Megan’s Law 20 Years Later: An Empirical Analysis and Policy Review

Abstract

This present study examines the sexual and general recidivism rates of 547 convicted sex offenders released before and after the enactment of Megan’s Law in New Jersey. Presenting the longest Megan’s Law evaluation, participants were followed for an average of 15 years after release (range = 10-29 years). Bivariate and multivariate logistic regression equations were estimated to identify covariates significantly associated with both sexual and general recidivism. Group-based trajectories of general recidivism within the 10 years post–prison release were also estimated and compared according to pre–Megan’s Law and post–Megan’s Law release status. No differences in recidivism rates were noted between the cohorts, but differences emerged in the offending trajectories of the high-risk group of offenders within 10 years of release. These results highlight the lack of impact that sex offender registration and notification (SORN) laws have on sexual and general reoffending rates postrelease.

Keywords

sex offenders sexual recidivism general recidivism sex offender registration and notification Megan’s Law

In July 1994, 7-year-old Megan Kanka was raped and murdered by a twice-convicted sex offender. Although local law enforcement was aware of the perpetrator’s criminal history and address, the victim’s family was not. The family’s outrage over this lack of knowledge resonated with the American public and became a rallying cry for the adoption of community notification of sex offender residence. In October 1994, just 3 months after the crime, Megan’s Law was passed in New Jersey. New Jersey’s Megan’s Law required that information on sex offender registrants (specifically those registrants deemed most likely to reoffend) be disseminated to members of the surrounding community (Logan, 2009). President Clinton signed the federal version of Megan’s Law in 1996 as an amendment to the Jacob Wetterling Crimes Against Children and Sexually Violent Offender Registration Act of 1994.

The year 2016 marked the 20th anniversary of the passage of the federal version of Megan’s Law. In the last two decades, states and the federal government have continued to revise and advance sex offender legislation with similar goals of registering sex offenders with local law enforcement and notifying communities of their presence in neighborhoods. Collectively, the laws are known as sex offender registration and notification (SORN) legislation. Research exploring the effectiveness of SORN laws to specifically decrease recidivism in convicted sex offenders supervised under the legislation, or general sexual crimes rates overall, has typically found little to no deterrent effect (e.g., Sandler, Freeman, & Socia, 2008; Schram & Milloy, 1995; Vasquez, Maddan, & Walker, 2008; Zgoba, Veysey, & Dalessandro, 2010; Zgoba, Witt, Dalessandro, & Veysey, 2008), with only a few studies finding evidence of a decrease in sexual offending (Duwe & Donnay, 2008; Freeman, 2012).

The present research examines the recidivism rates of sex offenders released before and after the enactment of Megan’s Law in New Jersey. It may be considered a continuation of the Zgoba et al. (2010) study due to the utilization of the same sample, or a replication of Jennings, Zgoba, and Tewksbury (2012) and Tewksbury and Jennings (2010) due to the effort of replication of the findings and the standardization of the length of follow-up. Unlike Zgoba and colleagues (2010), Jennings and colleagues (2012), or Tewksbury and Jennings (2010), which had follow-up periods of 6, 8, and 5 years, respectively, the current study examines sexual-specific and general reoffending in offenders for an average of 15 years after release. This exploration of recidivism rates utilizing a lengthier follow-up period provides a valuable glimpse into the long-term effects of SORN legislation.

History of SORN Legislation

Sex offender registration refers to the statutory requirement that released sex offenders register their personal information with local law enforcement authorities, whereas notification refers to the process by which the public is informed of such personal information. The goals of sex offender registration were to facilitate police apprehension of sexual recidivists and deter future repeat offending as registrants would be aware that they were under police and public scrutiny at all times. In comparison, the goal of notification was to improve communities’ abilities to protect themselves and their children from identified sex offenders, whether by altering their own behavior or informing the police about suspicious behavior by registrants (Ragusa-Salerno & Zgoba, 2012).

Washington State gained national attention for being the first state to legally mandate community registration guidelines for convicted sex offenders in 1991 (via the Community Protection Act), following highly publicized sexual crimes by repeat offenders. In 1994, sex offender registration was extended to all 50 states after the Wetterling Act was enacted into federal law as Title XVII of the Violent Crime Control and Law Enforcement Act. The Wetterling Act required all states to establish a sex offender registry by the end of 1997, or risk losing federal criminal justice funding.

Megan Kanka’s murder in New Jersey occurred a few months prior to the enactment of the Wetterling Act in September 1994, and has been described as a contributing factor for the inclusion of the community notification component of the Act (Logan, 2009). On October 31, 1994, the New Jersey Legislature voted into law the Registration and Community Notification Law, more commonly known as “Megan’s Law” throughout the state. In the summer of 1995, the New Jersey Attorney General appointed a committee to develop an instrument that would allow county prosecutors to assess sex offender risk in a uniform manner and the Registrant Risk Assessment Scale (RRAS) was developed (Lanterman, Boyle, & Ragusa-Salerno, 2014; Witt, DelRusso, Oppenheim, & Ferguson, 1996). The score from the RRAS places an offender into a designated risk tier (i.e., low [Tier 1], moderate [Tier 2], or high [Tier 3]), which guides the level of community notification. In 2001, a law was enacted that created the New Jersey Sex Offender Internet Registry, whereby the risk tiers produced by the RRAS were used to determine which offenders were included in the Internet Registry. The law has remained unchanged in New Jersey since these initial stages (Lanterman et al., 2014).

In 1996, a federal version of Megan’s Law was enacted into law as an amendment to the Wetterling Act; specifically, Megan’s Law extended the reach of the Wetterling Act’s guidelines of community notification by revising the expectation of registrant nondisclosure to a new requirement of disclosure to the public (Kabat, 1998). This disclosure was obtained via notification through an online sex offender registry or through in-person notifications by local law enforcement. The information disseminated to the public via notification was the same information provided to law enforcement agencies when the sex offender registered upon release from a correctional facility (Terry & Ackerman, 2009). At the inception of the federal Megan’s Law, states adopted provisions to notify the public only about those sex offenders who were deemed to be of highest risk. As states began adopting their own guidelines for notification over the years however, the notification provisions evolved (Vasquez et al., 2008), producing various procedures for the tiering of sex offenders (i.e., using actuarial risk scores) or implementing broad notification procedures without a distinction between high and low risk (Chajewski & Mercado, 2009; Levenson, D’Amora, & Hern, 2007; Zgoba et al., 2016).

The disparities in registration and notification procedures between states, and the belief that some sex offenders were being “lost” in the system as a result (see Logan, 2009, for a more thorough discussion), were eventually noted by federal legislators, and in 2006, Congress enacted the Adam Walsh Child Protection and Safety Act (herein, Adam Walsh Act). The Adam Walsh Act seeks to standardize SORN provisions so as to create a widespread national system for the registration of persons convicted of a sexual crime or a crime against a child (Adam Walsh Act, 42 USC 16,901 “Declaration of Purpose”). According to the Office of Sex Offender Sentencing, Monitoring, Apprehending, Registering and Tracking (SMART), not all states have adopted the legislation, however; at the time of this writing, only 17 states and three U.S. territories have successfully implemented the provisions set forth by the Act (Office of Justice Programs, n.d.). New Jersey, the site of the present study, has not adopted the Adam Walsh Act and has continued to abide by the original provisions set forth in the state’s Megan’s Law legislation. As SORN provisions have changed very little in the state in the past two decades, this makes New Jersey an exemplary study site for a long-term examination of SORN legislation effectiveness in the reduction of sex offender recidivism rates.

