A multi-state evaluation of sex offender risk and recidivism using the Adam Walsh act tiers.
Although managing sex offender risk has been a law enforcement initiative since the 1930s, exceptionally heinous sex crimes perpetrated by offenders during the 1990s brought forth intense public scrutiny and demands for more rigorous monitoring of sex offenders. Legislators responded by enacting national standards and procedures for sex offender registration and community notification, passing the Jacob Wetterling Act in 1994 and Megan's Law in 1996. Stricter registration requirements and public access to offenders' residential information initially assuaged some of the public's fear, as parents felt the risk to their children had been minimized. As media reports surfaced concerning sex offender registration violations, however, fear was reignited. Consequently, sex offender management and policy measures were again under public scrutiny.Legislators responded on July 27, 2006, with passage of the Adam Walsh Child Protection and Safety Act (AWA), which resulted in more stringent registration requirements and established a standardized, offense-based classification system. Enforcement of this legislation has undoubtedly shaped states' criminal justice responses to sexual crimes and sexual offenders. The proposed offense-based classification system was, however, fashioned without reliance on or guidance by empirical validation. The essential question, therefore, is "Does this classification system accurately represent the risk of reoffense and lead to more effective sex offender management?" Critics have argued that the AWA classification system relies too heavily on the circumstances of the offense, not the overall recidivism risk posed by the offender. States have used this reasoning to resist AWA adoption, contending that existing classification methods, which are based on actuarial risk, would provide a better means of ensuring public safety.
In an effort to identify best practices and responsible public policy, this study funded by the National Institute of Justice sought to assess the relative effectiveness of competing classification schemes used in sex offender management. (1) Although this research attempts to address several research questions, the overarching goal is to compare the nationally recommended AWA classification tiers to actuarial risk assessment instruments in their respective abilities to identify high-risk individuals and recidivists.
Methods
Offender data of formerly incarcerated sex offenders were sampled from Florida, Minnesota, New Jersey and South Carolina in order to achieve the following research goals:
* To compare the nationally recommended AWA classification tiers with actuarial risk assessment instruments in their respective abilities to identify high-risk individuals and recidivists;
* To evaluate the predictive accuracy of existing state risk assessment classification schemes;
* To examine the distribution of risk assessment scores within and across tier categories as defined by AWA; and
* To examine the role of offender age in recidivism risk across the adult lifespan.
Eligible offenders were convicted sex offenders who had been released into the community between Jan. 1, 1990, and Dec. 31, 2004. Sexual offenses were defined as any index crime requiring registration and/or an end-of-confinement review. Additionally, offenders must have been released after confinement to the community and not to a civil commitment program. This yielded a final sample of 1,789 sex offenders.
Data was collected using available automated databases, supplemented by a review of prison and probation records. The study proceeded in two phases. Phase 1 included coding from each offender's available archival records data to calculate recidivism risk scores for two commonly used actuarial risk assessment instruments--the Static-99R and Static-2002R--and extracting relevant demographic and criminal history data (including juvenile justice) at the time of release into the community. Each criminal contact was categorized by the most serious charge. Phase 2 involved the coding of recidivism data for each offender. Variables collected and coded during Phase 2 focused on charge information. Where available, sex offender registry information was also collected, including the registry status at the time of the charge (registered vs. not registered); registration requirement (number of times per year required to register); and start date of initial registration.
This project used several analytic strategies aimed at addressing its primary questions. These strategies included:
* Detailed review of statutory codes for each state;
* Assignment of baseline tiers for each type of offense across three age groups--12 and under, 13-17, and 18 and up;
* Review of both instant offense and most serious offense fields and assignment of initial tiers based on this information;
* Review of supplemental fields in the data set to identify other cases in which the offender had a history of two or more sexual offenses, history of victimizing children under age 12, and/or history of use of force in commission of offenses; and
* As applicable, adjustment of initial tiers based on this review.
To evaluate the degree to which classification systems correctly classified/accurately predicted offender risk, the measures were assessed using the Receiver Operating Curve analyses. At different risk times, the significance of area under the curve was calculated for the three different criteria: actuarial risk scores, AWA level and state-determined tier level. Because key data required to accurately calculate actuarial risk scores was sometimes missing in the available archival files, a reduced actuarial predictor based on the items that were most commonly available was created. This is referred to as the available predictor (AP). This simplified actuarial scale was used whenever one of the standard actuarial predictors would have greatly reduced the sample size (N) available for analysis.
