Protective Factors and Recidivism in Accused Juveniles Who Sexually Offended

Abstract

To date, research on juvenile sexual offender recidivism has tended to focus on risk factors rather than protective factors. Therefore, very little is known about protective factors in the population of juveniles who sexually offended. The aim of the present study was to examine the impact of protective factors on non-recidivism in a sample of accused juveniles who sexually offended (N = 71) in a mean follow-up period of 47.84 months. Protective factors were measured with the Protective Factor Scale of the Structured Assessment of Violence Risk in Youth (SAVRY), and the Structured Assessment of PROtective Factors for violence risk (SAPROF). Criminal charges served as recidivism data. The internal scale of the SAPROF, in particular, yielded moderate predictive accuracy for the absence of violent and general recidivism, though not for the absence of sexual recidivism. No protective factor of the SAVRY did reveal predictive accuracy regarding various types of the absence of recidivism. Furthermore, protective factors failed to achieve any significant incremental predictive accuracy beyond that captured by the SAVRY risk factors alone. The potential therapeutic benefit of protective factors in juvenile sexual offender treatment is discussed.

Keywords

juvenile sexual offender protective factors SAPROF SAVRY recidivism

In recent years, research on juveniles who sexually offended has focused predominantly on the prediction of recidivism and juvenile sexual reoffense risk. The state of research into factors that may affect recidivism has so far been dominated by a focus on risk factors, with a large number of risk assessment tools being developed and studies on predictive accuracy being conducted (e.g., Caldwell, 2013; Chu, Ng, Fong, & Teoh, 2012; Fanniff & Letourneau, 2012; Hempel, Buck, Cima, & van Marle, 2013; Viljoen, Mordell, & Beneteau, 2012; Worling, Bookalam, & Litteljohn, 2012). However, this mere focus on risk factors in applied forensic assessment may not be sufficient to either identify motivational aspects of criminal behavior or intervene on an individual level in terms of prevention or treatment (Jessor, 2013) and could result in inaccurate and biased predictions of recidivism. This is especially relevant when considering that most young people who offended go on to desist from such behavior across their life span (Moffitt, 1993, 2003). Hence, a comprehensive recidivism risk assessment—especially for young offenders—should also consider dynamic and changeable risk and protective factors, and should not rely on risk factors alone (e.g., de Vogel, de Ruiter, Bouman, & de Vries Robbé, 2009; Miller, 2006; Rogers, 2000; Ullrich & Coid, 2011).

Defining Protective Factors

The conceptualization referring to protective factors remains diverse among researchers. It has been suggested that such constructs related to pro-social behavior as resilience (Rutter, 1985) or protective factors (Costa, Jessor, & Turbin, 1999; Jessor, 1991) are also related to the prevention of risk behaviors. However, some researchers recommended considering the absence of risk factors as a protective factor (Busch et al., 2009; Costa et al., 1999; Zagar, Busch, Grove, & Hughes, 2009). In mental health studies, protective factors have usually been defined as factors in personal, social, and external support systems, which modify, ameliorate, compensate, or alter a person’s response to risk factors for any maladaptive life event and thus reduce the probability of those outcomes (Fitzpatrick, 1997; Masten & Garmezy, 1985; Rutter, 1985). The focus of desistance models in youth criminal behaviors has rather been on identifying the factors that help youths to “grow .out” of a criminal career. As physical and mental maturity in adolescents changes over time, most adolescents desist from criminal behavior over their developmental life course as a result of changes in personal or social contexts, due either to being passively influenced by those changes or to actively calculating costs and benefits (Laub & Sampson, 2001; Moffitt, 1993, 2003; Sampson & Laub, 1993).

Protective Factors for Violent Behavior

Most research has concentrated on the possible influence of protective factors on violent behavior in school-age youths. Blum and Ireland (2004) suggested that family and school connectedness and religiosity constitute protective factors against such health-compromising behavior as violence involvement, impersonal sexual intercourse, tobacco use, or alcohol use in their sample of 15,695 Caribbean school-age youths. Resnick, Ireland, and Borowsky (2004) conducted a large national longitudinal study of 13,110 adolescents in which the impact of various risk and protective factors for involvement in violent behavior was examined a year after the first data collection. Various protective factors were identified such as school and adult connectedness for 6,913 male and 7,419 female students, which contributed toward minimizing violent behavior even when risk factors were simultaneously present. Stouthamer-Loeber, Loeber, Stallings, and Lacourse (2008) and Stouthamer-Loeber, Wei, Loeber, and Masten (2004) pointed out a varied spectrum of personal, socio-economic, and environmental factors in samples of U.S.-American school boys (n = 506, replicated in n = 1,009), for example, a low level of physical punishment, being employed or in school, believing that one is likely to be caught, and having a good relationship with peers.