SORN Policy Evaluations

Although most researchers have studied only individual states, little evidence of SORN impact has been found for both first-time sexual offending and reoffending in the two decades since the legislation gained momentum in the mid-1990s. Schram and Milloy (1995) published one of the first examinations of SORN legislation effectiveness, utilizing a sample of high-risk sex offenders in Washington State who were and were not subject to community notification under Washington’s Community Protection Act. In comparing 3 years of recidivism data for 180 high-risk sex offenders, the researchers found that sex offenders subject to community notification were rearrested more quickly for new sexual offenses compared with those offenders who were not subject to notification. However, there were no differences in the overall recidivism rates between the two groups. A similar cohort study also found negligible results associated with SORN. Zgoba et al. (2008, 2010) followed a sample of 550 sex offenders released from New Jersey correctional facilities both pre- and post–Megan’s Law implementation for an average period of nearly 7 years. According to the authors, measures of recidivism (i.e., rearrest, reconviction, and reincarceration), community tenure (the amount of time spent in the community before a rearrest), and harm reduction (i.e., decreased sexual reoffending) were not significantly different between the cohorts. They concluded that despite wide community support for Megan’s Law and other SORN legislation, there is little evidence to support claims that such legislation is effective in reducing sexual reoffending.

Adkins, Huff, and Stageberg (2000) received federal funding and examined the effects of registration in Iowa. A preregistration group of sex offenders (n = 201) and a postregistration group (n = 233) were identified for comparison purposes. The results indicated that sex offense registration appeared to have varied effects on recidivism rates over the follow-up period of 4.3 years. Sex-offense recidivism was low at 3% for the registry sample and 3.5% for the preregistry sample. Of those who were convicted of sex offenses, the registry sample had a lower volume of recidivism per person than the preregistry sample; however, the differences in recidivism were not found to be statistically significant.

Researchers Petrosino and Petrosino (1999) approached an evaluation of Megan’s Law from a slightly different angle, determining how many repeat sex crimes could have been avoided if Megan’s Law had been enacted. They examined criminal records of sexual psychopaths (N = 136) and found that 27% of the sample had a prior conviction that met the requirements of the Massachusetts Registry Law before their most recent sex crime (Petrosino & Petrosino, 1999). Of the 36 offenders who would have qualified under the registry, 12 committed a sexual offense against a stranger, leading the researchers to conclude that a small number of cases would have been prevented if the law had been in effect.

Building on the results of Zgoba et al. (2010) and utilizing a similar sample, Tewksbury, Jennings, and Zgoba (2012) created matched samples of sex offenders released pre- and post–Megan’s Law (N = 495). The treatment group was matched to the comparison group on a number of variables theoretically linked to recidivism including age, race, education, family history, marriage, employment, substance use, mental health issues, sexual offense characteristics, and criminal history. Offenders were tracked for up to 8 years postrelease from prison and no significant differences were noted between the groups for both general and sexual reoffending. Like prior quasi-experiment cohort analyses, time-series analyses of sexual offending rates pre- and post–SORN enactment also find limited to no evidence of effectiveness. Vasquez and colleagues (2008) compared rates of forcible rape in 10 states for a minimum of 3 years prior to and following implementation of SORN laws with no specific attention toward recidivism. One of the particular strengths of this study was the inclusion of the multiple states and the increased ability to generalize beyond a single state. The authors concluded overall that SORN had no influence on the number of rapes reported postimplementation; although several states demonstrated a nonsignificant increase in the number of rapes, a significant reduction in rates was only found in three states studied. In a similar manner, Sandler and colleagues (2008) explored differences in sexual offense rates in New York State before and after the implementation of statewide SORN provisions. The researchers collected sexual arrest counts for the years 1986 through 2006, analyzing a sample inclusive of more than 170,000 recorded sexual offense arrests. It was noted that the majority of sex offense arrests (i.e., 95%) were committed by first-time sex offenders. As such, the authors concluded that SORN laws neither reduced sexual offending by first-time offenders, nor have a meaningful impact on the recidivism of convicted rapists, child molesters, and other sexual recidivists.

Although the majority of prior research has found the effectiveness of SORN to be futile in reducing sexual crimes reported to law enforcement, of note are the few studies that have found limited positive outcomes associated with SORN. For example, Duwe and Donnay (2008) compared three groups of offenders: high-risk sex offenders released with SORN provisions, sex offenders released prior to implementation of SORN, and sex offenders released after SORN implementation, but not subject to community notification in Minnesota. Although a comparison of the notification group and the prenotification group suggested a reduction in general and nonsexual recidivism, no reduction was seen for the comparison of the notification and nonnotification group. Furthermore, it was discovered that the notification group had the lowest reoffending rates for all measures of recidivism. The authors, thus, concluded that community notification reduced recidivism for those sex offenders registered under its provisions only.

An analysis of more than 300,000 sex offenses done in 15 states found that whereas public notification did not reduce recidivism rates, registration with law enforcement did (Prescott & Rockoff, 2011; Zgoba & Ragbir, 2016). The authors studied how SORN effected the occurrence of sex offenses, as well as any change in police response to the reported crimes. They discovered that registration reduced the frequency of sex offenses because it provided law enforcement with information on local sex offenders. Their results showed a decrease in crime concentrated among “local” victims (e.g., friends, acquaintances, neighbors), with little evidence of a decrease in crimes against strangers (Prescott & Rockoff, 2011).

Positive results of SORN’s effects on timing were noted by Freeman (2012) in an examination of sexual and nonsexual rearrest rates among convicted sex offenders eligible and ineligible for community notification and supervision (i.e., parole and probation) in New York State. Overall, offenders eligible for community notification were rearrested for nonsexual crimes 47% sooner than offenders not subject to community notification. Furthermore, sex offenders not under supervision but subject to community notification requirements were arrested for a nonsexual offense more quickly than probationers and parolees, regardless of notification status. Sex offenders under supervision who were also subject to notification laws were rearrested for nonsexual offenses sooner than those not subject to notification requirements. Such findings led Freeman (2012) to conclude that notification status, and not supervision, has an effect on sexual offender recidivism.

In sum, a review of the empirical literature examining the consequences of SORN legislation leans heavily toward the assumption that despite widespread public support and politician approval (e.g., Anderson & Sample, 2008; Edwards & Hensley, 2001; LaFond, 2005; Phillips, 1998), SORN laws are limited in their reductive benefits. Although informative, the extant literature lacks analyses that include lengthy follow-up time frames, specifically, follow-up time periods of 10 years or more. Such analyses are needed for a number of reasons. First, it is imperative to study the effects of SORN legislation 20 years after its inception to determine the long-term effects, both intended and unintended, of the legislation. Second, there is contradiction regarding sex offenders’ risk of recidivism postrelease. Although some criminologists suggest that sex offenders continue to commit new offenses for lengthy periods of time after an initial conviction, sometimes even after decades have passed (e.g., Prentky, Lee, Knight, & Cerce, 1997), more recent research has found that the risk of reoffending decreases after lengthy spans of time spent in the community (i.e., Hanson, Harris, Helmus, & Thornton, 2014). Given that SORN legislation often targets those at highest risk of reoffense, a clearer understanding of the impact of the legislation on recidivism rates is needed. This includes not only long-term sexual criminality but also offenses of a general nature to determine whether there is a potential “spill-over effect” of SORN legislation on general offending rates. The present study, thus, attempts to build on the foundation of previous literature by extending the follow-up period of analysis under the pre-/posttest design. Specifically, the present study investigates sexual and general recidivism for an average of 15 years (range = 10-29 years) following release from a New Jersey correctional facility before and after the state’s implementation of Megan’s Law. Three research questions are addressed:

Research Question 1: Did Megan’s Law contribute to a reduction in sex reoffending in those offenders released after the law’s implementation?

