Amendment’s right to freedom of speech. The Supreme Court found the law unconstitutional for that reason.  2Id. at 1738. The Court left states with the option to rewrite their statutes in a much stricter manner to comply with the opinion’s First Amendment analysis. See Melissa Hamilton, Sex Offenders, Social Media and the Supreme Court: Why Have the Justices Relaxed Restrictions?, Newsweek (July 21, 2017, 11:39 AM), http://www.newsweek.com/sex-offenders-social-media-and-supreme-court-why-have-judges-relaxed-640260. But a legal commentator contends that the Internet’s pliability will likely not allow such attempts to prevail. See James Trusty, Do Criminal Defendants Have Web Rights?, The Crime Report (Aug. 17, 2017), https://thecrimereport.org/2017/08/17/do-criminal-defendants-have-web-rights/.

Yet another issue arose in the briefing and oral arguments before the Supreme Court. The litigants and certain amici curiae engaged in some debate about whether such a restriction is necessary in the first place. That is, various parties argued about whether the ban serves to protect the public from what North Carolina and the representatives of twelve other states in a collective amicus brief contend are high risk sex offenders who commonly use the internet to locate children for purposes of sexual exploitation.  3See Brief of Respondent-Appellee, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194); Brief for State of Louisiana and Twelve Other States as Amici Curiae Supporting Respondents, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194) [hereinafter States’ Amicus Brief]. In opposition, Packingham’s submissions, as the individual petitioner, and the amicus brief by a group of sex offender treatment professionals refute such allegations.  4See Petition for Writ of Certiorari, Packingham v. North Carolina, 137 S. Ct. 368 (Mar. 21, 2016) (No. 15-1194); Brief for Association the Treatment of Sexual Abusers et al. as Amici Curiae Supporting Petitioner, Packingham v. North Carolina, 137 S. Ct. 368 (Dec. 22, 2016) (No. 15-1194) [hereinafter ATSA Brief].

This debate is important because it goes to the heart of the foundational basis of North Carolina’s justification for the ban. The Supreme Court has previously approved civil restrictions on sex offenders, such as public registries and residency restrictions, based on the belief that their recidivism risk is “frightening and high.”  5McKune v. Lyle, 536 U.S. 24, 34 (2002). The Supreme Court’s support for this assertion derives from an article in the trade magazine Psychology Today, in which the authors claimed that the recidivism rate was up to 80%. Robert E. Freeman-Longo & Robert V. Wall, Changing a Lifetime of Sexual Crime, Psychol. Today, Mar. 1986, at 58, 58. The lead author has since admitted that the 80% estimate may once have been valid but now repudiates it as far too high a figure. Joshua Vaughn, Closer Look: Finding Statistics to Fit a Narrative, Sentinel (Mar. 25, 2016) http://cumberlink.com/news/local/closer_look/closer-look-finding-statistics-to-fit-a-narrative/article_7c4cf648-0999-5efc-ae6a-26f4b7b529c2.html. The author likewise argues that sex offender policies built upon such a façade are not supported by scientific evidence and are dysfunctional. Id. Yet some experts point out that the scientific evidence is to the contrary.  6See generally Ira Mark Ellman & Tara Ellman, “Frightening and High”: The Supreme Court’s Crucial Mistake About Sex Crime Statistics, 30 Const. Comment. 495 (2015); Melissa Hamilton, Constitutional Law and the Role of Scientific Evidence: The Transformative Potential of Doe v. Snyder, 58 B.C. L. Rev. E. Supp. 34 (2017). News reporters have noticed and have run stories about the Packingham case, specifically challenging the Supreme Court’s previous rulings that uphold sex offender restrictions.  7E.g., Radley Balko, The Big Lie About Sex Offenders, Wash. Post (Mar. 9, 2017), https://www.washingtonpost.com/news/the-watch/wp/2017/03/09/the-big-lie-about-sex-offenders/?utm_term=.ddb7fd592b34. The headline in the New York Times reads “Did the Supreme Court Base a Ruling on a Myth?”  8Adam Liptak, Did the Supreme Court Base a Ruling on a Myth?, N.Y. Times (Mar. 6, 2017), https://www.nytimes.com/2017/03/06/us/politics/supreme-court-repeat-sex-offenders.html?hp&action=click&pgtype=Homepage&clickSource=story-heading&module=second-column-region&region=top-news&WT.nav=top-news&_r=2. Similarly, Slate Magazine’s coverage leads with “The Supreme Court’s Sex-Offender Jurisprudence is Based on a Lie.”  9David Feige, The Supreme Court’s Sex-Offender Jurisprudence Is Based on a Lie, Slate (Mar. 7, 2017, 11:47 AM), http://www.slate.com/articles/news_and_politics/jurisprudence/2017/03/sex_offender_bans_are_based_on_bad_science.html.

The arguments concerning the government’s purported need for a social networking ban refer to various statistical studies of sex offenders. This Essay contends that the case materials in Packingham v. North Carolina in support of the ban contain significant misunderstandings in conceptualizing and conveying the scientific evidence about the dangerousness of sex offenders. Such a conclusion is particularly distressing, as these errors are contained in briefs and oral arguments before the Supreme Court of the United States in an important constitutional case. The Supreme Court’s majority opinion decided the case without relying, one way or the other, upon these scientific assertions. But if the justices had relied upon the version of the scientific evidence offered by the states in deciding Packingham, they would have continued to be misled about the risks involved. This Essay addresses why the studies that North Carolina and its amici offered are more akin to junk science than valid representations of the empirical evidence as applicable to the group of sex offenders to whom the ban targets.

This Essay proceeds as follows. It first summarizes the background to Packingham. The following three sections review the main arguments that North Carolina and the thirteen states as friends of the court make concerning the risk of sex offenders using social media to exploit minors. Alongside are analyses of the validity of the scientific evidence they cite.

I. The Packingham Case

The North Carolina legislature passed the Protect Children from Sexual Predators Act in 2008. The law bans the use of commercial social networking websites (SNS) by registered sex offenders.  10N.C. Gen. Stat. Ann. § 14-202.5 (West 2014), invalidated by Packingham v. North Carolina, 137 S. Ct. 1730 (2017). Violating the ban constitutes a felony.  11Id.

Lester Gerard Packingham (Petitioner), was convicted of violating the statute for creating a Facebook profile.  12State v. Packingham, 748 S.E.2d 146, 149 (N.C. Ct. App. 2013). He challenged his conviction on First Amendment grounds.  13Id. Petitioner won in the state appellate court, which ruled that the statute was vague and arbitrarily burdened registered sex offenders’ free speech rights.  14Id. at 154. The North Carolina Supreme Court reversed.  15State v. Packingham, 777 S.E.2d 738, 741 (N.C. 2015) In a non-unanimous ruling, the majority found the statute was neither vague nor did it unduly infringe upon First Amendment rights.  16Id. In addition, the majority determined that the law appropriately fit “the government’s interest in protecting children from registered sex offenders who are lurking on social networking Web sites and gleaning information on potential targets.”  17Id. at 749.

Packingham petitioned for a writ of certiorari.  18Petition for Writ of Certiorari, supra note 4. The United States Supreme Court granted the writ and held oral arguments on February 27, 2017.  19Oral Argument, Packingham v. North Carolina (No. 15-1194), https://www.supremecourt.gov/oral_arguments/audio/2016/15-1194.

In its briefing and in oral arguments before the Supreme Court, North Carolina made three claims about the risk of sex offenders. The State argued that these claims were supported by social science and were in keeping with common sense. The first was a broad allegation that registered sex offenders as a group have a “notoriously high recidivism rate.”  20Brief of Respondent-Appellee, supra note 3, at 14–15. The second claim was that sex offenders typically are crossover offenders, meaning individuals who have sexually offended adult victims also sexually victimize children.  21Id. at 41. The State relied upon evidence of crossover offending to justify the ban’s application to all registered sex offenders, not just those who have previously victimized minors. North Carolina’s third assertion concerning risk was that registered sex offenders “commonly” use SNS to sexually exploit children.  22Id. at 33. As a result, North Carolina contended that it needed to restrict registered offenders’ use of SNS to proactively prevent such exploitation.  23Id. at 37. The next Sections review each claim.

