Thursday, July 28, 2011

New Developments in COSA-land

Hello All:

There are a number of interesting developments in the world of Circles of Support & Accountability.

In an earlier blog, I referred to an international collaboration of COSA-friendly nations during a conference in Riga, Latvia. That collaboration has continued, and there is now a European Union group of folks continuing the dialog—proving that COSA is garnering truly world-wide interest.

At this fall's ATSA conference in Toronto, Mechtild Hoing and Bas Vogelvang of the Netherlands will present COSA in Europe: Circles Together for Safety during a time slot that also includes a presentation by Fresno, California’s Clare Ann Ruth-Hefflbower entitled Into the Future: Expansion of Circles of Support and Accountability (COSA) in the U.S. For those who are interested, Andrew McWhinnie, Eileen Henderson, and I will also be delivering a half-day preconference workshop on the topic, entitled Circles of Support and Accountability Comes Home.

To date, COSA has been seen largely as a Canadian and British endeavor—with those two nations having projects of considerable tenure (17 and 10 years, respectively). However, it will be important for projects outside of those two nations to take center-stage to ensure the ongoing development of the Circles movement (if you can call it that).

One of the greatest difficulties in seeding Circles of Support & Accountability—beyond recruitment of volunteers—has been establishment of the credibility of the model. Until recently, only two peer-reviewed evaluations have existed, each with admittedly small samples and short-term follow-up:

Wilson, R.J., Picheca, J.E., & Prinzo, M. (2007). Evaluating the effectiveness of professionally-facilitated volunteerism in the community-based management of high risk sexual offenders: PART TWO—A comparison of recidivism rates. Howard Journal of Criminal Justice, 46, 327-337. link to abstract

Wilson, R.J., Cortoni, F., & McWhinnie, A.J. (2009). Circles of Support & Accountability: A Canadian national replication of outcome findings. Sexual Abuse: A Journal of Research & Treatment, 21, 412-430. link to abstract

I am pleased to announce that a third, peer-reviewed evaluation of the COSA model is now available:

Bates, A., Macrae, R., Williams, D., & Webb, C. (2011). Ever-increasing circles: A descriptive study of Hampshire and Thames Valley Circles of Support and Accountability. Journal of Sexual Aggression. Published online first February 22, 2011 as DOI: 10.1080/13552600.2010.544415. link to abstract

In this recent review, data are presented showing the effectiveness of COSA in international jurisdictions. In the United Kingdom, COSA is managed by a registered charity known as Circles UK, which is charged with promoting use of this model in England and Wales. The UK COSA model is slightly modified from the Canadian original, in that most core members are still “under license” (British terminology for continuing to be supervised by Probation Services). As such, COSAs in the UK are more formally situated within the statutory framework of criminal justice (i.e., MAPPA), whereas Canadian COSAs tend to focus on core members who are entirely post-sentence. Notwithstanding this difference, the bulk of the COSA work in the UK model remains with the voluntary sector, rendering the two models more similar than not.

In their study, Bates et al. followed 60 core members for an average period of 36.2 months; however, with no control or comparison sample. Eight core members experienced behavioral difficulties related to sexual offending, resulting in five core members being recalled to prison for probation violations and two others being placed on a Sex Offence Prevention Order (i.e., an order of recognizance). One core member (1.7%) engaged in behavior leading to a new sexual offense conviction.

Overall, the UK findings are very much in line with those reported previously in Canada in the papers noted above. Interestingly, and as a further corollary to the Canadian data, the UK project also observed a harm reduction effect, in that the core member who reoffended was convicted for downloading child pornography whereas his previous convictions were for hands-on offenses against children.

So, with the Bates et al. data in hand, we can say with some strength that the COSA model is neither regionally nor nationally specific—meaning that similar findings can be found in COSA projects whether they be located throughout a particular nation or from nation to nation. However, as many will attest (cf. pop music), breaking through in the all-important American market is key. To date, only anecdotal data exist regarding the effectiveness of the COSA model in US jurisdictions.

Currently, there are well-established COSA projects in California, Minnesota, and Vermont, with additional projects in the development or early stages in Colorado, Washington, and North Carolina, among others. Reports from the Fresno, CA coordinator (the aforementioned Clare Ann Ruth-Heffelbower) are that there have been no documented cases of sexual reoffending among the 25 COSAs formed to date; although, there have been parole violations for technical reasons, some of which were related to risk for sexual reoffense.

