Review of Risk Assessment Instruments for Juvenile Sex Offenders

Abstract

Risk assessment is considered to be a key element in the prevention of recidivism among juvenile sex offenders (JSOs), often by imposing long-term consequences based on that assessment. The authors reviewed the literature on the predictive accuracy of six well-known risk assessment instruments used to appraise risk among JSOs: the Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II), Juvenile Sexual Offence Recidivism Risk Assessment Tool-II (J-SORRAT-II), Estimate of Risk of Adolescent Sexual Offence Recidivism (ERASOR), Juvenile Risk Assessment Scale (JRAS), Structured Assessment of Violent Risk in Youth (SAVRY), and Hare Psychopathy Checklist:Youth Version (PCL:YV). Through a systematic search, 19 studies were reviewed. Studies showed differences in the predictive accuracies for general, violent, and sexual recidivism, and none of the instruments showed unequivocal positive results in predicting future offending. Not unexpectedly, the accuracy of the SAVRY and PCL:YV appeared to be weaker for sexual recidivism compared with specialized tools such as the J-SOAP-II or the ERASOR. Because of the rapid development of juveniles, it is questionable to impose long-term restrictions based on a risk assessment only. New challenges in improving risk assessment are discussed.

Keywords

risk assessment juvenile sex offenders predictive validity J-SOAP-II PCL:YV J-SORRAT-II ERASOR JRAS SAVRY

Introduction

Juvenile sex offenders (JSOs) are regularly assessed by clinicians in the juvenile justice system on their risk of reoffending. Risk assessment is considered to be a key element in the prevention of recidivism, often by imposing long-term consequences on JSOs based on that assessment. To make decisions about sanctions, referrals, or rehabilitation, actuarial and structured risk assessment instruments are superior to unstructured clinical judgments (Hanson, 2000; Hanson & Morton-Bourgon, 2009). For the adult sex offender population, there has been progress in developing reliable actuarial risk assessment instruments (Epperson, Kaul, & Hesselton, 1999; Hanson & Thornton, 1999). However, in the last decades, only a few risk assessment instruments have been developed specifically for JSOs, and the literature on these instruments is limited. The purpose of this review is to give an overview of the literature on the predictive accuracy of risk assessment instruments among JSOs.

Sexual Offending

Until 20 years ago, studies on sexual offending were mainly focused on adults. Sexual behavior of juveniles was seen as experimental or developmental curiosity (Veneziano & Veneziano, 2002). However, juveniles are responsible for a considerable percentage of sexual offences. In 2008, 15% of the arrests for forcible rape in the United States were on account of juveniles aged below 18 years (Federal Bureau of Investigation, 2008). These arrests included assaults and attempts of rape by force or threat of force but did not include statutory rape (without force) or other sex offences. Alongside the lack of indictments and not having enough evidence to proceed with an arrest, the 15% arrest rate is likely to be an underrepresentation of the actual juvenile sex offending rate. Several attempts were made to classify JSOs. However, these attempts were mostly intuitively derived and have not been empirically validated (Hendriks, 2006; Veneziano & Veneziano, 2002). Most researchers are tempted to adapt models and methods for categorization and prediction that have proven to be reliable with adult offenders to juvenile offenders (Caldwell, 2002; Worling, 2001). For example, similar to the adult sex offender population, juveniles who offend against children (child molesters) are compared with those who offend against peers or adults (rapists). However, few meaningful differences have been found between juvenile child molesters and juvenile rapists, and researchers question the categorization based on victim age (Hagan & Cho, 1996; Hsu & Starzynski, 1990; Worling, 2001). Regarding personality, however, four distinct subtypes of JSOs are suggested based on cluster analysis of Minnesota Multiphasic Personality Inventory profiles (see Smith, Monastersky, & Deisher, 1987; Worling, 2001).

Reoffending among JSOs is not limited to sex offences only. Caldwell (2010) examined recidivism studies that included a total of 11,219 JSOs. Within a mean follow-up period of 5 years, he found a mean base rate of 7.08% for sexual reoffending and a much higher mean base rate for general reoffending (43.4%). In a 20-year prospective follow-up study, the base rates for sexual (9%), nonsexual violent (22%), nonviolent (28%), and general reoffending (38%) were significantly lower for juveniles who participated in specialized treatment, relative to a comparison group (21%, 39%, 52% and 57% base rates, respectively; Worling, Litteljohn, & Bookalam, 2010). Prentky et al. (2010) found that there is also a difference within the adolescent population, with notably higher sexual recidivism rates for preadolescent boys (25%-28%) than for adolescent boys (14%-16%). However, the ages of the preadolescent boys ranged from 3 to 11, and their “offences” were coded by evidence of highly sexualized, age inappropriate behavior. Specialized treatment is focused on reducing risk for reoffending. The focus of the empirical literature has largely been on the identification of individual risk predictors, and when used in isolation, these risk predictors are typically weak (Prentky, Pimental, Cavanaugh, & Righthand, 2009). In addition, risk factors for JSOs are often extrapolated from the adult literature, not empirically validating them for the JSO population. However, a number of risk factors have been empirically linked to sexual reoffending among JSOs. Seto and Lalumière (2010) conducted a meta-analysis on studies comparing male JSOs with male juvenile nonsex offenders on general delinquency risk factors or factors identified in special explanations of juvenile sex offending. Special factors are different from the factors that explain the offenses of other juvenile delinquents. They found many similarities between general delinquency risk factors for offending in both groups (e.g., measures of personality traits), but these factors alone were not sufficient to understand why a juvenile commits a sexual offence rather than a nonsexual offence. Special explanations suggest a role for sexual abuse history, exposure to sexual violence, other abuse or neglect, social isolation, early exposure to sex or pornography, anxiety, and low self-esteem. Especially notable was that JSOs reported more atypical sexual fantasies, behaviors, or interests, or were more often diagnosed with a paraphilia. Seto and Lalumière suggested that atypical sexual fantasies, behaviors, or interests should be given more prominence in theories of juvenile sex offending as they might be regarded as risk factors.

Risk assessment is the examination of possible risks for reoffending based on factors that are empirically related to reoffending. Risk assessment instruments differ in their combination of risk factors, resulting in different predictive accuracies. The purpose of this review is to give an up-to-date overview of the literature on the predictive accuracy of risk assessment instruments among JSOs. We will examine the six instruments that have generated the most research—the Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II), Juvenile Sexual Offence Recidivism Risk Assessment Tool-II (J-SORRAT-II), Estimate of Risk of Adolescent Sexual Offence Recidivism (ERASOR), Juvenile Risk Assessment Scale (JRAS), Structured Assessment of Violent Risk in Youth (SAVRY), and Hare Psychopathy Checklist: Youth Version (PCL:YV)—and give an overview of the predictive validities for sexual, violent (nonsexual), and general reoffending. Finally, we will discuss the characteristics of the instruments that might be associated with higher levels of predictive accuracy as well as new methods for risk assessment. For a summary of the scales and contents of the instruments, see Table 1.