Research Question 2: Did Megan’s Law effectuate a reduction in reoffending overall for those offenders released after the law’s implementation?

Research Question 3: Do the general offending trajectories differ for the pre-/post–Megan’s Law release cohorts within 10 years of release from prison?

Method

Participants

This study used a sample of male sex offenders released from New Jersey Department of Corrections facilities before and after the implementation of Megan’s Law in New Jersey. Persons convicted of a sexual offense according to New Jersey law were randomly selected for inclusion in the analysis, totaling 550 cases and spanning 11 years of release. The preimplementation group included offenders released between 1981 and 1994 and consisted of 250 cases; the postimplementation group included offenders released between 1995 and 2000 and consisted of 300 cases. After a review of missing data, three cases were dropped for incomplete information, yielding a final sample of 547 participants.

Data Collection

For each of the 547 cases, extensive demographic, clinical, institutional, criminal history, and crime offense information was collected from each of the participant’s folders. These data provide an opportunity to contrast multiple outcomes (i.e., general recidivism, sexual-specific recidivism, and time to failure before a first reoffense) of offenders convicted and released prior to the passing of Megan’s Law with offenders convicted and released after the legislation enactment, while statistically controlling for theoretically and empirically relevant variables.

Recidivism information was attached to each case using New Jersey Computerized Criminal History reports, which are maintained by the New Jersey State Police. These reports track individual-level arrests that occur within the state of New Jersey. Individual-level arrests that occurred outside of the state of New Jersey (inclusive of all 50 states and U.S. territories) were also reviewed by researchers utilizing Interstate Identification Index reports. Criminal histories were reviewed for new offenses until year-end 2010. The average length of follow-up for tracking recidivism post–prison release was 15.43 years, with a follow-up length ranging from 10 to 29 years.

Sexual recidivism was defined as any arrest that was the result of a sexually related incident. Pursuant to N.J.S.A. 2C:47-1, a participant convicted of any of the following offenses, or an attempt to commit any of the following offenses, was recorded as having a sexual reoffending event: aggravated sexual assault (N.J.S.A. 2C:14-2[a]), sexual assault (N.J.S.A. 2C:14-2[b], (c)), aggravated criminal sexual contact (N.J.S.A. 2C:14-3), kidnapping (N.J.S.A. 2C:13-1[c](2)), endangering the welfare of a child by engaging in sexual conduct (N.J.S.A. 2C:24-4[a]), or endangering the welfare of a child by recording the child in a sexual act (N.J.S.A. 2C:24-4[b](4)). Crimes may also be considered a sexually related incident if the intent was sexual in nature (e.g., kidnapping for the purposes of sexual assault).

General recidivism was defined as an arrest for any type of offense (i.e., an offense of either a sexual or nonsexual nature). These may include violent crimes (e.g., assault, robbery, homicide), property crimes (e.g., arson, theft), drug offenses, or a sexual offense (as defined above).

Analytic Strategy

The analysis proceeds in a number of stages. In the first stage, summary descriptive statistics are reported for the sample. Next, a series of bivariate logistic regression equations are estimated to identify covariates that are significantly associated with sexual recidivism and general recidivism, respectively. In the third stage, the covariates that were identified as being significantly associated with either or both types of recidivism (along with pre- or post–Megan’s Law release status) are entered into multivariate logistic regression equations to determine their independent effects on sexual and general recidivism, controlling for other relevant covariates.

Finally, in an effort to replicate prior research and standardize the length of follow-up (Jennings et al., 2012; Tewksbury & Jennings, 2010; Tewksbury et al., 2012), group-based trajectories of general recidivism for the 10 years post–prison release are estimated; the number, shape, and percentages/proportions of participants classified into the identified trajectory groups are compared according to pre–Megan’s Law and post–Megan’s Law release status.¹ More specifically, relying on widely known and commonly accepted modeling strategies in the criminological literature for estimating group-based trajectories (for review, see Jennings & Reingle, 2012; Nagin, 2005, 2010), the trajectories for each cohort are estimated via an iterative process where the associated Bayesian information criterion (BIC) values and the group-based mean posterior probabilities of group assignment were reviewed for the trajectory solutions of various parametric forms (constant only, linear, quadratic, cubic) in an effort to maximize the BIC and to assess model precision. Also, due to the apparent skewness in the frequency of recidivism over the 10 years of follow-up and the presence of excess zeros, the zero-inflated Poisson (ZIP) functional trajectory form is utilized (Nagin, 2005, 2010).

Results

Table 1 presents the summary descriptive statistics for the entire sample of study (N = 547). As can be seen, approximately half of the sample was White, 35.3% were African American, and 14.6% were Hispanic/Latino. On average, offenders were 38.22 years old at the time of release, and nearly two thirds were raised in a natural two-parent household. Roughly, one in 10 reported having experienced physical abuse as a child, whereas 26.0% reported having been sexually abused as a child. Almost half of the sample had at least a high school diploma or a GED equivalent, and 32.4% were married at the time of their index offense.

Table 1:

Descriptive Statistics (n = 547)

Variables	M/%	SD	Minimum	Maximum
Race of sex offender		—	—	—
White	50.1%
African American	35.3%
Hispanic	14.6%
Age (years)	38.22	11.94	17.00	46.00
Natural two-parent household	65.7%	—	—	—
Physical child abuse	9.9%	—	—	—
Sexual child abuse	26.0%	—	—	—
High school diploma/GED or higher	49.9%	—	—	—
Marital status	32.4%	—	—	—
Mental health problems	22.9%	—	—	—
Drug abuse problems	44.6%	—	—	—
Alcohol abuse problems	47.0%	—	—	—
Type of sex offender		—	—	—
Child molester	79.4%
Rapist	20.3%
Exhibitionist/voyeur	0.4%
Gender of sexual victim		—	—	—
Male-only sexual victims	16.3%
Female-only sexual victims	81.4%
Male and female sexual victims	2.4%
Number of sexual victims	1.44	1.07	1.00	12.00
Type of victim/s		—	—	—
Family member sexual victim	48.1%
Stranger sexual victim	16.1%
Acquaintance sexual victim	33.5%
Significant other sexual victim	2.0%
Other sexual victim	0.2%
Prior nonsexual offense	63.3%	—	—	—
Prior sexual offense	25.0%	—	—	—
Sexual recidivist (rearrest)	12.6%	—	—	—
General recidivist (rearrest)	59.8%	—	—	—
Total number of rearrests	3.15	4.87	0.00	30.00
Average length of follow-up in years	15.43	3.46	12.00	29.00
Post-ML cohort	49.7%	—	—	—

Note. ML = Megan’s Law; GED = general equivalency diploma.