II. Recidivism Risk

The State of North Carolina asserted that registered sex offenders have a “notoriously” high rate of sexual recidivism.  24Id. at 14–15. The only empirical support North Carolina provided was a statistic from a Department of Justice document published in 2003.  25Id. at 41 (citing Dominique A. Simons, Sex Offender Typologies, in Sex Offender Management and Assessment and Planning Initiative 55, 61–62 (Christopher Lobanov-Rostovsky et al. eds., 2014) [hereinafter Smart Report]). This report, aptly titled “Recidivism of Sex Offenders Released from Prison in 1994,” contains the findings of a study tracking the rearrests of almost 10,000 sex offenders released from fifteen state prisons in 1994 (the DOJ Recidivism Study).  26Patrick A. Langan et al., U.S. Dep’t of Justice, Recidivism of Sex Offenders Released from Prison in 1994, at 24 (2003), https://www.bjs.gov/content/pub/pdf/rsorp94.pdf [hereinafter DOJ Recidivism Study]. The study collected a fairly representative sample for the United States, as it consisted of two-thirds of all male sexual offenders released in the country in that year.  27Id. at 1.

North Carolina points to the DOJ Recidivism Study’s finding that the sex crime rearrest rate for convicted sex offenders was four times higher than for non-sex offenders.  28Brief of Respondent-Appellee, supra note 3, at 37–38 (citing DOJ Recidivism Study, supra note 26, at 61–62). The respondent’s brief pinpoint cites the DOJ Recidivism Study at pages 61–62, which is a mistake. The correct reference would be to page 93. The multiple of four that North Carolina highlighted is correct, but the State’s lawyers were also hiding the ball. The given result is not directly applicable to registered sex offenders, as the DOJ Recidivism Study did not differentiate registered from non-registered. Further, the DOJ Recidivism Study indicated that 5.3% of released sex offenders were arrested on a new sex crime.  29DOJ Recidivism Study, supra note 26, at 1. Then, as a sign that recidivism studies that rely upon arrest data may overreach in counting failures, the reconviction rate of sex offenders for new sex crimes was 3.5%.  30Id. at 2. This means that one-third of those arrested for new sex crimes were not convicted of those charges.  31Id. at 1–2. Moreover, neither statistic—rate of arrests or convictions—supports any type of “notoriously high” risk designation for sex offenders that North Carolina trumpets.

The Petitioner also cited the DOJ Recidivism Study, but to highlight additional results. He found it confirmed that “empirical evidence refutes widely-held assumptions about dangers posed by registrants.”  32Petition for Writ of Certiorari, supra note 4, at 40 n.6. In this respect, Packingham promoted two findings from the DOJ Recidivism Study: (1) the general recidivism rate (i.e., reoffending with any type of crime) for convicted sex offenders was significantly lower than for those convicted of other types of crimes, and (2) offenders previously incarcerated for nonsexual crimes accounted for six times more new sex crime arrests than those whose prior convictions were for sex crimes.  33Id. (citing DOJ Recidivism Study, supra note 26, at 14, 24). The implication from these results is if the government truly hopes to target reductions in general recidivism and in sexual recidivism specifically, then it ought to focus more on non-sex offenders.

Perhaps realizing that the DOJ Recidivism Study is not very supportive of a “notoriously high” recidivism rate for sex offenders, North Carolina resorted to reflecting that the state’s own legislature and the United States Supreme Court have previously subscribed to the notion that sex offenders are highly likely to recidivate.  34Brief of Respondent-Appellee, supra note 3, at 40. The lawyer representing the State pointedly reminded the justices at oral arguments that the high court had in a prior case recognized that sex offenders are highly likely to repeat their crimes.  35Transcript of Oral Argument at 35, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194). North Carolina alternatively couched its conclusion on simple “common sense.”  36Brief of Respondent-Appellee, supra note 3, at 37. Yet common sense is not science and can be factually inaccurate.

At least the brief of the thirteen states acting as amici curiae (the States’ Amicus Brief) attempted to bolster the claim that a high percentage of sex offenders are recidivists by citing three additional scientific studies.  37Infra notes 38, 43, 45 and accompanying text. However, none of the underlying studies has any strong relevance to the risk of registered sex offenders as none of them distinguished registered from nonregistered. Further, none of the studies are generalizable to a population of convicted sex offenders in North Carolina for the reasons that are discussed below.

The States’ Amicus Brief declares, “One study showed that, over a twenty-five year period, fifty-two percent of persistent child molesters were rearrested for a new sex offense and thirty-nine percent of rapists were rearrested for a new sex offense.”  38States’ Amicus Brief, supra note 3, at 2–3 (citing Tim Bynum et al., Ctr. for Sex Offender Mgmt., Recidivism of Sex Offenders 8–9 (2001)); see also Robert A. Prentky et al., Recidivism Rates Among Child Molesters and Rapists: A Methodological Analysis, 21 Law & Hum. Behav. 635 (1997) (cited in Bynum et al., supra). However, the underlying study is of limited value here. The study is dated, using a sample of sex offenders released between 1959 and 1985.  39Prentky et al., supra note 38, at 637, 640. The recidivism rates quoted by the States’ Amicus Brief are not the observed rates of recidivism, but merely projected rates using a technique called survival analysis.  40Id. at 642. As a result, one of the original study’s authors has warned that the estimated rates should not be cited as actual rates.  41Jill Levenson, Sex Offense Recidivism, Risk Assessment, and the Adam Walsh Act, 10 Sex Offender L. Rep. 1 (2009). More importantly, the study is not generalizable to any degree, as the study sample was entirely composed of men prosecuted as “sexually dangerous” persons and thereafter civilly committed to a secure, inpatient mental health hospital.  42Prentky et al., supra note 38, at 637. Hence, the sample is only representative of an extremely select group of those presenting the highest risk, and is further distinguishable as having been diagnosed with severe mental illness.

The States’ Amicus Brief touted two other studies in their efforts to promote the idea of repeat offending with respect to child molesters. It maintained, “A five-year follow-up study found that, of persons who had committed child molestation, fifty-three percent of same-sex offenders and forty-three percent of opposite sex offenders had already been convicted of previous sex offenses.”  43States’ Amicus Brief, supra note 3, at 10 (citing Bynum et al., supra note 38, at 8–9). In turn, the Bynum et al. report links these results to the report of Vikki Henlie Sturgeon & John Taylor, Report of a Five-Year Follow-Up Study of Mentally Disordered Sex Offenders Released from Atascadero State Hospital in 1973, 4 Crim. Just. J. 31 (1980). The underlying study poses similar problems in its ability to represent the recidivism of a general population of sex offenders. The sample was entirely comprised of civilly committed sex offenders with diagnosed severe mental disorders who were released from hospitalization in 1973.  44Sturgeon & Taylor, supra note 43, at 31.

Then the States’ Amicus Brief cited a third study, stating that it “showed that thirty-one percent of extra-familial child molesters were reconvicted of a second sexual offense within six years.”  45States’ Amicus Brief, supra note 3, at 10 (citing Bynum et al., at 8–9). In turn, the Bynum et al. report links these results to Marnie E. Rice et al., Sexual Recidivism Among Child Molesters Released From a Maximum Security Psychiatric Institution, 59 J. Consulting & Clinical Psychol. 381 (1991). Again, empirical issues plagued its relevance to this case. The underlying study was conducted on patients released from a maximum-security psychiatric institution between 1972 and 1983.  46Rice et al., supra note 45, at 381. This study is also not on point for another reason: the site of the study was in Canada. In empirical terms, the results of the three studies the States’ Amicus Brief cites here are biased, with obvious validity concerns being represented as relevant to the risk of a heterogeneous group of registered sex offenders in North Carolina.

In contrast, the amicus brief on behalf of the Association for the Treatment of Sexual Abusers (and other groups) provided evidence of sexual recidivism studies from more appropriate samples.  47ATSA Brief, supra note 4, at 10. This brief cited results from studies of released sex offenders in seven different states in America, showing sexual recidivism rates in the low single digits (most around 3%),  48Id. which is relatively consistent with the DOJ Recidivism Study results.