Table 1.

Risk Assessment Instruments for Juvenile Sex Offenders: Scales and Contents

Instrument	J-SOAP-II	J-SORRAT-II	ERASOR	JRAS	PCL:YV	SAVRY
Scales	Sexual drive/sexual preoccupation Impulsive/antisocial behavior Intervention Community stability/adjustment	Offending history and characteristics Abuse history School history Treatment history	Sexual interests, attitudes, and behaviors Historical sexual assaults Psychosocial functioning Family/environmental functioning Treatment	Antisocial Sexual deviance Adult rapist	Interpersonal Affective Antisocial Behavioral	Historical Social contextual Individual/clinical Protective
Content scales	Sexual offence history, sexual drive, and preoccupation responsibility Past serious school and behavior problems Treatment motivation and taking Current adjustment to community	Number of adjudications and number of victims Past victimization experiences Special education and school discipline Completion of sex offender treatment	Deviant preoccupations and attitudes, unwillingness to alter deviancy Number of victims, type of victim, and male victim Antisocial orientation, lack of relationships, and poor self-regulation No development of prevention strategies and incomplete sexual offence–specific treatment	History of antisocial acts, substance abuse, response to treatment, sex offender–specific therapy, residential support, and employment/educational stability Degree of contact, number of offenses/victims, duration of offensive behavior, and victim gender Degree of force, age of victim, and victim selection	Impression management, grandiose sense of self-worth, lying, and manipulation Callousness, lack of remorse, and shallow affect Stimulation seeking, impulsivity, irresponsibility, parasitic orientation, and lack of goals Poor anger control, early behavior problems, and criminal behavior	Early initiation and history of violence, and criminality of parents Peer rejection and delinquency, and limited education skills of parents Substance abuse and anger management Prosocial involvement, strong social support, strong attachment and bonds to positive figures, and strong commitment to school

Note: J-SOAP-II = Juvenile Sex Offender Assessment Protocol-II; J-SORRAT-II = Juvenile Sexual Offence Recidivism Risk Assessment Tool-II; ERASOR = Estimate of Risk of Adolescent Sexual Offence Recidivism; JRAS = Juvenile Risk Assessment Scale; PCL:YV = Hare Psychopathy Checklist:Youth Version; SAVRY = Structured Assessment of Violent Risk in Youth.

Risk Assessment Instruments for JSOs

Risk assessment has developed from unstructured clinical or professional judgments to actuarial methods and finally to structured professional judgments (SPJs). Actuarial prediction involves a strictly evidence-based selection of risk factors empirically related to criminal behavior. The SPJ approach provides guidelines for assessing risk in a systematic and structured manner, based on empirically supported risk factors, while permitting professional flexibility to consider unique characteristics of individual cases. SPJ is a model of decision making that underlies many of the successful risk assessment measures (Douglas, Ogloff, & Hart, 2003).

One of the most commonly used measures in the United Sates with JSOs is the J-SOAP-II (Prentky & Righthand, 2003). The J-SOAP-II is an empirically informed guide for the systematic review and assessment of a uniform set of risk factors that has been associated with sexual and violent offending. It is designed to be used for boys in the age range of 12 to 18 years who have been adjudicated for sexual offences as well as nonadjudicated youths with a history of sexually coercive behavior. The J-SOAP-II results in a total score. As at this point there are no cutoff scores available for categories of risk, scores from J-SOAP-II should not be used in isolation when assessing risk. Although not developed for that purpose, the 12 dynamic items of the J-SOAP-II might be used for assessing treatment needs and progress because of their changeability during the treatment process. This instrument is mandatory in two states in the United States and is often used to impose long-term consequences on JSOs. According to Prentky and Righthand (2003), the interrater reliability for all items is good to excellent, ranging from .75 to .91, with an average of .83. McCoy (2007), however, found interrater reliabilities ranging from .39 to .99. The internal consistency alphas ranged from .68 to .85 (McCoy, 2007; Prentky & Righthand, 2003).

Another widely used instrument is the J-SORRAT-II (Epperson, Ralston, Fowers, & DeWitt, 2005), an actuarial risk assessment instrument for male juveniles between 12 and 18 years who have offended sexually. The interrater reliability of .89 falls in the excellent range (Viljoen et al., 2008). Ralston (2008) found an interrater reliability of .96 and an internal consistency of .99.

Worling and Curwen’s (2000) ERASOR was modeled after the Historical, Clinical, Risk Management-20 and Sexual Violence Risk-20 (adult risk assessment instruments; Boer, Hart, Kropp, & Webster, 1997; Webster, Douglas, Eaves, & Hart, 1997) and is a SPJ tool to assess risk of sexual violence among juveniles aged 12 to 18 years. The final risk estimate derived from using the ERASOR is short term (i.e., maximum 1 year) and should not be used to address questions related to long-term risk. The ERASOR has 9 identified static items, but the majority of the items (16) tap dynamic risk factors. Worling (2004) suggested that the ERASOR may assist clinicians to discriminate juveniles who have, for the first time, been detected for their sexual offenses from those who have sexually reoffended despite being sanctioned by an adult for a prior sexual assault. As with the J-SOAP-II, parts of the ERASOR might be used for assessing treatment needs and progress. The ERASOR has adequate to excellent interrater reliability and internal consistency (Viljoen, Elkovitch, Scalora, & Ullman, 2009; Worling, 2004). For the ERASOR total score, McCoy (2007) found adequate interrater reliabilities (.86-.88) and an acceptable internal consistency for the pretreatment rating (α = .79) but unacceptable for the posttreatment rating (α = .38).

The JRAS (based on the Registrant Risk Assessment Scale [RRAS] for adult offenders; Hiscox, Witt, & Haran, 2007; New Jersey Attorney General’s Office, 2006) is a SPJ Scale developed for JSOs. The JRAS differentiates between low risk, moderate risk, and high risk that is tapped by nine static items and five dynamic items. Three factors account for 49% of the variance of recidivism risk (see Table 1). Limited psychometric properties are available, but Hiscox et al. (2007) found an interrater reliability of .66.