Approximately 22.9% had a mental health problem, and slightly less than half of the sample had drug abuse and/or alcohol problems. The majority of the sample were child molesters, sexually victimized females, and were intrafamilial sex offenders. Almost two out of every three participants had committed an offense (of any kind) prior to their index offense, and 25.0% had committed a sexual offense prior to their index offense. Upon their release from a New Jersey correctional facility, 12.6% of the sample recommitted a sexual offense and 59.8% committed any type of recidivism, including a sexual offense. The number of recidivistic offenses ranged from zero to 30 offenses (M = 3.15 offenses, SD = 4.87 offenses). The average length of follow-up for tracking recidivism post–prison release was 15.43 years (range = 10-29 years).

The bivariate associations between the 25 covariates (including Megan’s Law release status) and sexual and general recidivism are displayed in Table 2. According to the results presented in Column 1, nine of the 25 covariates were significantly associated with sexual recidivism. Specifically, being of African American race/ethnicity (odds ratio [OR] = 1.70, 95% confidence interval [CI] = [1.02, 2.83], p = .04), being a rapist (OR = 1.89, CI = [1.08, 3.31], p = .03), having a greater number of sexual victims (OR = 1.23, CI = [1.02, 1.48], p = .03), having offended against stranger victims (OR = 2.68, CI = [1.51, 4.77], p < .001), and having a prior sexual offense on record (OR = 1.99, CI = [1.18, 3.40], p = .01) all significantly increased the odds that an offender had a sexual reoffense after being released from prison. In contrast, age (OR = 0.96, CI = [0.93, 0.98], p < .001), being married (OR = 0.54, CI = [0.30, 0.99], p = .04), being a child molester (OR = 0.50, CI = [0.29, 0.88], p = .02), and having sexually offended against family members (OR = 0.34, CI = [0.19, 0.60], p < .001) significantly reduced the odds that the sex offender would sexually reoffend postrelease. Megan’s Law status was not significantly associated with the odds of recidivating with a sexual offense upon release from prison.

Table 2:

Bivariate Logistic Regression Results

Variables	Sexual recidivist			General recidivist
Variables	OR	95% CI	p	OR	95% CI	p
African American	1.70	[1.02, 2.83]	.04	3.96	[2.64, 5.94]	<.001
White	0.84	[0.51, 1.40]	.51	0.38	[0.27, 0.55]	<.001
Hispanic	0.42	[0.16, 1.08]	.07	0.67	[0.41, 1.07]	.09
Age	0.96	[0.93, 0.98]	<.001	0.92	[0.91, 1.00]	.05
Natural two-parent household	0.88	[0.52, 1.50]	.64	0.61	[0.42, 0.88]	.01
Physical child abuse	0.53	[0.18, 1.51]	.23	0.98	[0.55, 1.73]	.93
Sexual child abuse	0.54	[0.68, 2.08]	.54	0.56	[0.38, 0.82]	.01
High school diploma/GED or higher	1.27	[0.76, 2.11]	.36	0.83	[0.59, 1.17]	.28
Marital status	0.54	[0.30, 0.99]	.04	0.44	[0.31, 0.64]	<.001
Mental health problems	0.98.	[0.90, 1.06]	.54	1.00	[0.99, 1.02]	.98
Drug abuse problems	0.99	[0.92, 1.08]	.85	3.71	[2.56, 5.38]	<.001
Alcohol abuse problems	1.02	[0.99, 1.05]	.18	1.31	[0.93, 1.85]	.13
Child molester	0.50	[0.29, 0.88]	.02	0.29	[0.17, 0.48]	<.001
Rapist	1.89	[1.08, 3.31]	.03	3.60	[2.16, 6.00]	<.001
Male-only sexual victim	1.10	[0.56, 2.14]	.79	0.54	[0.34, 0.86]	.01
Female-only sexual victim	0.99	[0.51, 1.88]	.97	2.09	[1.35, 3.22]	<.001
Male and female sexual victims	0.57	[0.07, 4.46]	.59	0.29	[0.09, 0.96]	.04
Number of sexual victims	1.23	[1.02, 1.48]	.03	0.85	[0.72, 1.00]	.05
Family member sexual victim	0.34	[0.19, 0.60]	<.001	0.50	[0.35, 0.71]	<.001
Stranger sexual victim	2.68	[1.51, 4.77]	<.001	3.31	[1.89, 5.80]	<.001
Acquaintance sexual victim	1.33	[0.79, 2.23]	.29	1.06	[0.74, 1.52]	.77
Significant other sexual victim	1.56	[0.33, 7.35]	.58	6.91	[0.88, 54.35]	.07
Prior nonsexual offense	1.10	[0.65, 1.87]	.72	4.05	[2.80, 5.85]	<.001
Prior sex offense	1.99	[1.18, 3.40]	.01	1.18	[0.79, 1.76]	.41
Average length of follow-up	1.01	[0.94, 1.08]	.84	1.07	[1.02, 1.12]	.01
Post-ML cohort	0.92	[0.56, 1.53]	.76	0.76	[0.54, 1.07]	.12

Note. OR = odds ratio; CI = confidence interval; ML = Megan’s Law; GED = general equivalency diploma.

Comparatively, 15 of the 25 covariates were significantly associated with general recidivism. As reported in Column 2, being of African American race/ethnicity (OR = 3.96, CI = [2.64, 5.94], p < .001), having drug abuse problems (OR = 3.71, CI = [2.56, 5.38], p < .001), being a rapist (OR = 3.60, CI = [2.16, 6.00], p < .001), having female-only sexual victims (OR = 2.09, CI = [1.35, 3.22], p < .001), having offended against stranger victims (OR = 3.31, CI = [1.89, 5.80], p < .001), having committed a prior nonsexual offense (OR = 4.05, CI = [2.80, 5.85], p < .001), and having a longer follow-up period (OR = 1.07, CI = [1.02, 1.12], p = .01) significantly increased the odds that the sex offender would recidivate with any type of offense after being released from prison. In contrast, being White (OR = 0.38, CI = [0.27, 0.55], p < .001), having been raised in a natural two-parent household (OR = 0.61, CI = [0.42, 0.88], p = .01), having been sexually abused as a child (OR = 0.56, CI = [0.38, 0.82], p = .01), being married (OR = 0.44, CI = [0.31, 0.64], p < .001), being a child molester (OR = 0.29, CI = [0.17, 0.48], p < .001), having male-only sexual victims (OR = 0.54, CI = [0.34, 0.86], p = .01), having male and female sexual victims (OR = 0.29, CI = [0.09, 0.96], p = .04), and having sexually offended against family members (OR = 0.50, CI = [0.35, 0.71], p < .001) significantly reduced the odds that the offender recidivated with a new general offense postrelease. Megan’s Law status was not significantly associated with the odds of recidivating with a general offense upon release from prison.