III. Crossover Offending Risk

The next scientific debate concerns crossover offending. North Carolina argued that its SNS ban is not overbroad in applying to all registered sex offenders. The state contended that “[r]esearch shows a high cross-over rate for sexual offenders,” meaning that offenders with adult victims frequently molest children as well.  49Brief of Respondent-Appellee, supra note 3, at 41. North Carolina’s brief asserted that a “majority of studies find[] ‘rates in the range of 50 to 60 percent’” for crossover offending, citing a publication produced by the Department of Justice’s Office of Sex Offender Sentencing, Monitoring, Apprehending, Registering, and Tracking (with the office having the acronym “SMART,” thus herein the “Smart Report”).  50Id. at 61 (citing Simons, in Smart Report, supra note 25, at 55, 61–62). This statement is misleading in that a “majority of studies” does not refer generally to sexual recidivism studies. The Smart Report’s reference was actually to studies of only male offenders that specifically focused on crossover offending and used individual self-reports as the methodology (as opposed to other measurements such as official statistics in the form of arrests or convictions).  51Simons, in Smart Report, supra note 25, at 61.

For the proposition of the 50–60% marker, the Smart Report cites five self-report studies. At the outset, it is evident that the description of the five studies, representing the “majority” of self-report studies on sexual recidivism, is a gross overstatement; the literature contains many more.  52See generally Holly Kleban et al., An Exploration of Crossover Sexual Offending, 25 Sexual Abuse 427 (2012) (collecting studies). In any event, the next Part analyzes the validity and reliability of those five studies in terms of whether they provide sufficient evidence for the State’s claim on the prevalence of crossover offending.

A. Assessing the Evidence on Crossover Offending

None of the five studies that the Smart Report cites as evidence of high levels of crossover offending are representative samples suitable for generalizing to a U.S.-based population of those convicted of sex offenses of all varieties. For example, each study relied upon convenience samples of individuals who were voluntarily or involuntarily in treatment programs for sexual deviance.  53E.g., Gene G. Abel & Candace Osborn, The Paraphilias: The Extent and Nature of Sexually Deviant and Criminal Behavior, 15 Psychiatric Clinics N. Am. 675 (1992) (studying a sample of patients seeking treatment at two hospitals); Kim English et al., Colo. Dep’t of Pub. Safety, The Value of Polygraph Testing in Sex Offender Management (2000) (studying offenders in sex offender treatment programs). From a scientific perspective, this type of nonprobability sampling means there is a high likelihood of selection bias and sampling error.  54Reg Baker et al., Report of the AAPOR Task Force on Non-Probability Sampling 15–26 (2013), https://www.aapor.org/AAPOR_Main/media/MainSiteFiles/NPS_TF_Report_Final_7_revised_FNL_6_22_13.pdf.

There are additional grounds for regarding the five studies as not generalizable outside their own contexts. Four of the studies were based simply on one or two sites,  55E.g., Peggy Heil et al., Crossover Sexual Offenses, 15 Sexual Abuse 221, 224 (2003) (describing inmates and parolees in a sex offender treatment program in Colorado); Daniel Wilcox et al., Sexual History Disclosure Using the Polygraph in a Sample of British Sex Offenders in Treatment, 34 Polygraph 171, 173 (2005) (describing individuals in a sex offending treatment program at a fixed site in Britain); Michael A. O’Connell, Using Polygraph Testing to Assess Deviant Sexual History of Sex Offenders 38 (June 12, 1997) (unpublished Ph.D. dissertation, University of Washington) (individuals seeking treatment at a clinic in Washington), UMI No. 9807012. thus severely limiting inferences to other populations. Two of the studies included individuals in their samples who had not committed officially recognized sex crimes,  56Abel & Osborn, supra note 53, at 675; O’Connell, supra note 55, at 38–40. rendering those samples inapposite to a population of convicted sex offenders, as there are risk relevant differences between the two populations. For the foregoing reasons, researchers in at least three of the five studies conceded in their papers that their research subjects did not constitute representative samples.  57Gene G. Abel et al., Self-Reported Sex Crimes of Nonincarcerated Paraphiliacs, 2 J. Interpersonal Violence 3, 20–21 (1987) (discussing same sample and results as in Abel & Osborn, supra note 56); O’Connell, supra note 55, at 77; Wilcox et al., supra note 55, at 177.

Additionally, the studies are bedeviled by questionable methodological choices that render them inherently unreliable. A few of these troubling issues are discussed herein.

1. Imprecision in Defining Sexual Offending

First, each of the five studies counted as offenses various behaviors that do not necessarily constitute crimes in the first instance, and do not necessarily involve human contact. For instance, one study (English et al.) counted as offenses with victims such things as obscene phone calls, voyeurism, stalking, and Internet pornography viewing.  58English et al., supra note 53, at app. D. Another study (O’Connell) defined sexually deviant acts to include group sex, prostitution, peeping, and any sex with a male.  59O’Connell, supra note 55, at 46–47, 95 app. A. Then a third (Wilcox et al.) recorded as offenses to be acts such as obscene phone calls, prostitution, calls to sex hotlines, adultery, threesomes, nude bars, and homosexual behavior.  60Wilcox et al., supra note 55, at 182–83 app. 1. This means that the rates of crossover offending with adults and children as victims are likely exaggerated due to counting the foregoing types of behaviors along with forcible rapes and child molestation. The inclusion of behaviors that may be minor and fail to rise to the level of crimes is a facial validity problem, meaning that the definition applied in the studies does not truly reflect the concept of criminal offending.  61Josine Junger-Tas & Ineke Haen Marshall, The Self-Report Methodology in Crime Research, 25 Crime & Just. 291, 320 (1999).

2. Childhood Sexual Activity

Second, to the extent that the idea of crossover offending is suggestive of adults who offend against victims both above and below the age of eighteen, the studies provide weak support for such a vision. Researchers in each study tabulated sexual acts over the subjects’ lives; that is they obtained self-reports of lifetime sexual histories. Thus, offenses against minors included acts when the subjects themselves were minors. As an example, English et al.’s study indicated that nearly three-fourths of the sample recalled sexual offenses they committed when they were age thirteen or younger.  62English et al., supra note 53, at 40 tbl. 7. It appears that at least some of the “offenses” against child victims may not have constituted crimes either. As further evidence of this, English et al.’s results also counted as child molestation any sexual behaviors with other minors that the subjects engaged in when they themselves were eight years old or younger.  63Id. It is unlikely for children at such tender ages to be legally culpable of such crimes.

Nor do the researchers seem to distinguish perpetrator from victim when two minors engaged in sexual acts. Wilcox et al. counted as self-reported sexual offenses those that occurred when individuals were as young as six years of age.  64Wilcox et al., supra note 55, at 175. Another study defined child molestation based simply on age differences, including when both were minors.  65O’Connell, supra note 55, at 46, 47, 50–59. In sum, it appears that the so-called crossover-offending counts in these studies were not limited to conduct with children when the offenders were adults.

3. Reliability of Self-Reports

Third, all five studies relied upon self-reports by subjects during interviews with treatment staff, and the studies are further subject to question because researchers failed to externally validate the self-reported victims and offenses.  66Theodore P. Cross & Leonard Saxe, Polygraph Testing and Sexual Abuse: The Lure of the Magic Lasso, 6 Child Maltreatment 195, 201 (2001). In empirical terms, this means they could not establish concurrent validity, which would entail testing whether the information gleaned from self-reports is consistent with other sources.  67Junger-Tas & Marshall, supra note 61, at 322.

The failure to validate is particularly troublesome here as the victim and offense counts reported in these studies yield unrealistic numbers of victims and sexual offenses per interviewee. For example, Heil et al.’s report indicated that individual inmate subjects recounted sexually offending against up to 215 different victims (on average reporting 18 victims) and up to 6,075 specific offenses (on average identifying 137 offenses).  68Heil et al., supra note 55, at 228 tbl. I. O’Connell’s study of patients referred for sexual deviance assessments found that subjects admitted to an average of 290 specific instances of sexually deviant behavior through their youth and adulthood.  69O’Connell, supra note 55, at 48. In Wilcox et al.’s small sample of British probationers, subjects on average reported 82 contact sexual offenses (standard deviation of 188) plus 81 noncontact sexual behaviors (standard deviation of 218).  70Wilcox et al., supra note 55, at 174. The standard deviations in Wilcox et al. suggests that multiple subjects were somehow able to identify and recount literally hundreds of contact and noncontact sexual acts they had committed. These extreme numbers suggest that most of the behaviors were nonserious, and as experience with self-reports in criminological studies informs, such studies are ripe with overreporting when eliciting events that are nonserious or high frequency occurrences.  71Junger-Tas & Marshall, supra note 61, at 322.