Although not specifically developed for estimating risk of sexual offending, the SAVRY (Borum, Bartel, & Forth, 2003) is sometimes used, in addition to other instruments, for assessing risk among JSOs. The SAVRY is a SPJ assessment designed to assess violent risk. Nevertheless, it can also be used to predict recidivism among juveniles who have sexually offended (Viljoen et al., 2008). The SAVRY consists of one static scale and two dynamic scales (Borum et al., 2003). In addition to these factors, the SAVRY is also includes protective factors. The interrater reliability is .91 (Viljoen et al., 2008). The risk (.82) and protective factor (.73) scores had good internal consistency (Borum et al., 2003)

The PCL:YV (Forth, Kosson, & Hare, 2003) is a SPJ assessment designed to measure psychopathy traits among juveniles aged 12 to 18 years. Psychopathy traits are seen as a risk factor for future violent offending (Hare, 1999), and the PCL:YV has also been used to predict recidivism among JSOs. To score the PCL:YV a semistructured interview and additional information from available judicial files are required. Forth and Burke (1998) reported acceptable levels of internal consistency across several studies (α = .75-.89). The PCL:YV also has acceptable rates of interrater reliability (intra-class correlation coefficient of .93; Caldwell, Ziemke, & Vitacco, 2008).

Method

Sample

This study included retrospective and prospective studies on the predictive validity of risk assessment instruments for JSOs. A search of five electronic databases (PubMed, National Criminal Justice Reference Service, Web of Science, PiCarta, and Scopus) using the following keywords risk assessment, recidivism, predictive validity, young sex offenders, JSOs, juvenile delinquency, J-SOAP-II, J-SORRAT, ERASOR, JRAS, SAVRY, and PCL:YV yielded 296 articles. Inclusion was restricted to studies that estimated the predictive validity of structured risk assessment instruments for JSOs. The predictive validity criterion variable was restricted to a measure of recidivism such as rearrests and readjudication. Titles and abstracts were then examined to select articles that met these criteria. Of the 296 articles, 277 failed to meet these criteria, yielding a total of 19 published and unpublished articles that were used in this review. We have to note that there might be a publication bias favoring studies reporting significant results.

Effect Size Coding

Two effect sizes for risk assessment studies are suggested in the literature: point biserial correlation (r) and area under the curve (AUC) statistic from receiver operator characteristic curve analysis (Rice & Harris, 2005). The AUC statistic reflects the predictive validity of a given assessment instrument. AUC values range from 0 to 1, where 0 is perfect negative prediction, .5 is prediction at chance level, and 1 is perfect positive prediction. An advantage of using the AUC statistic is that its value is independent of the base rate of recidivism, selection ratios, and distributions in the tested sample (Harris et al., 2003). When available, the AUC, r, and Cohen’s d will be reported, based on the base rates that were found in the studies. When provided, AUC values and r were directly adopted from studies and converted to Cohen’s d, using the formula (Equation 1) from Rosenthal (1991) and Swets (1986). In this formula, r is the point biserial correlation, d is Cohen’s d, p is the base rate, and q is 1 − p.

d = \frac{r}{\sqrt{p q (1 - r^{2})}}

Results

Table 2 presents the number of participants, the follow-up period, and the base rates of recidivism found in each study. Table 3 presents the predictive validities (AUC values) and correlations (if provided) and Cohen’s d of the instruments for sexual, violent (nonsexual), and general recidivism. Regarding the J-SOAP-II, results are mixed for the total J-SOAP-II and the individual subscales. The total score on the J-SOAP-II was a significant predictor of sexual recidivism (Martinez, Flores, & Rosenfeld, 2007; Prentky et al., 2010; Rajlic & Gretton, 2010), nonsexual recidivism (Rajlic & Gretton, 2010), and general recidivism (Martinez et al., 2007). However, other studies found that the total J-SOAP-II score did not predict reoffending of any type (Caldwell et al., 2008; McCoy, 2007; Viljoen et al., 2008). Furthermore, instrument-informed clinical judgments based on the J-SOAP-II and the SAVRY also did not predict sexual or nonsexual violence (Elkovitch, Viljoen, Scalora, & Ullman, 2008). Several studies found that the Sexual Drive/Sexual Preoccupation Scale significantly predicted sexual recidivism (Hecker, Scoular, Righthand, & Nangle, 2002; Prentky et al., 2010; Rajlic & Gretton, 2010), whereas others did not (Caldwell et al., 2008; Martinez et al., 2007; McCoy, 2007; Parks & Bard, 2006). According to Viljoen et al. (2008), this subscale was associated with sexual aggression during treatment but not after treatment. This subscale was not able to predict nonsexual recidivism (Powers-Sawyer & Miner, 2009; Rajlic & Gretton, 2010) or general recidivism (Martinez et al., 2007). McCoy (2007) even found a negative correlation of the Sexual Drive/Sexual Preoccupation Scale with general recidivism. Although more studies found that the Impulsive/Antisocial subscale was not able to predict sexual recidivism (Caldwell et al., 2008; Hecker et al., 2002; Martinez et al., 2007; Prentky et al., 2010; Rajlic & Gretton, 2010), McCoy (2007) and Parks and Bard (2006) found that it was. This subscale was also able to predict nonsexual violent recidivism (Caldwell et al., 2008; Rajlic & Gretton, 2010) and general recidivism (Martinez et al., 2007; Waite et al., 2005), but Powers-Sawyer and Miner (2009) found that the Impulsive/Antisocial subscale was a poor predictor for nonsexual violent recidivism and nonsexual general recidivism. The Intervention and the Community Stability/Adjustment Scale predicted sexual (Caldwell et al., 2008; Martinez et al., 2007; Prentky et al., 2010, Rajlic & Gretton, 2010), nonsexual (Rajlic & Gretton, 2010), and general recidivism (Martinez et al., 2007), but this was not found in the studies of Parks and Bard (2006) and Hecker et al. (2002).

Table 2.