Protective Factors for Sexually Aggressive Behavior

Protective factors have only seldom been studied in the context of the development of sexually aggressive behavior in adolescence. In an extensive survey, 71,594 student participants were asked whether they had ever forced someone into a sexual act. The following factors were found to protect against involvement in sexually aggressive behavior: emotional health and the presence of bonds with friends and adults (Borowsky, Hogan, & Ireland, 1997).

Although the studies were conducted on a large scale and with representative samples, a number of the studies listed can be recognized as deriving from the same projects, for example, the National Longitudinal Study on Adolescents Health (Blum & Ireland, 2004; Borowsky et al., 1997; Resnick et al., 2004), or the Association for Prevention Teaching and Research Award funded by Centers for Disease Control and Prevention (Loeber & Farrington, 2012; Stouthamer-Loeber et al., 2008; Stouthamer-Loeber et al., 2004). It is, thus, questionable whether the results described can be generalized.

Assessment of Protective Factors and Criminal Recidivism in Juvenile Violent Offenders

Protective factors in delinquents and their impact on recidivism have been researched even less. Hoge, Andrews, and Leschied (1996) studied a sample of 338 delinquent adolescents and identified the following protective factors as being associated with reduced recidivism: positive peer relations, good school achievement, positive response to authority, and effective use of leisure time. Salekin, Lee, Schrum-Dillard, and Kubak (2010) suggested that motivation to change was protective in a sample of 140 U.S.-American child and adolescent offenders (92 males and 48 females).

A growing number of studies have been conducted regarding the reliability and validity of assessment tools for protective factors. The focus of the research conducted on the Structured Assessment of Violence Risk in Youth (SAVRY; Borum, Bartel, & Forth, 2003), the only Structured Professional Judgment (SPJ; Douglas & Reeves, 2010) instrument for risk and protective factors in juveniles, has been placed primarily on its risk factors. Nevertheless, recent studies showed significant relationships between a lack of protective factors in the SAVRY and a higher recidivism risk. A study by Lodewijks, de Ruiter, and Doreleijers (2010), in particular, investigated the relevance of the Protective Factors Scale of the SAVRY in 224 violent adolescent offenders. In their study, low-risk adolescent offenders had significantly more protective factors than the high-risk group. Overall, a lack of protective factors in the high-risk group was significantly associated with higher rates of violent reoffending in the follow-up period. In addition, Rennie and Dolan (2010) investigated the influence of protective factors in reducing risk among 135 juvenile delinquents, with the result that those with higher scores in the Protective Factors Scale were older at the time of their first arrest and had less psychopathological problems than offenders without any protective factors. Furthermore, a group comparison of adolescents who did and did not reoffend during the 12-month follow-up showed that those without reoffending had significantly more protective factors.

Protective Factors in Juveniles Who Sexually Offended

It is still the case that very little is known about protective factors in the population of juveniles who sexually offend. The first study to be conducted on the influence of protective factors on recidivism in a sample of juveniles who sexually offended was reported by Spice, Viljoen, Latzman, Scalora, and Ullman (2013). The SAVRY was used for a sample of 193 juveniles who sexually offended to examine the association between protective factors and risk factors on sexual and non-sexual recidivism with a mean follow-up period of 7.24 years. No protective factor included in the SAVRY was correlated with sexual recidivism, and only the protective factor strong attachment and bonds of the SAVRY was negatively associated with non-sexual recidivism. Spice et al. (2013) assumed, therefore, that specific protective factors for sexual recidivism might potentially exist, which are different from those for general and violent recidivism. In a study examining the prevalence of protective factors in a sample of 66 young accused juveniles who sexually offended, also using parts of the sample assessed within the current study, Klein et al. (2012) reported a negative correlation between protective factors measured by the Structured Assessment of PROtective Factors for violence risk (SAPROF; de Vogel et al., 2009), and the SAVRY risk factors. Furthermore, a negative relationship between protective factors and psychopathological problems was reported. This was the first published study using the SAPROF, a SPJ instrument for adult forensic patients designed exclusively for the assessment of protective factors, in a juvenile sample.