I recently had the pleasure of meeting with folks involved in the Minnesota Department of Corrections’ COSA project, known as MnCoSA. They have established nearly 30 COSAs and are gearing up to evaluate data that will speak to the effectiveness of the model in their state. Of particular interest is the fact that the MN-COSA project has used a random assignment design—by far the most statistically-elegant research project on COSA to date. In speaking with Joann Dillavou and Bill Donnay (collaborator Grant Duwe was unable to attend) of the MN DOC, it appears that preliminary data are showing markedly similar reductions to those observed in Canada and the UK.

One last thing to note: I recently had the great pleasure of meeting with folks involved in two US-based COSA projects—specifically, those located in Denver, CO and Brattleboro, VT. Both Colorado and Vermont have been working on developing COSA projects for nearly a decade. At present, both projects are in receipt of funding from grants associated with the Second Chance Act.

All of us in the COSA world are particularly excited about developments emanating from projects run by our American and international partners.

Thursday, July 21, 2011

Guest Blog from Jill Levenson

Hello All.

Here is a guest blog from Florida's Jill Levenson, a public policy researcher and expert. In this piece, Jill discusses a variety of issues associated with evidence-based risk management of sexual offenders--focusing specifically on the SORNA. Please note that the majority of Jill's references are in Sage Journals, which means that ATSA members can access them for free from the Members Only side of RJW

With the SORNA deadline approaching next week, only eight states are in substantial compliance with the law. The Sex Offender Registration and Notification Act (SORNA), which is Title 1 of the Adam Walsh Act (AWA), requires states to meet minimum standards in their sexual offender tracking and disclosure procedures.

Why is it that over 80% of states are not complying, especially when facing a 10% loss of federal crime funding each year? Since its inception, certain pieces of SORNA have been quite controversial and, in many cases, "deal-breakers" for states. Aside from the fiscal reality that SORNA will cost more for most states to implement than they would lose through penalties, parts of the law contradict evidence and best practices that some states have spent years refining.

Take, for example, the rigid offense-based tier system that classifies offenders by their statutes of conviction. There is a seemingly intuitive logic to the belief that statute titles, crime degrees, and descriptions accurately portray the offense in question. We all know, however, that in many cases statutory language does a poor job of characterizing the crime—and an even worse job of characterizing the criminal. In other words, SORNA tiers confuse the seriousness of the crime with the risk to reoffend. Although the SORNA guidelines cleverly avoid using the word risk, there is a clear implication, based on the associated longer registration durations and more frequent updates, that the Tier 3 offenders are more dangerous and therefore necessitate closer monitoring for longer periods. As well, "tier enhancements based on recidivism" can be made, allowing an offender with more than one conviction to be upgraded to a Tier 3 based on the repetitive nature of the criminal behavior—again suggesting to the public that higher tier assignments are associated with an increased likelihood of recidivism.

Unfortunately, in most instances, offense definitions are an inadequate means for evaluating risk or determining supervision needs. They overestimate risk for most offenders, and underestimate risk for those who plea-bargained to lesser offenses. Andrew Harris and colleagues (2010) found that applying SORNA criteria in Ohio and Oklahoma greatly skewed the distribution of risk into the highest tier compared to the states' prior classification schemes using risk assessment tools. Naomi Freeman and Jeffrey Sandler (2010) found that SORNA tiers were less effective in identifying sexual recidivists in New York than assessment tools utilizing risk factors statistically associated with recidivism. Many states have spent the past two decades developing and refining evidence-based risk management systems, and although SORNA does not technically require states to abandon the use of those tools, it does require registration and notification decisions to be based on the federal guidelines rather than the state's risk assessment.

And, speaking of longer registration durations (25 years to life), they contradict research indicating that sexual offenders become less likely to reoffend with increased time spent offense-free in the community. According to Static-99 guidelines, an offender's risk declines by half after 5 years, and by half again after 10 years. As well, reoffending declines with age. Right now, 20% of America's sexual offenders are over 56 years of age. Over time, with longer registration durations, we should anticipate an aging sexual offender population of which a greater proportion will pose a lower risk to public safety.