Base Rates of Recidivism

Authors	n	Follow-up (years)	% sexual recidivism	% violent (nonsexual) recidivism	% general recidivism
Caldwell, Ziemke, and Vitacco (2008)	91^a	4-8	12.1^a	NA	69.0^a
Caldwell, Ziemke, and Vitacco (2008)	174^b		11.6^b		88.4^b
Elkovitch, Viljoen, Scalora, and Ullman (2008)	166	1-12	8.4	10.2	20.5
Epperson, Ralston, Fowers, and DeWitt (2005)	636	12.5	19.8	NA	NA
Gretton, McBride, Hare, O’Shaughnessy and Kumka (2001)	220	1-8	15.0	30.0	50.9
Hecker, Scoular, Righthand, and Nangle (2002)	54	10-12	11.1	37.0	NA
Hiscox, Witt, and Haran (2007)	231	3-13	16.5	51.5	NA
Martinez, Flores, and Rosenfeld (2007)	60	NA	13.3	NA	20.0
McCoy (2007)	128	NA	5.6	56.0	57.8
Morton (2003)	77	5.6	17.0	26.0	52.0
Powers-Sawyer and Miner (2009)	96	1-15	NA	NA	NA
Parks and Bard (2006)	156	4-16	6.4	30.1	NA
Prentky et al. (2010)	336^c	7	24.7^c	NA	NA
	223^d		13.9^d
Rajlic and Gretton (2010)	286^e	6.6	9.4^e	33.9^e	43.4^e
	128^f		7.0^f	21.1^f	26.6^f
	140^g		12.9^g	59.3^g	64.3^g
Ralston (2008)	566	NA	12.4	NA	NA
Skowron (2005)	110	4	18.0	36.0	71.0
Viljoen et al. (2008)	169	12	8.3	12.7	42.8
Viljoen, Elkovitch, Scalora and Ullman (2009)	193	7.2	8.3	13.0	42.0
Waite et al. (2005)	256	10	4.7	38.3	57.4
Worling (2004)	136	1	NA	NA	NA

Note: NA = not available.

Juvenile sex offenders.

Juvenile offenders who have never been charged for a sexual offence.

Preadolescent sample.

Adolescent sample.

Total sample of adolescent sex offenders.

Adolescent sex offenders with sex offenses only.

Delinquent adolescent sex offenders.

Table 3.

Predictive Validity of the J-SOAP-II for Sexual, Violent (nonsexual), and General Recidivism

			Sexual recidivism			Violent (nonsexual) recidivism			General recidivism
Instrument	Authors	Scale	AUC	d	r	AUC	d	r	AUC	d	r
J-SOAP-II	Caldwell, Ziemke, and Vitacco (2008)	Total	—	0.246	.08	—	—	—	—	—	—
		3	—	0.529*	.17*	—	—	—	—	—	—
	Elkovitch, Viljoen, Scalora, and Ullman (2008)	Total^a	.44	0.216	−.06	.58	0.265	.09	.53	0.199	.08
	Hecker, Scoular, Righthand, and Nangle (2002)	1	.79*	—	—	—	—	—	—	—	—
	Martinez, Flores, and Rosenfeld (2007)	Total	.78*	0.960	.31*	—	—	—	.76**	0.904	.34**
		1 and 2	.63*	0.386	.13	—	—	—	.68	0.673	.26*
		3 and 4	.86***	0.324	.41**	—	—	—	.74**	0.874	.33*
	McCoy (2007)	Total	.60	0.705	.16	—	—	—	.50	0.020	.01
	Parks and Bard (2006)	1	—	0.123	−.03	—	0.109	−.05	—	—	—
		2	—	1.284	.30*	—	0.109	.05	—	—	—
		3	—	0.082	.02	—	0.153	−.07	—	—	—
	Powers-Sawyer and Miner (2009)	1 and 2	.75*	—	—	.57	—	—	—	—	—
		1	.72	—	—	.46	—	—	—	—	—
		2	.64	—	—	.64	—	—	—	—	—
	Prentky et al. (2010)	Total	.80*^b	—	—	—	—	—	—	—	—
			.83*^c
	Rajlic and Gretton (2010)	Total	.69**^d	—		.77***^d	—	—	—	—	—
			.80*^e	—	—	.62^e	—	—	—	—	—
			.51^f	—	—	.74*^f	—	—	—	—	—
	Viljoen et al. (2008)	Total	.54	—	—	.56	—	—	.56	—	—
	Waite et al. (2005)	2	NA
J-SORRAT-II	Epperson, Ralston, Fowers, and DeWitt (2005)	Total	.79*
	Ralston (2008)	Total	.64*
	Viljoen et al. (2008)	Total	.53	—	—	.56	—	—	.54	—	—
ERASOR	McCoy (2007)	Total	.50	0.087	−.02	—	—	—	.46	0.163	−.08
	Morton (2003)	Total	.74*^g	—	—	—	—	—	.65*	—	—
	Rajilc and Gretton (2010)	Total	.71***^d			.71***^d			—
			.86*^e	0.19***		.64*^e		.07
			.54^f	0.02		.58^f		.03
	Skowron (2005)	Total	.71*	—	—	.64*	—	—	.67*	—	—
	Viljoen, Elkovitch, Scalora, and Ullman (2009)	Total	.60			.56			.53
	Worling (2004)	Total	.72*	—	—	—	—	—	—	—	—
JRAS	Caldwell et al. (2008)	Total	<.50	0.00	.00	—	—	—
	Hiscox, Witt, and Haran (2007)	Total	—	0.409*	.15**	—	0.495*	.24**
		1	.669*	—	—	.699*	—	—
		2	.542	—	—	—	—	—
SAVRY	Elkovitch et al. (2008)	Total^a	.44	0.216	−.06	.58	0.265	.09	.53	0.199	.08
	Viljoen et al. (2008)	Total	.53	0.218	.06	.58	0.456	.15	.58	0.307	.15
PCL:YV	Caldwell et al. (2008)	Total	—	1.20***	.36***	—	—	—	—	—	—
	Gretton, McBride, Hare, O’Shaughnessy and Kumka (2001)	Total	.550	0.253	.09	.607**	0.422	.19**	.642**	0.516	.25**
	Parks and Bard (2006)	Total	—	—	—	—	0.175	.08***	—	—	—
		1	—	1.575	.36*	—	0.022	−.01	—	—	—
		2	—	0.082	−.02	—	0.153	.07	—	—	—
		3	—	0.535	−.13	—	0.219	.10*	—	—	—
		4	—	2.000	−.44*	—	0.539	.24***	—	—	—
	Viljoen et al. (2009)	Total	.49			.71**			.63*

Note: J-SOAP-II = Juvenile Sex Offender Assessment Protocol-II; J-SORRAT-II = Juvenile Sexual Offence Recidivism Risk Assessment Tool-II; ERASOR = Estimate of Risk of Adolescent Sexual Offence Recidivism; JRAS = Juvenile Risk Assessment Scale; SAVRY = Structured Assessment of Violent Risk in Youth; PCL:YV = Hare Psychopathy Checklist:Youth Version. AUC = area under the receiver operator characteristic curve; d = effect size based on base rates.

Clinical judgments based on the SAVRY; J-SOAP-II.

Preadolescent higher risk sample.

Adolescent higher risk sample.

Total sample.

Adolescent sex offenders with sex offenses only.

Delinquent adolescent sex offenders.

Modified prediction based on nine ERASOR items.

p < .05. **p < .01. ***p < .001.