Results
The racial background of the sample group was 51 percent white, 31 percent black, 7 percent Latino, and 2 percent Native American. The mean age of the sample group was 33 years at sentencing, and the mean age of the sample at release was 37 years. The majority of offenders had no prior convictions for a sexual crime, but two-thirds had prior involvement in the criminal justice system for a criminal offense. For the cases in which victim characteristics were available, three-quarters of these offenders had unrelated victims (anyone not related to the offender, such as an acquaintance, stranger, friend, etc.), and about one-fifth had stranger victims (not known to the victim prior to the offense). Nearly half of all victims were age 12 or under.
The sexual recidivism rate for the sample was 5.1 percent during a five-year period and 10.3 percent during a 10-year period. There was an apparent trend for sexual recidivism rates to differ among states after five years, but this trend failed to reach significance. The trend reached significance after 10 years, with the highest rate occurring in Florida and the lowest rate in South Carolina.
Sex offenders were assigned to an AWA tier according to a consultant using the criminal codes and guidelines. Few offenders fit into the Tier 1 category--which defines offenses that are eligible for a sentence of less than one year in prison--and typically met criteria for a misdemeanor rather than a felony.
The researchers examined whether AP risk bands varied by AWA tier designation. If tier designations identify higher-risk offenders, cases with Tier 3 designations should also on average have higher AP scores. Concordance between AP risk band and tier designation was not, however, consistently identified. Cases with Tier 2 designations had higher mean AP scores (or higher percent of upper-risk band scores). Reflecting this surprising result, the AP risk hands only significantly predicted sexual recidivism for Tier 3 cases. An unexpectedly high sexual recidivism rate was observed for the AWA Tier 2/lower AP band cell.
The association between state and AWA tier designations and the 10-year recidivism rate was also examined. Results indicated that while the higher state-assigned tier was significantly associated with sexual recidivism in the expected, positive direction, a higher AWA tier was significantly associated with sexual recidivism in the unexpected, negative direction. In other words, riskier AWA tiers were associated with lower odds of sexual recidivism. The comparable analysis for five-year sexual recidivism yielded similar results, but were only statistically significant for the AWA tier.
Additionally, age has typically been found to have an inverse relation with sexual recidivism. This was true for both five- and 10-year sexual recidivism predictions, although the effect was greater and significant only for the longer follow-up period. Similarly, the interaction between age and whether or not the offender had perpetrated a crime against a child under age 13 was not statistically significant. The results indicated that increased age is protective of future reoffending, regardless of whether it is the age at which the commitment offense occurred, age at sentencing or age at release from incarceration.
This study found that the higher AWA tier was unrelated to reoffending in three of the four states, and was negatively associated with reoffending in one state. Where there is an effect for the AWA tier, Tier 2 offenders, those with shorter registration terms and fewer registration expectations, were more likely to reoffend than Tier 3 offenders. It is noteworthy that this unexpected inverse finding was in Florida, which is the only state in the sample that has been certified as substantially compliant with the AWA by the federal government.
Policy Implications
This study is one of the first of its kind to investigate procedures commonly used to classify risk among contemporary American sex offenders. The study is potentially useful for facilitating the interface between science and practice, as well as for informing the development and implementation of sex offender policies in the U.S. More important from a policy standpoint, the already existing tiering systems in the states studied outperformed AWA tiers in predicting reoffending, as did the AP scores. Additionally, the data indicate that the aging-out phenomenon does exist for sexual reoffending. This finding has implications for policy related to lifetime registration and suggests the need for some reconsideration of the relative weight that policy decisions should place on short-term versus longer-term management goals--a key component of the AWA.
ENDNOTE
(1.) To view the results of this study, visit http://www.cdjrs.gov/pdffiles1/nij/grants/240099.pdf.
Kristen M Zgoba, Ph.D., is supervisor of research at the New Jersey Department of Corrections. Michael Miner, Ph.D., is a professor at the University of Minnesota Medical School. Raymond Knight, Ph.D., is a professor at Brandeis University. Elizabeth Letourneau, Ph.D., is an associate professor at Johns Hopkins University. Jill Levenson, Ph.D., is an associate professor at Lynn University. David Thornton, Ph.D., is treatment director at Sand Ridge Secure Treatment Center.
 Printer friendly
 Cite/link
 Email
 Feedback
| |
| Title Annotation: | Research Notes |
|---|---|
| Author: | Zgoba, Kristen M.; Miner, Michael; Knight, Raymond; Letourneau, Elizabeth; Levenson, Jill; Thornton, |
| Publication: | Corrections Today |
| Geographic Code: | 1USA |
| Date: | May 1, 2013 |
| Words: | 1696 |
| Previous Article: | NIC's inaugural virtual conference. |
| Next Article: | Day reporting centers in New Jersey: no evidence of reduced recidivism. |
| Topics: | |