Table 3 provides the results from the multivariate logistic regressions² relying on the covariates that were identified as being significantly associated with either or both types of recidivism in the bivariate analyses (along with Megan’s Law release status). As illustrated in Column 1, having a greater number of sexual victims (OR = 1.37, CI = [1.12, 1.68], p = .01) and having had a prior sexual offense (OR = 2.29, CI = [1.28, 4.09], p = .01) both significantly increased the odds that an offender would recidivate with a sexual offense controlling for other relevant covariates, whereas age (OR = 0.95, CI = [0.92, 0.98], p < .001) was significantly associated with a reduction in the odds of sexual recidivism controlling for other relevant covariates. With regard to general recidivism, being of African American race/ethnicity (OR = 2.94, CI = [1.53, 5.63], p < .001), having drug abuse problems (OR = 2.58, CI = [1.68, 3.97], p < .001), having committed a prior nonsexual offense (OR = 2.38, CI = [1.56, 3.63], p < .001), and having a greater length of follow-up (OR = 1.14, CI = [1.01, 1.28], p = .04) significantly increased the odds that an offender would have a general offense postrelease controlling for other relevant covariates. In addition, Megan’s Law status was not significantly associated with the odds of recidivating for either a sexual offense or for a general offense upon release from a correctional facility.

Table 3:

Multivariate Logistic Regression Results

Variables	Sexual recidivist			General recidivist
Variables	OR	95% CI	p	OR	95% CI	p
African American	1.47	[0.83, 2.62]	.19	2.94	[1.53, 5.63]	<.001
White	Reference category			1.12	[0.63, 2.01]	.71
Hispanic	Reference category			Reference category
Age	0.95	[0.92, 0.98]	<.001	—
Natural two-parent household	—			0.76	[0.48, 1.18]	.22
Physical child abuse	—			—
Sexual child abuse	—			0.63	[0.39, 0.99]	.05
High school diploma/GED or higher	—			—
Marital status	1.31	[0.65, 2.64]	.45	0.64	[0.41, 1.00]	.05
Mental health problems	—			—
Drug abuse problems	—			2.58	[1.68, 3.97]	<.001
Alcohol abuse problems	—			—
Child molester	0.94	[0.50, 2.09]	.94	0.75	[0.39, 1.44]	.39
Rapist	Reference category			Reference category
Male-only sexual victims	—			2.29	[0.59, 8.83]	.23
Female-only sexual victims	—			3.49	[0.98, 12.40]	.05
Male and female sexual victims	—			Reference category
Number of sexual victims	1.37	[1.12, 1.68]	.01	—
Family member sexual victim	0.53	[0.27, 1.03]	.06	0.72	[0.45, 1.16]	.18
Stranger sexual victim	1.53	[0.75, 3.13]	.24	1.08	[0.52, 2.24]	.84
Acquaintance sexual victim	—			Reference category
Significant other sexual victim	—			Reference category
Prior nonsexual offense	—			2.38	[1.56, 3.63]	<.001
Prior sexual offense	2.29	[1.28, 4.09]	.01	—
Average length of follow-up	—			1.14	[1.01, 1.28]	.04
Post-ML cohort	0.54	[0.69, 2.04]	.54	1.88	[0.85, 4.14]	.12
Model diagnostics
−2 log likelihood	–371.35			–567.66
Nagelkerke R²	.14			.32

Note. Analysis only includes variables that were statistically significant at the bivariate level (see Table 2). OR = odds ratio; CI = confidence interval; GED = general equivalency diploma; ML = Megan’s Law.

Following the modeling process for estimating the trajectories described above and relying on an evaluation of the BIC values and mean posterior probabilities of group assignment as outlined, the most optimal trajectory solution for both Megan’s Law pre- and postcohorts was a three-group, quadratic solution. In addition, the group-based mean posterior probabilities of group assignment were all well above the 0.70 threshold provided by Nagin (2005, 2010), indicating that the models demonstrated precision when assigning individuals to the group-based trajectories that best characterized their recidivism pattern.

As illustrated in Figure 1, the three recidivism trajectories for the pre–Megan’s Law cohort were nonrecidivists, moderate-rate recidivists, and high-rate recidivists.³ The largest trajectory group was the nonrecidivists (57.6% of the cohort), followed by the moderate-rate recidivists (39.9% of the cohort), and the high-rate recidivists (2.5% of the cohort). The moderate-rate recidivists began their trajectory averaging nearly 0.5 offenses per year following their release from prison before gradually declining to an average of less than 0.4 offenses at 10 years postrelease. In contrast, the high-rate recidivists demonstrated a pattern of recidivism that began at a near-zero rate, but their recidivism rate markedly increased over the 10 years of follow-up to an average of greater than 0.6 offenses at 10 years post–prison release.

Figure 1:

Results of the Pre–Megan’s Law Cohort’s General Recidivism Trajectory Analysis (n = 275)

Comparatively, the three recidivism trajectories for the post–Megan’s Law cohort are illustrated in Figure 2.⁴ Similar to the pre–Megan’s Law trajectories, the nonrecidivists represented the most prevalent trajectory group (65.3% of the cohort), followed by the moderate-rate recidivists (32.5% of the cohort), and the high-rate recidivists (2.2% of the cohort). The moderate-rate recidivists averaged nearly 0.4 offenses per year before peaking at an average of 0.6 offenses per year at 5 years post–prison release, and then declining to approximately averaging 0.3 offenses per year at 10 years postrelease. In stark contrast, the high-rate recidivists began recidivating at a high rate immediately upon release from prison (averaging nearly one offense per year) before precipitously declining to zero offenses by 7 to 10 years postrelease.

Figure 2:

Results of the Post–Megan’s Law Cohort’s General Recidivism Trajectory Analysis (n = 272)

Table 4 presents the results from a series of chi-square tests for the differences in percentages/proportions between the trajectory group compositions for the pre- and post–Megan’s Law cohorts. As can be seen, the prevalence of the specific trajectory groups are largely comparable for the two cohorts for the nonrecidivists (57.6% for the pre–Megan’s Law cohort and 65.3% for the post–Megan’s Law cohort), the moderate-rate recidivists (39.9% for the pre–Megan’s Law cohort and 32.5% for the post–Megan’s Law cohort), and the high-rate recidivists (2.5% for the pre–Megan’s Law cohort and 2.2% for the post–Megan’s Law cohort). As such, not surprisingly, none of these compositional differences in the percentages/proportions of individuals assigned to the nonrecidivists, moderate-rate recidivists, or to the high-rate recidivists trajectories was determined to be significantly different when comparing the compositions for the pre–Megan’s Law cohort with those of the post–Megan’s Law cohort. This lack of significant compositional differences suggests that Megan’s Law had little to no effect on the number of trajectory groups or on the percentage of nonrecidivists, moderate-rate recidivists, or high-rate recidivists. These null effects notwithstanding, it is important to note that although there were no significant differences in terms of the prevalence/percentage of high-rate offenders for the pre- and post–Megan’s Law cohorts, these two groups were distinctly different, in that the pre–Megan’s Law high-rate offenders began recidivating slowly and then markedly increased over the 10 years post–prison release, whereas the post–Megan’s Law cohort began recidivating at high rates and then markedly declined to zero for years 7 to 10 post–prison release.

Table 4:

Group Membership Comparisons for Pre- and Post–Megan’s Law Samples

Sample distribution	Nonrecidivist trajectory	Moderate-rate recidivist trajectory	High-rate recidivist trajectory
Pre-ML sample	57.6%	39.9%	2.5%
Post-ML sample	65.3%	32.5%	2.2%
Difference in percentages/proportions	7.7%	7.4%	0.3%
Chi-square test of difference of percentages/proportions	2.09 (p = .15)	1.15 (p = .28)	0.001 (p = .97)

Note. ML = Megan’s Law.