In their risk prediction, some studies differentiated between subtypes. Mixed-type offenders scored higher on the Impulsive/Antisocial Scale, Intervention Scale, and Total Scale of the J-SOAP-II than child molesters and peer or adult molesters. On the Sexual Drive/Preoccupation Scale, mixed-type offenders scored higher than child molesters, who scored higher than peer or adult molesters (Parks & Bard, 2006). Rajlic and Gretton (2010) differentiated between juveniles with a history of general offending and juveniles with a sex offence only. The J-SOAP-II predicted sexual recidivism for the sex-only group and nonsexual recidivism for juveniles with a history of general offending.

Very little research has been done on the predictive validity of the J-SORRAT-II. Epperson et al. (2005) noted that the J-SORRAT-II is able to predict sexual reoffending very well. However, the predictive validity was based on the exact same sample that was used to develop the instrument. Ralston (2008) extended this sample with independent participants and found a lower but significant predictive validity for sexual recidivism. A total independent sample showed that this instrument did not predict reoffending of any type (Viljoen et al., 2008).

The results on the ERASOR are also equivocal. Several studies found the ERASOR to be a moderate to strong predictor of sexual, nonsexual, and general recidivism (Morton, 2003; Rajlic & Gretton, 2010; Skowron, 2005; Worling, 2004). However, Morton (2003) used a modified version of nine ERASOR items to obtain significant results for sexual recidivism, otherwise no significant results were found. In the study of Rajlic and Gretton (2010), the ERASOR predicted sexual and nonsexual recidivism for the sex-offence-only group but not for the juveniles with a history of general offending. In contrast to these findings, Viljoen et al. (2009) and McCoy (2007) did not find support for the validity of this measure to predict recidivism of any type. SPJs, however, nearly reached significance in predicting sexual reoffending (Viljoen et al., 2009). So far, no study has been able to use the ERASOR as it was originally developed by Worling and Curwen (2000), resulting in different findings.

The JRAS moderately predicted sexual and nonsexual recidivism in a group of young males who were adjudicated for a sexual offence (Hiscox et al., 2007). The antisocial factor moderately predicted sexual and nonsexual recidivism. However, the sexual deviance factor did not predict sexual or nonsexual recidivism (Hiscox et al., 2007). In another study (Caldwell et al., 2008), the JRAS failed to predict sexual, nonsexual violent, or general reoffending.

Two studies have examined the predictive value of the SAVRY for reoffending of JSOs (Elkovitch et al., 2008; Viljoen et al., 2008). As mentioned earlier, Elkovitch et al. (2008) found that instrument-informed clinical judgments based on the J-SOAP-II and SAVRY did not significantly predict sexual and nonsexual violence. Total scores on the SAVRY did not predict serious nonsexual violent offending (defined as at least one violent felony charge; Viljoen et al., 2008). In addition, it was not able to predict violent felonies or violent misdemeanors, sexual violent reoffending, or reoffending of any type. Only the historical scale was able to predict any nonsexual violent reoffending (Viljoen et al., 2008).

The PCL:YV was a moderate to strong predictor of nonsexual and general recidivism (Caldwell et al., 2008; Gretton, McBride, Hare, O’Shaughnessy, & Kumka, 2001; Parks & Bard, 2006; Viljoen et al., 2009). The PCL:YV could not predict sexual recidivism. Only Caldwell et al. (2008) found the total score of the PCL:YV to be predictive of new sex offense charges. This result might be due to extremely high PCL:YV scores (i.e., more than 34) of juveniles who sexually offended in the follow-up period.

Parks and Bard (2006) examined the PCL:YV for predicting sexual and nonsexual recidivism, resulting in a three-predictor model for sexual recidivism that included the Impulsive/Antisocial Scale of the J-SOAP-II and the Interpersonal factor and Antisocial factor of the PCL:YV. A two-factor predictor model for nonsexual recidivism included the Behavioral factor and Antisocial factor of the PCL:YV. Neither of the J-SOAP-II nor PCL:YV total scores was identified as a predictor of sexual recidivism; however, the PCL:YV total was identified as a predictor of nonsexual recidivism. Parks and Bard found higher scores on the PCL:YV among mixed-type offenders as compared with those who exclusively offend sexually against children or peers/adults. However, they did not further examine the predictive validity of the PCL:YV for each of the offender types.

Discussion

This study reviewed six measures commonly used for risk assessment in JSOs: the J-SOAP-II, J-SORRAT-II, ERASOR, JRAS, SAVRY, and PCL:YV. In search of their qualities, we gave an overview of literature on the predictive validities of risk assessment instruments for sexual, nonsexual, and general reoffending. Although some of the instruments seem promising for risk assessment among JSOs, there is no one instrument that shows unequivocal positive results in predicting future offending among this population. Not unexpectedly, the results obtained by the SAVRY and PCL-YV for sexual recidivism appeared to be weaker than specialized tools such as the J-SOAP-II or the ERASOR. Nonsexual, that is, violent and general recidivism among JSOs, can best be predicted by the ERASOR or the PCL:YV. As little research has been done on the J-SORRAT-II and JRAS, it is too early to draw conclusions about the predictive accuracy of these instruments. Note that studies that found a significant predictive validity for an instrument were often conducted by the individual or group that had developed the measure. More independent research is needed to draw objective conclusions.

Although it is one of the most commonly used measures in the United Sates with JSOs, the results of the J-SOAP-II were mixed across studies, a problem that also applies to the other instruments. Mixed results might have been influenced by differences in procedures or samples: was it based on an actuarial approach or on a structured clinical judgment; the mean age of the sample; the index offences for which the JSOs were adjudicated; the type of sample (incarcerated, community, never charged); the follow-up times, and whether the follow-up time commenced immediately after the risk assessment, or after a period of incarceration or treatment; and the definition of recidivism. Most of the instruments currently used are SPJs, with a partially subjective interpretation of the risk by a clinician. Mixed findings might also be due to low rates of sexual reoffending, which may make sexual reoffending challenging to predict. Furthermore, it is important to notice that results vary when differentiating between subtypes. Bartosh, Garby, Lewis, and Gray (2003) suggested that taking adult offender type into account might increase predictive accuracy. This is also likely to be true for the juvenile population. However, the attempts to classify JSOs were mostly intuitively derived and have not been empirically validated (Hendriks, 2006; Veneziano & Veneziano, 2002).