Overall, it seems preposterous to assume that the examinees’ recollections were sound enough and sufficiently reliable to enable them to recount specifics about so many events and persons. Coupled with these studies’ tendencies to count sexual offenses perpetrated when the examinees were as young as six,  72Wilcox et al., supra note 55, at 175. the high number of “admissions” seems implausible. To this point, Abel and Osborn’s research found that adult offenders who reported having had a deviant sexual interest during childhood also admitted to committing an average of 380 sex offenses before reaching adulthood.  73Gene G. Abel & Joanne-L. Rouleau, The Nature and Extent of Sexual Assault, in Handbook of Sexual Assault 9, 13 (W.L. Marshall et al. eds., 1990) (discussing the same study and sample as in Abel & Osborn, supra note 56.

4. Controversies with Polygraph

A fourth issue is evident as researchers in four of the studies allegedly supported the idea of a high crossover offense rate by using polygraph testing to intentionally increase the number and scope of admissions.  74English et al., supra note 53, at 31; Heil et al., supra note 55, at 226; O’Connell, supra note 55, at 35; Wilcox et al., supra note 55, at 172. North Carolina’s brief did not mention the disputed nature of polygraph exams, but the Smart Report itself warns that using polygraph exams with sex offenders is a “controversial” practice, in part because of the “possibility of false admissions and an overstating of the number of victims.”  75Christopher Lobanov-Rostovsky, Sex Offender Management Strategies, in Smart Report 145, 150–51 (Christopher Lobanov-Rostovsky et al. eds., 2014). Critics contend that the way polygraph exams for sex offenders are orchestrated enhances the likelihood that honest polygraph takers will be judged untrue, while frequent liars will be judged as truthful.  76Gershon Ben-Shakhar, The Case Against the Use of Polygraph Examinations to Monitor Post-Conviction Sex Offenders, 13 Legal & Criminological Psychol. 191, 196 (2008). Indeed, studies of polygraph exams of sex offenders have indicated false-positive rates (innocent examinees judged as deceptive) are higher than false-negative rates (lying examinees perceived as truthful).  77Ewout H. Meijer et al., Sex Offender Management Using the Polygraph: A Critical Review, 31 Int’l J. L. & Psychiatry 423, 425 (2008) (citing studies).

Experts note several explanations for false positives in sex offender polygraph results. For one, innocent individuals who fear being wrongfully accused experience stronger physiological responses, which can read as deception.  78Id. at 424. It is also recognized that sexual history disclosure tests often include ambiguous questions, such that the individual’s deceptive results may simply mean that he is having difficulty determining whether his behavior fits within the definition.  79Tim F. Branaman & Sheree N. Gallagher, Polygraph Testing in Sex Offender Treatment: A Review of Limitations, 23 Am. J. Forensic Psychol. 45, 54 (2005). For example, the Heil et al. study posed this question: “Have you physically forced or threatened anyone 15 or older into having sexual contact with you?”  80Heil et al., supra note 55, at 226. The fluidity of language and behaviors in human interaction is so variable that it might not be entirely clear to an examinee whether persuasive strategies count as force or threat, or whether a specific contact qualifies as sexual in nature. Thus, an examinee’s confusion as to the question may influence a deceptive response. There is also a strong potential for confirmation bias in which the polygraph investigator’s own judgment may be influenced by preconceived expectations about the true extent of the examinee’s sexual deviance.  81Ben-Shakhar, supra note 76, at 198. An examiner may have internalized the presumption that most sex offenders are repeat offenders, which could influence the tone of the questions and the resulting judgments on the subject’s veracity if he denies having additional victims or committing more offenses.  82Nat’l Res. Council, The Polygraph and Lie Detection 89–90 (2003).

5. False Confessions

Another issue to acknowledge is that false confessions are often elicited in the context of sex offender treatment. English et al. noted that uncooperative sex offenders may exaggerate their sexual deviance in the treatment and polygraph process.  83English et al., supra note 53, at 46. O’Connell likewise conceded that the result of examinees in his study reporting on average about 300 sexually deviant behaviors may in part be explained as their “[w]anting to ‘pass’ the [polygraph] examination may have led them to over-estimate their deviant sexual histories, and the polygraph charts may not have picked up their exaggeration.”  84O’Connell, supra note 55, at 78. North Carolina’s brief then cited the Smart Report for the finding that “64–66 percent of incest offenders report sexually assaulting children who they were not related to.” Brief of Respondent-Appellee, supra note 3, at 41 (citing Simons, in Smart Report, supra note 25, 55, 61–62). Yet the three studies underlying the Smart Report’s assertion here were among those cited for the assertion that 50–60% of sex offenders with adult victims have also abused children just discussed. Id. Thus, this assertion also lacked sufficient and appropriate empirical support for the same reasons.

Incentives for progress in treatment may increase false admissions. The Heil et al. study compared polygraph-induced admissions between a group of prisoners and a group of parolees.  85Heil et al., supra note 55, at 225. The prisoners were rewarded for success in treatment with a transfer to a less secure facility and a reduction in sentence.  86Id. at 226–27. The parolees did not receive an analogous reward.  87Id. Thus, Heil et al.’s finding that the number of additional disclosures (whether true) following polygraphs for the sample of prisoners was significantly greater than the increased disclosures from the parolees who did not receive equivalent incentives may be evidence of this carrot-like effect of inducing potentially false admissions by the prisoners.  88Id. at 227–28.

Additional reasons may explain the role of polygraphs in inducing exaggerations. A polygraph examiner familiar with post-conviction sex offender treatment programs observes that program officials routinely challenge the credibility of every examinee, regardless of the polygraph results.  89Raymond Nelson, Testing the Limits of Evidence Based Polygraph Practices, 45 Polygraph 74, 78 (2016). He explains that as a result,

examinees [are] faced with a limited range of options, which may include accepting arbitrary consequences for making no admissions, making false admissions, or developing their skill at making safe admissions to placate or manipulate the polygraph examiner and referring agent into a sense of complacent satisfaction that they are extracting additional information by routinely questioning truthful examinees.  90Id.

Sex offenders may likewise falsely confess because they believe it is expected that they had previously unknown victims.  91Cross & Saxe, supra note 66, at 195, 200–01. Thus, observers note that many “[sex] offenders might have fabricated stories after deceptive test outcomes, in order to satisfy examiners or to obscure the actual reason for failing the test.”  92Meijer et al., supra note 77, at 426. The National Resource Council, a research committee of the National Academy of Sciences, recognizes that false confessions are more common than people may think and that polygraph interrogations, particularly those involving false-positive test results, are prone to generating erroneous admissions.  93Nat’l Res. Council, supra note 82, at 56.

For the foregoing reasons, North Carolina did not provide sufficient empirical evidence for its claim of a high rate of crossover offending against adults and children by sex offenders.

B. Assessing the Risk of Registered Sex Offenders to Children

North Carolina next proclaimed that “[r]egistered sex offenders are proportionately far more likely than members of the general public to sexually assault minors,” emphasizing such statement in bold and underlined type.  94Brief of Respondent-Appellee, supra note 3, at 37. The State asserted that this higher risk for registered offenders regarding children is “supported by social science.”  95Id. Yet the statistical measures it provided under that heading offer no authority for the risk of registered sex offenders, as opposed to nonregistered sex offenders. Nor did the State present any evidence for the conclusion that child victims are at higher risk of victimization by known sex offenders, registered or not. Instead, the State simply claimed that reported recidivism rates of sex offenders are underestimates because of the gross underreporting of sex crimes due to the victims’ shame.  96Id. at 38 (citing Reporting of Sexual Violence Incidents, Nat’l Inst. of Just. (2010), https://www.nij.gov/topics/crime/rape-sexual-violence/Pages/rape-notification.aspx (last modified Oct. 26, 2010). Here, North Carolina pointed to the Smart Report’s reference to a study finding that just 5% of rapes and child sexual assaults self-reported during treatment were reflected in official records.  97Brief of Respondent-Appellee, supra note 3, at 38 (citing Roger Przybylski, Adult Sex Offender Recidivism, in Smart Report 91 (Christopher Lobanov-Rostovsky et al. eds., 2014)). For the basis of this statistic, as cited in the Smart Report, see generally D. Simons et al., Utilizing Polygraph As a Risk Prediction/Treatment Progress Assessment Tool, 23rd Annual Conference of the Association for the Treatment of Sexual Abusers (2004).