SMART Office staff have claimed that the purpose of SORNA was never to prevent recidivism, but to provide a monitoring tool for law enforcement and to increase public awareness. Indeed, the AWA legislative intent does not mention reduced reoffending as a goal. But, isn't that a little disingenuous? Why the need for enhanced law enforcement monitoring and increased public awareness if not for the hope of preventing future sex crimes and protecting potential victims from the trauma of sexual assault? Do Americans really want to spend tens of millions of dollars on sexual offender notification if it doesn't pay off in terms of protecting communities? Surveys of the American public have found that most citizens support SORN laws because, in fact, they believe them to be effective in reducing sexual recidivism. About a dozen empirical investigations (see summary in Levenson, 2009) have been conducted to date on the effectiveness of SORN laws to reduce recidivism. Most have not shown declines in sex crime rates generally or sexual recidivism specifically that can be attributed to SORN. Interestingly, the two studies that have noted significant effects on recidivism were conducted in Washington and Minnesota—both states with empirically derived risk assessment systems and concordant monitoring and management plans.

Or, consider the requirement that minors adjudicated delinquent in juvenile courts must be registered as sexual offenders. Though the SORNA guidelines conceded to states that public notification was discretionary, even registering juveniles with law enforcement contradicts a century of juvenile justice philosophy in the United States. The labeling of minors, and the subsequent potential obstacles to educational, occupational, and social opportunities, seems counter-intuitive to both juvenile advocates and victim advocates. This component of SORNA has been a noted barrier to implementation for many states.

We've seen the harrowing story of Jaycee Dugard in recent weeks, and all of us wondered how a registered sexual offender could get away with holding someone captive for 18 years. We've heard the scathing reports of parole agent incompetence, and we've used our 20/20 hindsight to shrewdly point out the missed clues. There is no doubt in anyone's mind that Garrido was high risk; he'd had two previous arrests, one resulting in a conviction for the abduction and rape of a woman whom he held hostage until she escaped.

But, perhaps, the real answer to the unfathomable question—How could this happen?? —is simply that Philip Garrido was a needle in a haystack. Philip Garrido is an example of how overzealous SORN laws might actually be counterproductive. As the registered sexual offender population nears three quarters of a million, fiscal and personnel resources are spread thin, law enforcement agents' capacity to target their monitoring toward more dangerous individuals is compromised, and the public's ability to distinguish truly high-risk predators is diluted. As Robin Wilson astutely pointed out in his recent San Francisco public radio interview, cases like Jaycee (and Adam Walsh, Jessica Lunsford, and Megan Kanka) are exceedingly rare, which is why they command so much attention—they shock us. They are statistically improbable events not representative of the vast majority of sexual assault cases and, thus, not the best cases on which to base public policy designed to address a rampant social problem.

Oh, and what about those 100,000 missing sexual offenders discussed repeatedly in AWA legislative hearings and in news reports? The ones who supposedly take advantage of discrepancies in laws to shop around for lax states where they can more easily neglect to register and avoid detection while seeking new child victims? My colleagues and I (Levenson & Harris, 2011) analyzed data downloaded directly from the online registries, along with surveys of registry managers, and we discovered that approximately 4% of the nation's RSOs have addresses that cannot be verified. We estimate the number of noncompliant sexual offenders to be between 25,000 and 30,000—but less than half of them are officially designated by states to have truly absconded. And, incidentally, research conducted in Minnesota, South Carolina, New York, and New Jersey all conclude that failure to register does not typically contribute to the likelihood for sexual recidivism.

Perhaps the apparent lack of state enthusiasm for SORNA will provide an impetus to re-open the dialogue about sexual offender management. SORNA in its current form sacrifices precision for over-inclusion, quality for quantity, efficiency for symbolism. By going back to the drawing board, with input from victim advocates, law enforcement, psychologists, criminal justice practitioners, and academic researchers, Congress has an opportunity to create model legislation that is evidence-based and more likely to achieve the goal of preventing repeat sexual violence.

Jill S. Levenson, Ph.D., LCSW

Associate Professor, Department of Psychology

Lynn University

Boca Raton, FL


Freeman, N. J., & Sandler, J. C. (2010). The Adam Walsh Act: A False Sense of Security or an Effective Public Policy Initiative? Criminal Justice Policy Review, 21, 31-49.

Harris, A. J., Lobanov-Rostovsky, C., & Levenson, J. S. (2010). Widening the Net: The Effects of Transitioning to the Adam Walsh Act Classification System. Criminal Justice and Behavior, 37(5), 503-519.

Levenson, J. S. (2009). Sex offense recidivism, risk assessment, and the Adam Walsh Act. Sex Offender Law Report, 10(1).

Levenson, J. S., & Harris, A. J. (2011). 100,000 Sex Offenders Missing … Or Are They? Deconstruction of an Urban Legend. Criminal Justice Policy Review, doi: 10.1177/0887403411415398.