More knowledge about how the instruments differ may shed light on why one instrument is a good predictor of recidivism and another instrument is not. One explanation for the promising predictive validity of the ERASOR is that, in contrast to the other instruments, this instrument focuses more on dynamic than on static risk factors. Unlike static risk factors (e.g., offence history), dynamic risk factors are factors that are amenable to deliberate interventions (e.g., substance abuse, unemployment). It would seem that a risk assessment instrument mainly based on dynamic risk factors can pick up small improvements in risk factors such as sexual interests and social functioning of sex offenders. This is the reason that this risk estimate is short term and should not be used to address questions related to long-term risk. Furthermore, the ERASOR includes more items assessing cognitive factors and sexual deviance. Dynamic risk factors such as atypical sexual fantasies, behaviors or interests, and unwillingness to alter deviant sexual interests or attitudes have been shown to be important in the initiation and maintenance of sexual offending behavior among JSOs and elevate the risk of sexual offending (Hunter, Goodwin, & Becker, 1994; Seto & Lalumière, 2010). However, more studies need to confirm this finding, especially as we do not know what factors are related to the persistency of sexual offending. In contrast, other instruments, particularly the J-SORRAT-II, focus more on the history of offenders with respect to behavioral problems and offences, which are static factors.

The Sexual Deviance subscale of the JRAS did not predict recidivism (Hiscox et al., 2007). This is remarkable because sexual deviance is regarded as a potentially important risk factor among adults and juveniles (Beech & Ford, 2006; Hanson & Morton-Bourgon, 2009; Långström & Grann, 2000; Seto & Lalumière, 2010; Ward, 2000; Ward & Keenan, 1999). However, examination of the JRAS Scales shows that static and quantitative items like degree of contact, number of sexual offences or victims in the past, duration of offensive behavior, and victim gender comprise the sexual deviance scale while sexual deviance is mostly regarded as a dynamic and qualitative concept. Besides, there seems to be a problem with the definition of sexual deviance. This is particularly true in children. As paraphilias do not apply to children under 16, what, then, can be regarded as deviant and as predictive of persistent sexual behavior? Although their items are static, the developers of the JRAS explained the lack of predictability by the possibility that sexual deviance and identities among juveniles have not been completely formed. Furthermore, sexual deviance may only be a predictor in some subtypes among sex offenders but not in all. As the study by Hiscox et al. (2007) used a sample of mixed-type offenders, this may be an alternative explanation as to why the sexual deviance factor did not predict recidivism.

Another problem of the JRAS Scale, and of the scales of other instruments, is that the validity of the scales is mainly based on low-risk samples, and the inclusion of items is mostly based on their proven predictive value in adults. Risk factors for JSOs are often extrapolated from the adult literature instead of empirical determination. The development of the JRAS, for example, was initially based on a rational analysis, reviewing items of the RRAS and reaching consensus on what criteria needed to be modified or added to make the scale more suitable for juveniles (Hiscox et al., 2007). The J-SOAP-II variables were developed after reviews of the literature that covered five areas, including risk assessment or outcome studies of adult sex offenders, and risk assessment studies on mixed populations of adult offenders (Prentky & Righthand, 2003). The PCL:YV and ERASOR are also based on adult risk assessment instruments. Risk assessment might be beneficial to the juvenile, for example, when assessing its urgent treatment needs. However, the extrapolation of juvenile risk factors from the adult literature may lead to inaccuracies in the prediction of risk. Juveniles are still developing their personality, cognitions, and moral judgment, processes that reflect considerable plasticity (Nelson & Bloom, 1997). Nevertheless, based on the assessment, long-term consequences, that is, restrictions, are imposed on JSOs, assuming that sex offending in juveniles is something immutable and caused by stable internal traits. For example, the clinical use of the PCL:YV might lead to the assumption that a person has many traits of psychopathy and must therefore be at high risk for reoffending, as psychopathy is regarded as untreatable. Yet, from the time of their offence, there are still many possible developmental pathways, and no one knows what causes persistent sexual offending. Often, scales are not sensitive to child and adolescent development and to mutable and unstable traits. The developmental processes are viewed as being irrelevant once a juvenile has engaged in sexual misconduct, evaluating them as adults. If the assessments are used to make important decisions, we should be at a point where there is consensus about what causes the onset and persistency of sexual offending in juveniles. Clearly, this is not the case, but still judgments are based on these uncertainties.

To conclude, the predictive validities of the risk assessment instruments for JSOs are still insufficient to accurately predict recidivism. It is difficult to assess the risk accurately when this is based on changing risk factors because of the rapid development of juveniles. Therefore, a significant proportion of uncertainty remains. Because of this uncertainty, it is highly questionable whether it is ethical to impose long-term consequences on juveniles based on these assessments. So, how can risk predictions be improved? To this end, three recommendations can be made.

First, there is a need for reliable and valid typologies for JSOs. While knowing that there are differences between types of JSOs, we do not know what the shared factors are, what discriminates them, or how to define each type while excluding others. Studies need to clarify what dynamic forces influence the onset of sexual offending and what causes persistency. Discriminating between subtypes might contribute to a better understanding of certain etiological pathways of offending as well as specific treatment needs for subtypes of JSOs based on the factors that are most susceptible for intervention. Furthermore, such discrimination is likely to improve the successfulness of the risk predictions and the derived decisions for referrals or rehabilitation (Hendriks, 2006; Worling, 2001).

Second, although the emphasis of risk assessment is now based on empirically derived factors and SPJ, more attention should be paid to the psychological representatives of atypical sexual fantasies, behaviors, or interests as risk factors, such as cognitive distortions (distorted fantasies, paraphilias, distorted thoughts, and interpretations), deviant sexual arousal patterns and preferences, sexual preoccupations, and intimate sexual behavior. As Seto and Lalumière (2010) stated, the differences found between male JSOs and male juvenile nonsex offenders might indicate possible relationships with sexual offending. However, these variables are now underexposed in risk assessment instruments but might also be valuable in assessing JSOs. Nevertheless, we have to be careful with the evaluation of atypical sexual fantasies, behaviors, or interests in juveniles, as juveniles are still in a developmental phase, and no clear definitions of “atypical” or “deviant” are available. Therefore, studies need to examine whether these factors are indeed strongly related to the onset of sexual offending among juveniles and which factors cause persistent sexual offending.