Yet, the fact that self-reports of offenses do not equal official record counts does not in itself show that registered sex offenders are more likely to assault minors. The study underlying the 5% figure likely did not differentiate registered versus unregistered offenders as few registered offenders existed at the time of the research.  98News Release, Nat’l Ctr. for Missing & Exploited Children, Number of Registered Sex Offenders in the U.S. Nears Three-Quarters of a Million (Jan. 23, 2012), http://www.missingkids.com/News/page/4615. North Carolina did not cite any empirical research to substantiate its seeming presumption that underreporting is a greater problem when the perpetrators are previously identified sex offenders—as opposed to the general public. To the contrary, the same Smart Report the State so frequently cited indicates the opposite. The Smart Report states that those who have had prior contact with police are most likely to be arrested, charged, and prosecuted for new sex crimes.  99Przybylski, in Smart Report, supra note 97, at 91 (citing Wendy Larcombe, Sex Offender Risk Assessment: The Need to Place Recidivism Research in the Context of Attrition in the Criminal Justice System, 18 Violence Against Women 482, 493 (2012)). In sum, North Carolina failed to reveal what “social science” might bolster its claim about the higher risk to children specifically presented by registered sex offenders.

IV. Online Risk

The next empirical issue the Packingham materials address concerns evidence to support the notion that registered sex offenders pose a significant risk of sexually exploiting minors by means of SNS. North Carolina’s brief to the Supreme Court made the following claim: “Sexual predators commonly use social networking sites to cull information about minors.”  100Brief of Respondent-Appellee, supra note 3, at 33. It supported this assertion by citing two studies.

A. British Reports of Suspicious Online Activity

North Carolina’s brief claimed that “[o]ne study found that ‘48.5% of online child sexual exploitation reports received were linked to social networking sites.’”  101Id. at 33–34 (citing Christopher Hope, Facebook Is a ‘Major Location for Online Child Sexual Grooming’, Head of Protection Agency Says, Telegraph (Oct. 15, 2013. 3:47 PM), http://www.telegraph.co.uk/technology/facebook/10380631/Facebook-is-a-major-location-for-online-child-sexual-grooming-head-of-child-protection-agency-says.html. This statistic derives from an article in the British newspaper the Telegraph102Hope, supra note 101. The underlying source is a document generated by a division of the British national police agency concerning communications it had received from the public about possible online sexual exploitation.  103Child Exploitation & Online Prot. Ctr., Threat Assessment of Child Sexual Exploitation and Abuse 10 (2013). However, the report does not differentiate complaints that were substantiated as constituting a crime. Nor does the report distinguish whether the online exploiters were previously known sex offenders as compared to members of the general public. Plus, many of the reports did not suggest the involvement of any adults. For example, the report indicates that a majority of the reports involving sexually suggestive images of minors were self-generated without any coercion or exploitive acts by adults.  104Id. at 12. Further, the Telegraph article also quotes a British police official warning that much of the use of online social networks to contact children is by foreigners acting from outside Britain.  105Hope, supra note 101(“British children were being ‘harvested’ by foreign abusers who were getting access to children in their homes over the internet.”). The chief of the Child Exploitation and Online Protection Centre commented, “It is not uncommon to encounter situations where offenders in one country will target and harvest victims in a completely different part of the world.” Id. In sum, the 48.5% statistic fails to substantiate North Carolina’s need to ban registered residents from SNS.

B. The Online Exploitation Study

North Carolina’s brief cited a second study as purportedly supporting its claim about the prevalence of sexual predators gaming SNS to prey on children:

Another study showed that “in 82 percent of online sex crimes against minors, the offender used the victim’s social networking site to gain information about the victim’s likes and dislikes,” and “in 62 percent of online sex crimes against minors, the offender used the victim’s social networking site to gain home and school information about the victim.”  106Brief of Respondent-Appellee, supra note 3, at 34 (quoting Kimberly J. Mitchell et al., Use of Social Networking Sites in Online Sex Crimes Against Minors: An Examination of National Incidence and Means of Utilization, 47 J. Adolescent Health 183, 185 (2010)).

However, neither of the statements in quotation marks are actual excerpts from the underlying study’s report.  107The study found that “[t]hese cases involved offenders who were using victims’ SNSs to get information about the victims’: (a) likes or interests (82% of cases involving offenders using SNSs to access information), (b) home or school (65%).” Mitchell et al., supra note 106, at 185. The reference to “these cases” is only of those cases in which offenders used SNS to try to make contact with identified minors. Id. Moreover, the statistical measures reported by North Carolina significantly misrepresent the study’s actual findings. The study at issue was conducted by researchers with the Crimes Against Children Resource Center using results from the National Juvenile Online Victimization survey.  108Id. at 184. This survey queried a national sample of law enforcement agencies concerning arrests for online sex crimes against children (the Online Exploitation Study).  109Id. The Online Exploitation Study did not distinguish sexual predators as North Carolina’s claim suggested.  110Id. The study concerns individuals arrested for online sexual exploitation of children, whether or not previously known as sex offenders.  111Id.

Additionally, the 82% figure is not of all arrests for online sex crimes against minors as North Carolina’s brief conveys; it is the figure representing a small subset thereof.  112Id. at 185. The Online Exploitation Study did concern arrests for online sex crimes against minors, but cases in which offenders used SNS in such crimes comprised just one-third of those arrests.  113Id. Then the study divided arrests involving SNS into three groups: those involving identified victims, those with victims who were not identified, and those in which there were no real victims in the sense that they were cases in which undercover officers portrayed minors online.  114Id. The 82% figure actually concerns just those cases in which the offenders used SNS with identified victims, a small subset of the larger sample.  115Id.

Crunching data contained in the Online Exploitation Study, the calculated percentage of offenders arrested for online sex crimes against children who used SNS to gain information about the victim’s likes and dislikes is actually 22%. Also, in a significant majority of those cases, the offenders were not gaining access to details about actual minors, but to data manufactured by undercover officers posing as children online.  116Id. Thus, North Carolina’s version inflated the number of cases in which offenders explored SNS to access minors’ likes and interests by a factor of four.

Next, the 62% that North Carolina’s brief cited is mistaken on its face; the actual percentage is 65%.  117Id. But again, the State misquotes what the rate represents. The Online Exploitation Study found that 65% of the cases involving offenders using SNS in cases with identified minors specifically gained information about home or school. As with the other statistical measure, cases involving SNS and identified minors were only a small subset of online arrests. Overall, only 5% of cases of online sex crimes with minors included access to a child’s home or school information through SNS.  118Id. Hence, North Carolina greatly exaggerated the frequency of offenders using SNS to gain information about home or school in cases of online sexual exploitation. To make matters worse, the lawyer representing North Carolina at oral arguments in February 2017 repeated the same mistakes, erroneously reporting these same two results before the Supreme Court:

[W]e know from studies that about 82 percent of online sex crimes against children, social networking websites were used to gain information about their likes and dislikes. And 62 percent of online sex crimes use . . . social networking websites to gain home and school information. So we know that there’s a very high percentage of these offenders . . . who are using social networking websites to find out information.  119Transcript of Oral Argument, supra note 35, at 45.

Regrettably, these significant overstatements of the prevalence of offenders exploiting SNS could have misled the Supreme Court about the online dangers that registered sex offenders pose.

North Carolina’s brief also ignored an important conclusion from this same study. The researchers reflected that

when considered in the context of the entire spectrum of places online where police are arresting people for illegal sexual intentions, SNSs do not appear to present risk in and of themselves or a greater risk than other online sites where people can meet and interact (e.g., chat rooms). Findings from this article support previous data that suggests the reality about Internet-initiated sex crimes, particularly those in which sex offenders meet juvenile victims online, is different, more complex, and more serious but less archetypically frightening than the publicity about these crimes suggest.  120Mitchell et al., supra note 106, at 187.

So, while North Carolina’s representatives distorted the statistics, they also ignored this important warning from the same study.