Discussion

Study Findings

The present study was completed to explore the long-term effects, both intended and unintended, of the enactment of SORN legislation nearly 20 years after implementation. Using a sample of 547 sex offenders released from New Jersey correctional facilities prior to and after the enactment of the state’s Megan’s Law, the present study examined three specific research questions: (1) Did Megan’s Law contribute to a reduction in sexual reoffending in those offenders released after the law’s implementation? (2) Did Megan’s Law effectuate a reduction in reoffending overall for those offenders released after the law’s implementation? and (3) Do the general offending trajectories differ for the pre-/post–Megan’s Law release cohorts within 10 years of release from prison? Similar to previous studies (e.g., Sandler et al., 2008; Schram & Milloy, 1995; Vasquez et al., 2008; Zgoba et al., 2010; Zgoba et al., 2008), the results of the analyses indicate that SORN legislation has not had a significant impact on sexual or general reoffending rates for sex offenders overall in the past two decades. With that said, there is evidence that SORN legislation may slightly affect reoffenses generally for those offenders who are most likely to reoffend—and at high rates—postrelease.

There were three factors that were found to be related to increased odds of sexual recidivism postrelease after controlling for a number of other covariates in the multivariate analysis: sex offender age, the number of victims of the index offense, and having a prior history of sexual offending. Such findings are supportive of the established literature. It is well known that sexual recidivism declines with age (e.g., Fazel, Sjöstedt, Långström, & Grann, 2006; Hanson, 2002; Hanson & Bussière, 1998), and that having a sexual criminal history increases the likelihood of sexual reoffending (Hanson & Bussière, 1998; Harris & Hanson, 2004). However, Megan’s Law release status was not a predictor of sexual recidivism, indicating that sex offenders who were released prior to the enactment of Megan’s Law in New Jersey (and, thus, not subject to many of the various provisions of SORN) did not have higher rates of sexual offending in the nearly two decades of follow-up when compared with those sex offenders who were released after the enactment of Megan’s Law.

The above correlates are also problematic from a legislative and statistical perspective. The federally enacted Adam Walsh Act seeks to broaden the registrant eligibility in terms of longer registration and online notification requirements, which runs counterintuitive to the result that sex offenders age out of crime similar to non–sex offenders. In general, risk of sexual reoffense decreases with advancing age, suggesting that the longer registration periods in the Adam Walsh Act may be inefficient. As sex offenders age, they pose less of a threat to society and a requirement of lifetime registration and online notification may make it difficult for the public to discern true risk (Zgoba et al., 2016). In addition, a basic tenet of both the federal Megan’s Law and the Adam Walsh Act is that identification and notification of known sex offenders will deter reoffending. The current sample of sex offenders released before and after Megan’s Law that was enacted in New Jersey highlights that 75% of the sample had no previously detected sexual offense. This minimizes the strength of one of the more basic principles of sexual offense legislation.

Furthermore, several risk factors such as having a history of sexual child abuse victimization, being a drug abuser, and having a prior nonsexual offense on record were all found to be predictors of general reoffending. Again, Megan’s Law release status was not predictive of general offending postrelease. This provides evidence that SORN legislation is ineffective at reducing not only sexual-specific recidivism in the long term but also general reoffending overall.

Of note are the results of the group-based trajectories of general recidivism within 10 years post–prison release. Although Megan’s Law had little to no effect on the number of trajectory groups that were established or on the percentage of nonrecidivists, moderate recidivists, or high-rate recidivists in the sample, there were differences in the pre- and postcohorts in terms of each group’s reoffending trends. Those offenders who were deemed to be at highest risk of recidivism and released before Megan’s Law began a slow recidivism pattern, which increased markedly by the 10-year postrelease time point, whereas those high-rate offenders released after Megan’s Law began recidivating quickly postrelease and declined to reoffense rates of near zero by the 7-year postrelease mark. The high rate of offending for the post–Megan’s Law cohort is presumed to be a result of the high rate of detection of criminality soon after release from prison given the group’s registry presence and increased monitoring and surveillance under SORN provisions. The increased rates are also a likely artifact of the myriad collateral consequences of SORN legislation (detailed extensively in Tewksbury, 2005).

The results for the pre–Megan’s Law group bear more scrutiny, however. One considered explanation for these findings was the timing of the implementation of Megan’s Law in New Jersey in 1994, and its retroactivity for certain offenders in the pre–Megan’s Law cohort. Although the sex offender laws in New Jersey have changed very little since the implementation of Megan’s Law, an offender sentenced under the New Jersey Sex Offender Act (originally enacted in 1950 [c.207 2A:192])—even if convicted and released prior to the passage of New Jersey’s Megan’s Law—is required to abide by its guidelines.⁵ This includes registering with local law enforcement and advising law enforcement to changes of address, among other provisions. Thus, any sex offender included within the pre–Megan’s Law cohort who was originally sentenced under the Sex Offender Act was forced to comply with the Megan’s Law provisions that went into effect in 1994. A more in-depth look at the pre–Megan’s Law sample shows that 92.3% were released between 1990 and 1994. The results of the trajectory models show an increase in rearrest recidivism for high-rate offenders beginning 5 years postrelease and beyond, which overlaps with the timeline of implementation of Megan’s Law for the majority of the precohort. This could lead one to assume that the increased rate of offending beginning 5 years after release for the high-rate offenders is an artifact of increased monitoring brought on by Megan’s Law provisions. However, in examining the pre–Megan’s Law group trajectories further, it was discovered that the majority of the high-rate group (i.e., 77.2%) were not released from the Adult Diagnostic and Treatment Center (ADTC), and thus, not sentenced under the Sex Offender Act or eligible for Megan’s Law retroactivity. For this reason, we cannot conclude that the increase in rearrests years after release for this group is related to the timing of SORN legislation and increased surveillance of those offenders who were at highest risk of reoffending in the pregroup.

Rather, high-rate offenders who were released pre–Megan’s Law may have been able to markedly increase their offending frequency over time, and were perhaps able to commit offenses soon following their release that did not come to the attention of law enforcement because there was no registry or increased monitoring and surveillance in place for the majority of the time that they were on the streets. These findings for the group-based trajectories are interesting, given that the proportions of offenders released pre-/post–Megan’s Law who recidivated with a sexual offense in the entire follow-up were not found to be significantly different across the cohorts (pre–Megan’s Law = 13% vs. post–Megan’s Law = 12.2%; χ² = 0.09, p = .76). It can, thus, be concluded that although SORN may not affect the rates of reoffending for sexual and nonsexual recidivism in the 10 years postrelease, it may affect the progression of offending rates. It is recommended that future research explore recidivism in this population of high-rate offenders to further disentangle the effects of lack of supervision and increased follow-up time at risk on rearrest events. Specifically, it is proposed that offense type be analyzed to determine whether these unsupervised sex offenders are committing crimes of a sexual or nonsexual nature, and whether recidivism events continue beyond the 10-year follow-up mark. Unfortunately, such an exploration was outside the scope of the present research and could not be completed.