Finally, other ways of assessing sexual risk than assessments based only on items and clinical impressions should be developed, and, as juveniles are rapidly developing, there is a need for reliable measures concerning short-term risk. Risk assessment may be expanded by the introduction of psychiatric, psychological, and biopsychological measurement tasks on reactive behavior on certain sexual cues. We believe that much can be gained by measuring implicit cognitions and associations about sex, estimated by implicit association tests (e.g., Greenwald, McGhee, & Schwartz, 1998). Brown, Gray, and Snowden (2009) found differences between types of adult offenders in implicit associations between children and sex. This association was present in pedophilic offenders irrespective of their denial of offence history. Furthermore, the measurement of viewing time (Glasgow, Osborne, & Croxen, 2003) could also be useful in the assessment of sexual interests. Viewing time is the length of time spent viewing an image of a person and has been reported to be significantly correlated with sexual interest in adults (Harris, Rice, Quinsey, & Chaplin, 1996). In a study of Worling (2006), viewing time significantly differentiated juveniles who assaulted male children from juveniles who assaulted other individuals. Spoken or read vignettes describing situations of sexual activities or by operations in a virtual reality setting might also be helpful in detecting sexual preoccupations and risks. These are promising methods for the adult population, and studies should examine their value for the JSO population. Using these new measures alongside existing assessment instruments may greatly improve the accuracy of risk assessment and treatment.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Bartosh

D. L.

Garby

Lewis

Gray

(2003). Differences in the predictive validity of actuarial risk assessments in relation to sex offender type. International Journal of Offender Therapy and Comparative Criminology, 47, 422-438.

Beech

Ford

(2006). The relationship between risk, deviance, treatment outcome and sexual reconviction in a sample of child sexual sexual abusers completing residential treatment for their offending. Psychology, Crime & Law, 12, 685-701.

Boer

D. P.

Hart

S. D.

Kropp

P. R.

Webster

C. D.

(1997). Manual for the sexual violence risk-20: Professional guidelines for assessing risk of sexual violence. Vancouver: British Columbia Institute on Family Violence.

Borum

Bartel

Forth

(2003). Manual for the Structured Assessment of Violence Risk in Youth, Version 1.1. Tampa: University of South Florida.

Brown

A. S.

Gray

N. S.

Snowden

R. J.

(2009). Implicit measurement of sexual associations in child sex abusers: Role of victim type and denial. Sexual Abuse, 21, 166-180.

Caldwell

M. F.

(2002). What we do not know about juvenile sexual reoffense risk. Child Maltreatment, 7, 291-302.

Caldwell

M. F.

(2010). Study characteristics and recidivism base rates in juvenile sex offender recidivism. International Journal of Offender Therapy and Comparative Criminology, 54, 197-212.

Caldwell

M. F.

Ziemke

M. H.

Vitacco

M. J.

(2008). An examination of the sex offender registration act as applied to juveniles: Evaluating the ability to predict sexual recidivism. Psychology, Public Policy, and Law, 14, 89-114.

Douglas

Ogloff

Hart

(2003). Evaluation of a model of violence risk assessment among forensic psychiatric patients. Psychiatric Services, 54, 1372-1379.

10.

Elkovitch

Viljoen

J. L.

Scalora

M. J.

Ullman

(2008). Assessing risk of reoffending in adolescents who have committed a sexual offense: The accuracy of clinical judgments after completion of risk assessment instruments. Behavioral Sciences & the Law, 26, 511-528.

11.

Epperson

D. L.

Kaul

J. D.

Hesselton

(1999). Minnesota sex offender screening tool-revised: Development, performance, and recommended risk level cut off scores. St. Paul: Minnesota Department of Corrections.

12.

Epperson

D. L.

Ralston

C. A.

Fowers

DeWitt

(2005). Development of a sexual offense recidivism risk assessment tool—II (JSORRAT-II) Unpublished manuscript, University of Iowa, Ames.

13.

Federal Bureau of Investigation. (2008). Uniform crime reports for the United States: Crime in the United States, 2008. Washington, DC: U.S. Department of Justice.

14.

Forth

A. E.

Burke

(1998). Psychopathy in adolescence: Assessment, violence and developmental precursors. In Cooke

Forth

A. E.

Hare

R. D.

(Eds.), Psychopathy: Theory, research and implications for society (pp. 205-229). Boston, MA: Kluwer Academic.

15.

Forth

A. E.

Kosson

D. S.

Hare

R. D.

(2003). The Psychopathy Checklist: Youth Version. Toronto, Ontario, Canada: Multi-Health Systems.

16.

Glasgow

D. V.

Osborne

Croxen

(2003). An assessment tool for investigating paedophile sexual interest using viewing time: An application of single case methodology. British Journal of Learning Disabilities, 31, 96-102.

17.

Greenwald

A. G.

McGhee

D. E.

Schwartz

J. L. K.

(1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464-1480.

18.

Gretton

H. M.

McBride

Hare

R. D.

O’Shaughnessy

Kumka

(2001). Psychopathy and recidivism in adolescent sex offenders. Criminal Justice and Behavior, 28, 427-449.

19.

Hagan

Cho

(1996). A comparison of treatment outcomes between adolescent rapists and child sexual offenders. International Journal of Offender Therapy and Comparative Criminology, 40, 113-122.

20.

Hanson

R. K.

(2000). Risk assessment. Beaverton, OR: ATSA.

21.

Hanson

R. K.

Morton-Bourgon

K. E.

(2009). The accuracy of recidivism risk assessments for sexual offenders: A meta-analysis of 118 prediction studies. Psychological Assessment, 21, 1-21.

22.

Hanson

R. K.

Thornton

(1999). Static 99: Improving actuarial risk assessments for sex offenders (User Report No. 98 01). Ottawa, Ontario: Department of the Solicitor General of Canada.

23.

Hare

R. D.

(1999). Psychopathy as a risk factor for violence. Psychiatric Quarterly, 70, 181-197.

24.

Harris

G. T.

Rice

M. E.

Quinsey

V. L.

Chaplin

T. C.

(1996). Viewing time as a measure of sexual interest among child molesters and normal heterosexual men. Behaviour Research and Therapy, 34, 389-394.

25.

Harris

G. T.

Rice

M. E.

Quinsey

V. L.

Lalumière

M. L.

Boer

Lang

(2003). A multisite comparison of actuarial risk instruments for sex offenders. Psychological Assessment, 15, 413-425.

26.

Hecker

Scoular

Righthand

Nangle

(2002, October). Predictive validity of the J-SOAP over 10-plus years: Implications for risk assessment. Paper presented at the Annual Meeting of the Association for Treatment of Sexual Abusers, Montreal, Quebec, Canada.

27.

Hendriks

(2006). Jeugdige zedendelinquenten: Een studie naar subtypen en recidive [Juvenile sexual offenders: A study of subtypes and recidivism].Utrecht: Forum Educatief.

28.

Hiscox

S. P.

Witt

P. H.

Haran

S. J.

(2007). Juvenile Risk Assessment Scale (JRAS): A predictive validity study. Journal of Psychiatry & Law, 35, 503-539.

29.

Hsu

L. K.

Starzynski

(1990). Adolescent rapists and adolescent child sexual assaulters. International Journal of Offender Therapy and Comparative Criminology, 34, 23-30.

30.