C. The Online Predators Study

North Carolina’s brief additionally attempted to highlight the risk of registered sex offenders by referencing findings from a study titled “Trends in Arrests of ‘Online Predators’” (the Online Predators Study).  121Janis Wolak et al., Crimes Against Children Research Ctr., Trends in Arrests of “Online Predators” (2009) [hereinafter Online Predators Study], http://scholars.unh.edu/cgi/viewcontent.cgi?article=1051&context=ccrc. Researchers with the Crimes Against Children Research Center also conducted this study using data from a survey of law enforcement agencies about arrests during two time periods, roughly 2000 and 2006.  122Id. at 2. The Petitioner’s brief in Packingham filed earlier had used the Online Predators Study results to highlight that 96% of those arrested for soliciting minors online were not registered sex offenders.  123Brief for Petitioner-Appellant at 58, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194). This result derives from the underlying study’s finding that 4% of online solicitation arrestees were registered sex offenders.  124Online Predators Study, supra note 121, at 2.

To counter this statistic, North Carolina instead pointed to the finding in the Online Predators Study that the percentage of such arrests involving registered sex offenders actually doubled from 2000 to 2006.  125Brief of Respondent-Appellee, supra note 3, at 38–39 (citing Online Predators Study, supra note 121, at 7–8). This argument is problematic because it undercuts the State’s implication that the increase in the percentage of registered persons to be arrested for soliciting youth online means that greater restraint of registered offenders is required. As more and more Americans become registered sex offenders because of the expanding scope of such laws, it makes statistical sense that the proportion of those arrested for any crime would happen to be registered. Indeed, if a state simply required everyone to register, regardless of their histories, the percent of online solicitors who were registered would rise to 100%. Then governmental officials could (albeit unreasonably) argue that there was an even greater need to monitor all residents because everyone is at risk of being an online solicitor. North Carolina explained that the increase was not surprising, as registries were in their infancy in 2000 and the percentage increased as more people were registered during that time frame.  126Id. Oddly, North Carolina’s argument here failed to explain why registered sex offenders are at greater risk considering the significant surge in the number of registered sex offenders as registries ramped up during the same time period might explain the increase. Moreover, North Carolina ignored the researchers’ conclusion in the Online Predators Study. Considering that the percentage of registered sex offenders in the online sex crimes was not more than 4%, the study authors found that these small statistics mean “aiming strategies to prevent online predation at [the] population [of registered sex offenders] may have limited utility because so few online predators are registered sex offenders.”  127Online Predators Study, supra note 121, at 6.

The Online Predators Study also disavows North Carolina’s claim that the Internet is fueling a wave of sexual exploitation of children. The researchers concluded, “While there was an increase in arrests of offenders using the Internet to seek sex with minors [from 2000 to 2006], there was during the same period a decrease in reports of overall sex offenses against children and adolescents and a decrease in arrests for such crimes.”  128Id. at 3. The researchers further explain:

[T]he facts do not suggest that the Internet is facilitating an epidemic of sex crimes against youth. Rather, increasing arrests for online predation probably reflect increasing rates of youth Internet use, a migration of crime from offline to online venues, and the growth of law enforcement activity against online crimes.  129Id. at 2.

At least the States’ Amicus Brief attempted to specifically address the risk of registered sex offenders in claiming that “[r]egistered sex offenders account for four to five percent of online solicitors of undercover police officers.”  130States’ Amicus Brief, supra note 3, at 9 (citing Michael C. Seto, Internet Sex Offenders 183 (2013)). The source cited for this statistic in turn referred to three studies.  131Seto, supra note 130, at 183. However, none of the three studies actually addressed registered sex offenders. Instead, the three studies found that of the samples investigated, 4%–5% of those arrested for online solicitations had prior sex offense arrests or convictions, not that they were registered.  132Id. The two groups are not synonymous. Individuals with prior sex offenses may not be registered and those registered might not have been charged or convicted of sex crimes.  133See generally Ofer Raban, Be They Fish or Not Fish: The Fishy Registration of Nonsexual Offenders, 16 Wm. & Mary Bill Rts. J. 497 (2007).

The trouble with statistics does not end here. The States’ Amicus Brief attempted to compute the percentage of adult Americans who are registered sex offenders. In what it calls “the States’ math” the brief concluded that one-third of 1% of American adults are registered sex offenders.  134States’ Amicus Brief, supra note 3, at 9–10 n.7. The States’ math is wrong. It included in the numerator the number of registered sex offenders in the United States. The denominator used the number of adult Americans.  135Id. The “States’ math” notes that “there are about 323,127,513 Americans, of which 22.9% are adults and 77.1% are not.” Id. at 10 n.7. Obviously, the “States’ math” erroneously switches the percentage of adults and children here. The mathematical lapse is that the numerator contains registered sex offenders who are juveniles. Hence, the formula is incorrect, rendering the final percentage also mistaken.

Conclusion

In its amicus brief, an association of sex offender treatment professionals correctly emphasized the “myth of homogeneity” concerning sex offenders.  136ATSA Brief, supra note 4, at 7. Instead, scientific research indicates “registrants are not a homogenous group of ‘sex offenders’ that should be monolithically managed. Rather, registrants comprise a diverse group of individuals, each different from the next in terms of past criminal history, behavioral patterns, and risk of recidivism.”  137Id. at 3. Further, the experts properly warned that policies that target sex offenders, which are not based on some empirical reality, are unlikely to be effective.  138Id.

In the end, North Carolina and thirteen other states weighing in as friends of the court in Packingham v. North Carolina offered a troubling version of the scientific evidence in an attempt to support a significant ban on registered sex offenders’ use of SNS. It is not clear if the states’ legal representatives were merely naïve and uneducated on the true science behind the empirical studies they touted. The alternative—that they intentionally tried to mislead the Supreme Court on the risks of sex offenders as a group—would be regrettable for ethical and political reasons. Perhaps fortunately, the Supreme Court was able to render its finding on the unconstitutionality of the North Carolina statue without such questionable evidence of risk.

The case may be momentous for another reason. As Professor Wayne Logan reads the opinion, Packingham “suggest[s] a possible softening of the Court’s customary unequivocal backing of laws imposing harsh sanctions on convicted sex offenders.”  139Wayne Logan, SCOTUS Invalidates Law Criminalizing Sex Offender Access to Social Media, Collateral Consequences Resource Ctr. (June 20, 2017), http://ccresourcecenter.org/2017/06/20/scotus-invalidates-law-criminalizing-sex-offender-access-to-social-media/.

Footnotes

*Senior Lecturer of Law & Criminal Justice, University of Surrey School of Law; J.D., The University of Texas School of Law; Ph.D (criminology), The University of Texas at Austin.

1Packingham v. North Carolina, 137 S. Ct. 1730 (2017).

2Id. at 1738. The Court left states with the option to rewrite their statutes in a much stricter manner to comply with the opinion’s First Amendment analysis. See Melissa Hamilton, Sex Offenders, Social Media and the Supreme Court: Why Have the Justices Relaxed Restrictions?, Newsweek (July 21, 2017, 11:39 AM), http://www.newsweek.com/sex-offenders-social-media-and-supreme-court-why-have-judges-relaxed-640260. But a legal commentator contends that the Internet’s pliability will likely not allow such attempts to prevail. See James Trusty, Do Criminal Defendants Have Web Rights?, The Crime Report (Aug. 17, 2017), https://thecrimereport.org/2017/08/17/do-criminal-defendants-have-web-rights/.

3See Brief of Respondent-Appellee, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194); Brief for State of Louisiana and Twelve Other States as Amici Curiae Supporting Respondents, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194) [hereinafter States’ Amicus Brief].

4See Petition for Writ of Certiorari, Packingham v. North Carolina, 137 S. Ct. 368 (Mar. 21, 2016) (No. 15-1194); Brief for Association the Treatment of Sexual Abusers et al. as Amici Curiae Supporting Petitioner, Packingham v. North Carolina, 137 S. Ct. 368 (Dec. 22, 2016) (No. 15-1194) [hereinafter ATSA Brief].