Finally, it is important to discuss that although this article focuses on an assessment of Megan’s Law ability (or inability) to deter subsequent sexual crimes, other limitations or drawbacks have been highlighted in the literature. First, various states have recognized that implementation and compliance with increased standards under new evolutions of Megan’s Law would cost millions of dollars over expectations; in particular, the State of Virginia has noted that the first year of compliance with SORNA would cost US$12 million (Virginia Department of Planning and Budget, 2008). Second, as comprehensively highlighted in Connor (2016), a large majority of the public does not actively utilize the information that is disseminated through sex offender registries. This presents both a practical and financial dilemma because although states are paying millions of dollars, the public is not utilizing a tool that could broaden their awareness. When the public does utilize the registry, it often promotes excessive precautionary or irrational behavior among users, and it has been well documented that there is a large number of errors in the offender information that is posted (Connor, 2016; Levenson & Cotter, 2005). For example, Levenson and Cotter (2005) found that out of a sample of 183 registrants, half had either erroneous or missing information listed. Finally, sex offender registries have also promoted a false sense of security among the public, in both their belief that the government is protecting them and their perception of who the perpetrators are. The registry has done little to sway the conversation in the direction of who is responsible for sexual abuse, as the public continues to believe that strangers must be feared the most.

Limitations and Directions for Future Research

The present findings must be interpreted with caution. Recidivism is defined by offenses that are known to the authorities and, thus, appear on one’s criminal history. It has been well established that sexual offenses are underreported to police and that official records may underestimate the number of sexual offenses that are actually occurring (Furby, Weinrott, & Blackshaw, 1989; Grubin, 1998). As noted, the rates of sexual recidivism in the present sample were quite low (pre–Megan’s Law cohort = 13% vs. post–Megan’s Law cohort = 12.2%) and may be an artifact of the unreporting of events by victims. Unfortunately, the low rate of recidivism in the present sample made the addition of group-based trajectory models predicting sexual reoffending to be untenable. In a similar vein, only 25% of the sample had a prior sexual offense on record at the time of release, whereas 75% were first-time convicted sex offenders. Given that having a more extensive criminal history puts a sex offender at increased risk of reoffending (Hanson & Bussière, 1998; Harris & Hanson, 2004), the present sample lacks offenders who are at an overall higher risk of recidivating. It is advised that future studies replicate the present analyses utilizing a larger heterogeneous sample to adequate power analyses examining the trajectories of sexual-specific recidivism within a 10-year time frame.

Although the group-based trajectory analyses divide offenders by offending rate, the present study does not include any analyses in which offenders are categorized according to Megan’s Law tier designation. As noted, sex offenders in New Jersey are given a risk tier designation (i.e., Tier 1, Tier 2, and Tier 3) after release that guides community notification provisions for the registrant and determines placement on the New Jersey Sex Offender Internet Registry (specifically, inclusion for Tier 2 and Tier 3 registrants only). Unfortunately, tier designation could not be collected for the sample at this time. Thus, it is unknown how tier designation—above and beyond the simple pre-/post–Megan’s Law designation—affects sexual and general recidivism in this sample, particularly for those in the post–Megan’s Law release cohort.

It is also important to note that this study’s findings are based on an individual state, and although the majority of studies that evaluate the effectiveness of SORN legislation are also completed within one state, the study generalizability may be reduced. As the birthplace of Megan’s Law, New Jersey’s law has undergone few modifications since its inception (whereas other states have made modifications) and exists in a relatively original state. Due to the variances in Megan’s Law across the country (as evidenced by the enactment of the Adam Walsh Act to standardize SORN), these findings may be more specific to New Jersey as the offenders have abided by New Jersey provisions. Despite the majority of research confirming these results, more research continues to be necessary to determine whether this study is generalizable to the nation as a whole.

Finally, one must approach these findings with the recognition that proponents of registries believe they hold value and purpose separate from crime reduction (Bierie, 2016). In particular, research indicates that registrants are at an increased risk of committing a sex crime compared with the general public and that registration has aided law enforcement investigations in the capture or prevention of sexual crimes (Bierie, 2016). In addition, the advent of sexual offense laws has prompted a national conversation on sexual abuse and has increased parental awareness of child safety. Researchers should recognize these potential values to have a balanced debate on efficacy.

Conclusion

There is much evidence to suggest that SORN legislation for offenders convicted of sexual crimes does not have a demonstrable effect on future offending. The present study has added to this research by illustrating that such effects are not limited to the short term, but also the long term, as no differences in recidivism rates were noted in cohorts of offenders released before and after the implementation of Megan’s Law in New Jersey. There was, however, evidence that being subjected to SORN provisions has an impact on the trajectory of offending within 10 years of release from a correctional facility, in that high-rate offenders released under SORN tend to commit crimes immediately upon release and then decrease offending to a near-zero rate after a term of approximately 7 years. Thus, although SORN legislation may not have an effect on recidivism rates, it may have a limited effect on the timing of offenses committed after release from prison.

Footnotes

Notes

Kristen M. Zgoba, PhD, received her doctoral degree from the School of Criminal Justice at Rutgers, the State University of New Jersey in 2004. She is an assistant professor at the University of Central Florida where she teaches graduate statistics. She is the recipient of multiple grants, awards, and a Fulbright Scholarship to conduct research in the United Kingdom. She has authored more than 40 peer-reviewed articles.

Wesley G. Jennings, PhD, is professor and doctoral program coordinator in the School of Criminal Justice at Texas State University. In addition, he has a courtesy appointment in the Department of Health Outcomes and Biomedical Informatics and is a faculty affiliate of the Institute for Child Health Policy in the College of Medicine at the University of Florida. He received his doctorate degree in criminology from the University of Florida. He has more than 225 publications, his h-index is 41 (i-index of 121), and he has nearly 6,000 citations to his published work.

Laura M. Salerno, PhD, is a research scientist with the New Jersey Department of Corrections, Research and Evaluation Unit. Her research interests lie primarily in the fields of corrections, reentry, and public policy. She has published works in Criminology and Public Policy, Journal of Research in Crime and Delinquency, and Journal of Interpersonal Violence, among others.

References

Adkins

Huff

Stageberg

(2000). The Iowa sex offender registry and recidivism. Des Moines: Iowa Department of Human Rights.

Anderson

A. L.

Sample

L. L.

(2008). Public awareness and action resulting from sex offender community notification laws. Criminal Justice Policy Review, 19, 371-396.

Bierie

D. M.

(2016). The utility of sex offender registration: A research note. Journal of Sexual Aggression, 22, 263-273.

Chajewski

Mercado

C. C.

(2009). An evaluation of sex offender residency restriction functioning in town, county, and city-wide jurisdictions. Criminal Justice Policy Review, 20, 44-61.

Connor

D. P.

(2016). Rare acts, sympathetic victims, and the emergence of laws to regulate sex offenders. In Fradella

H. F.

Sumner

J. M.

(Eds.), Sex, sexuality, law, and (in)justice (pp. 340-366). New York, NY: Routledge.

Duwe

Donnay

(2008). Impact of Megan’s Law on sex offender recidivism: The Minnesota experience. Criminology, 46, 411-446.

Edwards

Hensley

(2001). Restructuring sex offender sentencing: A therapeutic jurisprudence approach to the criminal justice process. International Journal of Offender Therapy and Comparative Criminology, 45, 646-662.