Hunter

J. A.

Goodwin

D. W.

Becker

J. V.

(1994). The relationship between phallometrically measured deviant sexual arousal and clinical characteristics in juvenile sex offenders. Behavior Research and Therapy, 32, 533-538.

31.

Långström

Grann

(2000). Risk for criminal recidivism among young sex offenders. Journal of Interpersonal Violence, 15, 855-871.

32.

Martinez

Flores

Rosenfeld

(2007). Validity of the juvenile sex offender assessment protocol-II (J-SOAP-II) in a sample of urban minority youth. Criminal Justice and Behavior, 34, 1284-1295.

33.

McCoy

W. K.

(2007). Predicting treatment outcome and recidivism among juvenile sex offenders: The utility of the J-SOAP-II and ERASOR in an outpatient treatment program (Unpublished doctoral dissertation). Sam Houston State University, Huntsville, TX.

34.

Morton

K. E.

(2003). Psychometric properties of four risk assessment measures with male adolescent sex offenders (Unpublished master’s thesis). Carleton University, Ottawa, Ontario, Canada.

35.

Nelson

C. A.

Bloom

F. E.

(1997). Child development and neuroscience. Child Development, 68, 970-987.

36.

New Jersey Attorney General’s Office. (2006). Juvenile Risk Assessment Scale manual. Retrieved from http://www.state.nj.us/lps/dcj/megan/jras-manual-scale-606.pdf

37.

Parks

G. A.

Bard

D. E.

(2006). Risk factors for adolescent sex offender recidivism: Evaluation of predictive factors and comparison of three groups based upon victim type. Sexual Abuse, 18, 319-342.

38.

Powers-Sawyer

A. B.

Miner

M. H.

(2009). Actuarial prediction of juvenile recidivism: The static variables of the juvenile sex offender assessment protocol-II (J-SOAP-II). Sexual Offender Treatment, 4, 1-11.

39.

Prentky

R. A.

N-C.

Righthand

Schuler

Cavanaugh

Lee

A. F.

(2010). Assessing risk of sexually abusive behavior among youth in a child welfare sample. Behavioral Sciences & the Law, 28, 24-45.

40.

Prentky

R. A.

Pimental

Cavanaugh

D. J.

Righthand

(2009). Predicting risk of sexual recidivism in juveniles: Predictive validity of the J-SOAP-II. In Beech

A. R.

Craig

L. A.

Browne

K. D.

(Eds.), Assessment and treatment of sex offenders (pp. 265-292). West Sussex, UK: John Wiley.

41.

Prentky

R. A.

Righthand

(2003). Juvenile Sex Offender Assessment Protocol (J-SOAP-II): Manual. Bridgewater, MA: Justice Research Institute.

42.

Rajlic

Gretton

H. M.

(2010). An examination of two sexual recidivism risk measures in adolescent offenders: The moderating effect of offender type. Criminal Justice and Behavior, 37, 1066-1085.

43.

Ralston

C. A.

(2008). Validation of the Juvenile Sexual Offense Recidivism Risk Assessment Tool—II (Unpublished doctoral dissertation). Iowa State University, Iowa.

44.

Rice

M. E.

Harris

G. T.

(2005). Comparing effect sizes in follow-up studies: ROC Area, Cohen’s d, and r. Law and Human Behavior, 29, 615-620.

45.

Rosenthal

(1991). Meta-analytic procedures for social research. Newbury Park, CA: SAGE.

46.

Seto

M. C.

Lalumière

M. L.

(2010). What is so special about male adolescent sexual offending? A review and test of explanations through meta-analysis. Psychological Bulletin, 136, 526-575.

47.

Skowron

(2005). Differentiation and predictive factors in adolescent sexual offending (Unpublished doctoral dissertation). Carleton University, Ottawa, Ontario, Canada.

48.

Smith

W. R.

Monastersky

Deisher

R. M.

(1987). MMPI-based personality types among juvenile sex offenders. Journal of Clinical Psychology, 43, 422-430.

49.

Swets

J. A.

(1986). Indices of discrimination or diagnostic accuracy: Their ROCs and implied models. Psychological Bulletin, 99, 100-117.

50.

Veneziano

(2002). Adolescent sex offenders: A review of the literature. Trauma, Violence, & Abuse, 3, 247-260.

51.

Viljoen

J. L.

Elkovitch

Scalora

Ullman

(2009). Assessment of reoffense risk in adolescents who have committed sexual offenses: Predictive validity of the ERASOR, PCL: YV, YLS/CMI and static-99. Criminal Justice and Behavior, 36, 981-1000.

52.

Viljoen

J. L.

Scalora

Cuadra

Bader

Chávez

Ullman

Lawrence

(2008). Assessing risk for violence in adolescents who have sexually offended: A comparison of the J-SOAP-II, J-SORRAT-II, and SAVRY. Criminal Justice and Behavior, 35, 5-23.

53.

Waite

Keller

McGarvey

E. R.

Wieckowski

Pinkerton

Brown

G. L.

(2005). Juvenile sex offender re-arrest rates for sexual, violent nonsexual, and property crimes: A 10-year follow-up. Sexual Abuse: A Journal of Research and Treatment, 17, 313-331.

54.

Ward

(2000). Sexual offenders’ cognitive distortions as implicit theories. Aggression and Violent Behavior, 5, 491-507.

55.

Ward

Keenan

(1999). Child molesters’ implicit theories. Journal of Interpersonal Violence, 14, 821-838.

56.

Webster

C. D.

Douglas

Eaves

Hart

S. D.

(1997). HCR-20: Assessing risk for violence, version 2. Burnaby, Canada: Simon Fraser University and Forensic Psychiatric Services Commission of British Columbia.

57.

Worling

J. R.

(2001). Personality-based typology of adolescent male sexual offenders: Differences in recidivism rates, victim-selection characteristics, and personal victimization histories. Sexual Abuse: A Journal of Research and Treatment, 13, 149-166.

58.

Worling

J. R.

(2004). The estimate of risk of adolescent sexual offence recidivism (ERASOR): Preliminary psychometric data. Sexual Abuse, 16, 235-254.

59.

Worling

J. R.

(2006). Assessing sexual arousal with adolescent males who have offended sexually: Self-report and unobtrusively measured viewing time. Sexual Abuse: A Journal of Research and Treatment, 18, 383-400.

60.

Worling

J. R.

Curwen

(2000). The ERASOR: Estimate of risk of adolescent sexual offence recidivism. Toronto, Ontario, Canada: SAFE-T Program, Thistletown Regional Centre.

61.

Worling

J. R.

Litteljohn

Bookalam

(2010). 20-year prospective follow-up study of specialized treatment for adolescents who offended sexually. Behavioral Sciences & the Law, 28, 46-57.