5McKune v. Lyle, 536 U.S. 24, 34 (2002). The Supreme Court’s support for this assertion derives from an article in the trade magazine Psychology Today, in which the authors claimed that the recidivism rate was up to 80%. Robert E. Freeman-Longo & Robert V. Wall, Changing a Lifetime of Sexual Crime, Psychol. Today, Mar. 1986, at 58, 58. The lead author has since admitted that the 80% estimate may once have been valid but now repudiates it as far too high a figure. Joshua Vaughn, Closer Look: Finding Statistics to Fit a Narrative, Sentinel (Mar. 25, 2016) http://cumberlink.com/news/local/closer_look/closer-look-finding-statistics-to-fit-a-narrative/article_7c4cf648-0999-5efc-ae6a-26f4b7b529c2.html. The author likewise argues that sex offender policies built upon such a façade are not supported by scientific evidence and are dysfunctional. Id.

6See generally Ira Mark Ellman & Tara Ellman, “Frightening and High”: The Supreme Court’s Crucial Mistake About Sex Crime Statistics, 30 Const. Comment. 495 (2015); Melissa Hamilton, Constitutional Law and the Role of Scientific Evidence: The Transformative Potential of Doe v. Snyder, 58 B.C. L. Rev. E. Supp. 34 (2017).

7E.g., Radley Balko, The Big Lie About Sex Offenders, Wash. Post (Mar. 9, 2017), https://www.washingtonpost.com/news/the-watch/wp/2017/03/09/the-big-lie-about-sex-offenders/?utm_term=.ddb7fd592b34.

8Adam Liptak, Did the Supreme Court Base a Ruling on a Myth?, N.Y. Times (Mar. 6, 2017), https://www.nytimes.com/2017/03/06/us/politics/supreme-court-repeat-sex-offenders.html?hp&action=click&pgtype=Homepage&clickSource=story-heading&module=second-column-region&region=top-news&WT.nav=top-news&_r=2.

9David Feige, The Supreme Court’s Sex-Offender Jurisprudence Is Based on a Lie, Slate (Mar. 7, 2017, 11:47 AM), http://www.slate.com/articles/news_and_politics/jurisprudence/2017/03/sex_offender_bans_are_based_on_bad_science.html.

10N.C. Gen. Stat. Ann. § 14-202.5 (West 2014), invalidated by Packingham v. North Carolina, 137 S. Ct. 1730 (2017).

11Id.

12State v. Packingham, 748 S.E.2d 146, 149 (N.C. Ct. App. 2013).

13Id.

14Id. at 154.

15State v. Packingham, 777 S.E.2d 738, 741 (N.C. 2015)

16Id.

17Id. at 749.

18Petition for Writ of Certiorari, supra note 4.

19Oral Argument, Packingham v. North Carolina (No. 15-1194), https://www.supremecourt.gov/oral_arguments/audio/2016/15-1194.

20Brief of Respondent-Appellee, supra note 3, at 14–15.

21Id. at 41.

22Id. at 33.

23Id. at 37.

24Id. at 14–15.

25Id. at 41 (citing Dominique A. Simons, Sex Offender Typologies, in Sex Offender Management and Assessment and Planning Initiative 55, 61–62 (Christopher Lobanov-Rostovsky et al. eds., 2014) [hereinafter Smart Report]).

26Patrick A. Langan et al., U.S. Dep’t of Justice, Recidivism of Sex Offenders Released from Prison in 1994, at 24 (2003), https://www.bjs.gov/content/pub/pdf/rsorp94.pdf [hereinafter DOJ Recidivism Study].

27Id. at 1.

28Brief of Respondent-Appellee, supra note 3, at 37–38 (citing DOJ Recidivism Study, supra note 26, at 61–62). The respondent’s brief pinpoint cites the DOJ Recidivism Study at pages 61–62, which is a mistake. The correct reference would be to page 93.

29DOJ Recidivism Study, supra note 26, at 1.

30Id. at 2.

31Id. at 1–2.

32Petition for Writ of Certiorari, supra note 4, at 40 n.6.

33Id. (citing DOJ Recidivism Study, supra note 26, at 14, 24).

34Brief of Respondent-Appellee, supra note 3, at 40.

35Transcript of Oral Argument at 35, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194).

36Brief of Respondent-Appellee, supra note 3, at 37.

37Infra notes 38, 43, 45 and accompanying text.

38States’ Amicus Brief, supra note 3, at 2–3 (citing Tim Bynum et al., Ctr. for Sex Offender Mgmt., Recidivism of Sex Offenders 8–9 (2001)); see also Robert A. Prentky et al., Recidivism Rates Among Child Molesters and Rapists: A Methodological Analysis, 21 Law & Hum. Behav. 635 (1997) (cited in Bynum et al., supra).

39Prentky et al., supra note 38, at 637, 640.

40Id. at 642.

41Jill Levenson, Sex Offense Recidivism, Risk Assessment, and the Adam Walsh Act, 10 Sex Offender L. Rep. 1 (2009).

42Prentky et al., supra note 38, at 637.

43States’ Amicus Brief, supra note 3, at 10 (citing Bynum et al., supra note 38, at 8–9). In turn, the Bynum et al. report links these results to the report of Vikki Henlie Sturgeon & John Taylor, Report of a Five-Year Follow-Up Study of Mentally Disordered Sex Offenders Released from Atascadero State Hospital in 1973, 4 Crim. Just. J. 31 (1980).

44Sturgeon & Taylor, supra note 43, at 31.

45States’ Amicus Brief, supra note 3, at 10 (citing Bynum et al., at 8–9). In turn, the Bynum et al. report links these results to Marnie E. Rice et al., Sexual Recidivism Among Child Molesters Released From a Maximum Security Psychiatric Institution, 59 J. Consulting & Clinical Psychol. 381 (1991).

46Rice et al., supra note 45, at 381.

47ATSA Brief, supra note 4, at 10.

48Id.

49Brief of Respondent-Appellee, supra note 3, at 41.

50Id. at 61 (citing Simons, in Smart Report, supra note 25, at 55, 61–62).

51Simons, in Smart Report, supra note 25, at 61.

52See generally Holly Kleban et al., An Exploration of Crossover Sexual Offending, 25 Sexual Abuse 427 (2012) (collecting studies).

53E.g., Gene G. Abel & Candace Osborn, The Paraphilias: The Extent and Nature of Sexually Deviant and Criminal Behavior, 15 Psychiatric Clinics N. Am. 675 (1992) (studying a sample of patients seeking treatment at two hospitals); Kim English et al., Colo. Dep’t of Pub. Safety, The Value of Polygraph Testing in Sex Offender Management (2000) (studying offenders in sex offender treatment programs).

54Reg Baker et al., Report of the AAPOR Task Force on Non-Probability Sampling 15–26 (2013), https://www.aapor.org/AAPOR_Main/media/MainSiteFiles/NPS_TF_Report_Final_7_revised_FNL_6_22_13.pdf.

55E.g., Peggy Heil et al., Crossover Sexual Offenses, 15 Sexual Abuse 221, 224 (2003) (describing inmates and parolees in a sex offender treatment program in Colorado); Daniel Wilcox et al., Sexual History Disclosure Using the Polygraph in a Sample of British Sex Offenders in Treatment, 34 Polygraph 171, 173 (2005) (describing individuals in a sex offending treatment program at a fixed site in Britain); Michael A. O’Connell, Using Polygraph Testing to Assess Deviant Sexual History of Sex Offenders 38 (June 12, 1997) (unpublished Ph.D. dissertation, University of Washington) (individuals seeking treatment at a clinic in Washington), UMI No. 9807012.

56Abel & Osborn, supra note 53, at 675; O’Connell, supra note 55, at 38–40.

57Gene G. Abel et al., Self-Reported Sex Crimes of Nonincarcerated Paraphiliacs, 2 J. Interpersonal Violence 3, 20–21 (1987) (discussing same sample and results as in Abel & Osborn, supra note 56); O’Connell, supra note 55, at 77; Wilcox et al., supra note 55, at 177.

58English et al., supra note 53, at app. D.

59O’Connell, supra note 55, at 46–47, 95 app. A.

60Wilcox et al., supra note 55, at 182–83 app. 1.

61Josine Junger-Tas & Ineke Haen Marshall, The Self-Report Methodology in Crime Research, 25 Crime & Just. 291, 320 (1999).

62English et al., supra note 53, at 40 tbl. 7.

63Id.

64Wilcox et al., supra note 55, at 175.

65O’Connell, supra note 55, at 46, 47, 50–59.