Fazel

Sjöstedt

Långström

Grann

(2006). Risk factors for criminal recidivism in older sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 18, 159-167.

Freeman

N. J.

(2012). The public safety impact of community notification laws: Rearrest of convicted sex offenders. Crime & Delinquency, 58, 539-564.

10.

Furby

Weinrott

M. R.

Blackshaw

(1989). Sex offender recidivism: A review. Psychological Bulletin, 105, 3-30.

11.

Grubin

(1998). Sex offending against children: Understanding the risk (Vol. 99). London, England: Home Office, Policing and Reducing Crime Unit, Research, Development and Statistics Directorate.

12.

Hanson

R. K.

(2002). Recidivism and age: Follow-up data from 4,673 sexual offenders. Journal of Interpersonal Violence, 17, 1046-1062.

13.

Hanson

R. K.

Bussière

M. T.

(1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.

14.

Hanson

R. K.

Harris

A. J.

Helmus

Thornton

(2014). High-risk sex offenders may not be high risk forever. Journal of Interpersonal Violence, 29, 2792-2813.

15.

Harris

A. J. R.

Hanson

R. K.

(2004). Sex offender recidivism: A simple question (Vol. 3). Ottawa, Ontario: Public Safety and Emergency Preparedness Canada.

16.

Jennings

W. G.

Reingle

(2012). On the number and shape of developmental/life-course violence, aggression, and delinquency trajectories: A state-of-the-art review. Journal of Criminal Justice, 40, 472-489.

17.

Jennings

W. G.

Zgoba

Tewksbury

(2012). A comparative longitudinal analysis of recidivism trajectories and collateral consequences for sex and non-sex offenders released since the implementation of sex offender registration and community notification. Journal of Crime and Justice, 35, 356-364.

18.

Kabat

A. R.

(1998). Scarlet letter sex offender databases and community notification: Sacrificing personal privacy for a symbol’s sake. The American Criminal Law Review, 35, 333-370.

19.

LaFond

J. Q.

(2005). Preventing sexual violence: How society should cope with sex offenders. Washington, DC: American Psychological Association.

20.

Lanterman

J. L.

Boyle

D. J.

Ragusa-Salerno

L. M.

(2014). Sex offender risk assessment, sources of variation, and the implications of misuse. Criminal Justice and Behavior, 41, 822-843.

21.

Levenson

J. S.

Cotter

L. P.

(2005). The effect of Megan’s Law on sex offender reintegration. Journal of Contemporary Criminal Justice, 21, 49-66.

22.

Levenson

J. S.

D’Amora

D. A.

Hern

A. L.

(2007). Megan’s Law and its impact on community re-entry for sex offenders. Behavioral Sciences & the Law, 25, 587-602.

23.

Logan

W. A.

(2009). Knowledge as power: Criminal registration and community notification laws in America. Stanford, CA: Stanford University Press.

24.

Nagin

D. S.

(2005). Group-based modeling of development. Cambridge, MA: Harvard University Press.

25.

Nagin

D. S.

(2010). Group-based trajectory modeling: An overview. In Piquero

A. R.

Weisburd

(Eds.), Handbook of quantitative criminology (pp. 53-67). New York, NY: Springer.

26.

Nagin

D. S.

Tremblay

R. E.

(2005). Developmental trajectory groups: Fact or a useful statistical fiction. Criminology, 43, 873-904.

27.

Office of Justice Programs. (n.d.). SORNA implementation status. Washington, DC: Author. Retrieved from https://www.smart.gov/sorna-map.htm

28.

Petrosino

A. J.

Petrosino

(1999). The public safety potential of Megan’s Law in Massachusetts: An assessment from a sample of criminal sexual psychopaths. Crime & Delinquency, 45, 140-158.

29.

Phillips

D. M.

(1998). Community notification as viewed by Washington’s citizens. Olympia: Washington State Institute for Public Policy.

30.

Prentky

R. A.

Lee

A. F.

Knight

R. A.

Cerce

(1997). Recidivism rates among child molesters and rapists: A methodological analysis. Law and Human Behavior, 21, 635-659.

31.

Prescott

J. J.

Rockoff

J. E.

(2011). Do sex offender registration and notification laws affect criminal behavior? Journal of Law & Economics, 54, 161-206.

32.

Ragusa-Salerno

L. M.

Zgoba

K. M.

(2012). Taking stock of 20 years of sex offender laws and research: An examination of whether sex offender legislation has helped or hindered our efforts. Journal of Crime and Justice, 35, 335-355.

33.

Sampson

R. J.

Laub

J. H.

(2005). Seductions of method: Rejoinder to Nagin and Tremblay’s “developmental trajectory groups: Fact or fiction.” Criminology, 43, 905-914.

34.

Sandler

J. C.

Freeman

N. J.

Socia

K. M.

(2008). Does a watched pot boil? A time-series analysis of New York State’s sex offender registration and notification law. Psychology, Public Policy, and Law, 14, 284-302.

35.

Schram

D. D.

Milloy

C. D.

(1995). Community notification: A study of offender characteristics and recidivism. Olympia: Washington Institute of Public Policy.

36.

Terry

A. K.

Ackerman

A. R.

(2009). A brief history of major sex offender laws. In Wright

R. G.

(Ed.), Sex offender laws: Failed policies, new directions (2nd ed., pp. 65-98). New York, NY: Springer.

37.

Tewksbury

(2005). Collateral consequences of sex offender registration. Journal of Contemporary Criminal Justice, 21, 67-81.

38.

Tewksbury

Jennings

W. G.

(2010). Assessing the impact of sex offender registration and community notification on sex offending trajectories. Criminal Justice and Behavior, 37, 570-582.

39.

Tewksbury

Jennings

W. G.

Zgoba

(2012). A longitudinal examination of sex offender recidivism prior to and following the implementation of SORN. Behavioral Sciences & the Law, 30, 308-328.

40.

Vasquez

B. E.

Maddan

Walker

J. T.

(2008). The influence of sex offender registration and notification laws in the United States. Crime & Delinquency, 54, 175-192.

41.

Virginia Department of Planning and Budget. (2008). Fiscal impact statement. Richmond: Author.

42.

Witt

P. H.

DelRusso

Oppenheim

Ferguson

(1996). Sex offender assessment and the law. Journal of Psychiatry & Law, 24, 343-377.

43.

Zgoba

Miner

Letourneau

Levenson

Knight

Thornton

(2016). A multi-state recidivism study using Static-99 and Static-2002 risk scores and tier guidelines from the Adam Walsh Act. Sexual Abuse: A Journal of Research and Treatment, 28, 722-740.

44.

Zgoba

Ragbir

(2016). Sex Offender Registration and Notification Act (SORNA). In Calkins

Jeglic

(Eds.), Sexual violence: Evidence based policy and prevention (pp. 33-49). New York, NY: Springer.

45.

Zgoba

Veysey

B. M.

Dalessandro

(2010). An analysis of the effectiveness of community notification and registration: Do the best intentions predict the best practices? Justice Quarterly, 27, 667-691.

46.

Zgoba

Witt

Dalessandro

Veysey

B. M.

(2008, December). Megan’s Law: Assessing the practical and monetary efficacy (Document No. 225370). Washington, DC: U.S. Department of Justice.