66Theodore P. Cross & Leonard Saxe, Polygraph Testing and Sexual Abuse: The Lure of the Magic Lasso, 6 Child Maltreatment 195, 201 (2001).

67Junger-Tas & Marshall, supra note 61, at 322.

68Heil et al., supra note 55, at 228 tbl. I.

69O’Connell, supra note 55, at 48.

70Wilcox et al., supra note 55, at 174.

71Junger-Tas & Marshall, supra note 61, at 322.

72Wilcox et al., supra note 55, at 175.

73Gene G. Abel & Joanne-L. Rouleau, The Nature and Extent of Sexual Assault, in Handbook of Sexual Assault 9, 13 (W.L. Marshall et al. eds., 1990) (discussing the same study and sample as in Abel & Osborn, supra note 56.

74English et al., supra note 53, at 31; Heil et al., supra note 55, at 226; O’Connell, supra note 55, at 35; Wilcox et al., supra note 55, at 172.

75Christopher Lobanov-Rostovsky, Sex Offender Management Strategies, in Smart Report 145, 150–51 (Christopher Lobanov-Rostovsky et al. eds., 2014).

76Gershon Ben-Shakhar, The Case Against the Use of Polygraph Examinations to Monitor Post-Conviction Sex Offenders, 13 Legal & Criminological Psychol. 191, 196 (2008).

77Ewout H. Meijer et al., Sex Offender Management Using the Polygraph: A Critical Review, 31 Int’l J. L. & Psychiatry 423, 425 (2008) (citing studies).

78Id. at 424.

79Tim F. Branaman & Sheree N. Gallagher, Polygraph Testing in Sex Offender Treatment: A Review of Limitations, 23 Am. J. Forensic Psychol. 45, 54 (2005).

80Heil et al., supra note 55, at 226.

81Ben-Shakhar, supra note 76, at 198.

82Nat’l Res. Council, The Polygraph and Lie Detection 89–90 (2003).

83English et al., supra note 53, at 46.

84O’Connell, supra note 55, at 78. North Carolina’s brief then cited the Smart Report for the finding that “64–66 percent of incest offenders report sexually assaulting children who they were not related to.” Brief of Respondent-Appellee, supra note 3, at 41 (citing Simons, in Smart Report, supra note 25, 55, 61–62). Yet the three studies underlying the Smart Report’s assertion here were among those cited for the assertion that 50–60% of sex offenders with adult victims have also abused children just discussed. Id. Thus, this assertion also lacked sufficient and appropriate empirical support for the same reasons.

85Heil et al., supra note 55, at 225.

86Id. at 226–27.

87Id.

88Id. at 227–28.

89Raymond Nelson, Testing the Limits of Evidence Based Polygraph Practices, 45 Polygraph 74, 78 (2016).

90Id.

91Cross & Saxe, supra note 66, at 195, 200–01.

92Meijer et al., supra note 77, at 426.

93Nat’l Res. Council, supra note 82, at 56.

94Brief of Respondent-Appellee, supra note 3, at 37.

95Id.

96Id. at 38 (citing Reporting of Sexual Violence Incidents, Nat’l Inst. of Just. (2010), https://www.nij.gov/topics/crime/rape-sexual-violence/Pages/rape-notification.aspx (last modified Oct. 26, 2010).

97Brief of Respondent-Appellee, supra note 3, at 38 (citing Roger Przybylski, Adult Sex Offender Recidivism, in Smart Report 91 (Christopher Lobanov-Rostovsky et al. eds., 2014)). For the basis of this statistic, as cited in the Smart Report, see generally D. Simons et al., Utilizing Polygraph As a Risk Prediction/Treatment Progress Assessment Tool, 23rd Annual Conference of the Association for the Treatment of Sexual Abusers (2004).

98News Release, Nat’l Ctr. for Missing & Exploited Children, Number of Registered Sex Offenders in the U.S. Nears Three-Quarters of a Million (Jan. 23, 2012), http://www.missingkids.com/News/page/4615.

99Przybylski, in Smart Report, supra note 97, at 91 (citing Wendy Larcombe, Sex Offender Risk Assessment: The Need to Place Recidivism Research in the Context of Attrition in the Criminal Justice System, 18 Violence Against Women 482, 493 (2012)).

100Brief of Respondent-Appellee, supra note 3, at 33.

101Id. at 33–34 (citing Christopher Hope, Facebook Is a ‘Major Location for Online Child Sexual Grooming’, Head of Protection Agency Says, Telegraph (Oct. 15, 2013. 3:47 PM), http://www.telegraph.co.uk/technology/facebook/10380631/Facebook-is-a-major-location-for-online-child-sexual-grooming-head-of-child-protection-agency-says.html.

102Hope, supra note 101.

103Child Exploitation & Online Prot. Ctr., Threat Assessment of Child Sexual Exploitation and Abuse 10 (2013).

104Id. at 12.

105Hope, supra note 101(“British children were being ‘harvested’ by foreign abusers who were getting access to children in their homes over the internet.”). The chief of the Child Exploitation and Online Protection Centre commented, “It is not uncommon to encounter situations where offenders in one country will target and harvest victims in a completely different part of the world.” Id.

106Brief of Respondent-Appellee, supra note 3, at 34 (quoting Kimberly J. Mitchell et al., Use of Social Networking Sites in Online Sex Crimes Against Minors: An Examination of National Incidence and Means of Utilization, 47 J. Adolescent Health 183, 185 (2010)).

107The study found that “[t]hese cases involved offenders who were using victims’ SNSs to get information about the victims’: (a) likes or interests (82% of cases involving offenders using SNSs to access information), (b) home or school (65%).” Mitchell et al., supra note 106, at 185. The reference to “these cases” is only of those cases in which offenders used SNS to try to make contact with identified minors. Id.

108Id. at 184.

109Id.

110Id.

111Id.

112Id. at 185.

113Id.

114Id.

115Id.

116Id.

117Id.

118Id.

119Transcript of Oral Argument, supra note 35, at 45.

120Mitchell et al., supra note 106, at 187.

121Janis Wolak et al., Crimes Against Children Research Ctr., Trends in Arrests of “Online Predators” (2009) [hereinafter Online Predators Study], http://scholars.unh.edu/cgi/viewcontent.cgi?article=1051&context=ccrc.

122Id. at 2.

123Brief for Petitioner-Appellant at 58, Packingham v. North Carolina, 137 S. Ct. 368 (2017) (No. 15-1194).

124Online Predators Study, supra note 121, at 2.

125Brief of Respondent-Appellee, supra note 3, at 38–39 (citing Online Predators Study, supra note 121, at 7–8). This argument is problematic because it undercuts the State’s implication that the increase in the percentage of registered persons to be arrested for soliciting youth online means that greater restraint of registered offenders is required. As more and more Americans become registered sex offenders because of the expanding scope of such laws, it makes statistical sense that the proportion of those arrested for any crime would happen to be registered. Indeed, if a state simply required everyone to register, regardless of their histories, the percent of online solicitors who were registered would rise to 100%. Then governmental officials could (albeit unreasonably) argue that there was an even greater need to monitor all residents because everyone is at risk of being an online solicitor.

126Id.

127Online Predators Study, supra note 121, at 6.

128Id. at 3.

129Id. at 2.

130States’ Amicus Brief, supra note 3, at 9 (citing Michael C. Seto, Internet Sex Offenders 183 (2013)).

131Seto, supra note 130, at 183.

132Id.

133See generally Ofer Raban, Be They Fish or Not Fish: The Fishy Registration of Nonsexual Offenders, 16 Wm. & Mary Bill Rts. J. 497 (2007).

134States’ Amicus Brief, supra note 3, at 9–10 n.7.

135Id. The “States’ math” notes that “there are about 323,127,513 Americans, of which 22.9% are adults and 77.1% are not.” Id. at 10 n.7. Obviously, the “States’ math” erroneously switches the percentage of adults and children here.

136ATSA Brief, supra note 4, at 7.

137Id. at 3.

138Id.

139Wayne Logan, SCOTUS Invalidates Law Criminalizing Sex Offender Access to Social Media, Collateral Consequences Resource Ctr. (June 20, 2017), http://ccresourcecenter.org/2017/06/20/scotus-invalidates-law-criminalizing-sex-offender-access-to-social-media/.