Overall, the influence of protective factors on the occurrence or persistence of criminal behavior in juveniles who sexually offended has not been sufficiently investigated, yet. Investigations regarding the possible influence of protective factors on desistance from future offending or antisocial behavior could contribute toward an improvement in juvenile sexual offender treatment. The aim of the present study was therefore to explore the predictive accuracy of various protective factors concerning recidivism in accused juveniles who sexually offended as measured by the SAPROF and the SAVRY Protective Factor Scale. Furthermore, the incremental predictive validity of these protective factors beyond the SAVRY risk factors was examined.

Method

Sample Description

The present study was part of the Hamburg Model Project for children and adolescents at risk for sexual offending (for further descriptions of the Hamburg Model Project, see Driemeyer, Spehr, Yoon, Richter-Appelt, & Briken, 2013; Klein et al., 2012; Spehr, Yoon, & Briken, 2010). The longitudinal project aimed at assessing and intervening in all children and adolescents who had been accused of sexual offending according to the German Criminal Code in Hamburg, Germany. All youths had been reported by the police to the Family Intervention Team (FIT) of the youth welfare services from 2007 to 2010. The FIT constitutes of an interdisciplinary professional team including psychologists and social workers who conduct comprehensive diagnostics and assessments on every reported juvenile and refer them to community services or non-profit treatment centers. In total, 177 juveniles were registered and 83 gave their written informed consent to participate in the scientific evaluation (47%). All participants and their legal guardians signed a declaration of consent. Participants each received 20 Euros at the time they consented to the study. The ethics committee of the Hamburg Medical Council approved the completion of the study. In addition, Data Protection Supervisor and the police department of Hamburg approved the use of the recidivism data for research purposes. Furthermore, the authority of the City of Hamburg agreed to the publication of the study.

The initial sample consisted of 83 male juveniles. Charges were used as the index offense inclusion criteria. Therefore, all boys of the present sample had an index sex offense charge, but adjudication of that charge could not be determined. Three individuals were excluded from this initial sample for the purposes of this study due to missing recidivism data. Pre-adolescent participants, defined as individuals below the age of 12 (n = 9), were also excluded for the purposes of the present study, so that the final sample consists of 71 juveniles. The mean age at the time of the accused index offense was 14.55 years (SD = 1.43; range = 12-17). Although the participants were accused for sexual offending, only those who were older than 14 years (76.1%, n = 54) would possibly convicted based on the minimum age of criminal liability according to the German Criminal Code. Table 1 demonstrates the index charges of these juveniles. The conviction history regards only those above 14 years due to the above-mentioned law. For a detailed sample description, see Table 1.

Table 1.

Sample Description.

	n = 71	%
Accused index offenses^a
Sexual coercion (rape or sexual assault)	31	43.7
Sexual child abuse	23	32.4
Aggravated child abuse	3	4.2
Abuse of persons who are incapable of resistance	4	5.6
Exhibitionism	1	1.4
Distribution of pornography	1	1.4
Sexual insult	10	14.1
Education level
Special education school	21	29.6
Lower secondary education	21	29.6
Divorced parents	44	62
Criminal and treatment history
Known to the public authorities because of sexual offenses/charges	2	2.8
Previous treatment because of sexual offenses	0	0
Convictions because of sexual offenses	0	0
Known to the public authorities because of violent offenses/charges	18	25.4
Previous treatment because of violent offenses	4	5.6
Convictions because of violent offenses	3	4.2
Migrant background	45	63.4

Multiple answers possible.

Instruments

SAPROF

The SAPROF (de Vogel et al., 2009) is a SPJ instrument that was designed to assess protective factors and should be used in combination with an established SPJ risk assessment instrument such as the Historical-Clinical-Risk Management–20 Violence Risk Assessment Scheme (HCR-20; Webster, Douglas, Eaves, & Hart, 1997) or the Sexual Violence Risk–20 (SVR-20; Boer, Hart, Kropp, & Webster, 1997). The protective factors in the SAPROF are defined as personal characteristics as well as environmental or situational factors that protect the individual offender from violent behavior. The SAPROF consists of 17 items, which are divided into the following three scales: Internal items are the historical and dynamic characteristics of an individual (intelligence, secure attachment in childhood, empathy, coping, self-control), motivational items are those that arise from the motivation to be a positive member of society (work, leisure activities, financial management, motivation for treatment, attitudes toward authority), and external items refer to beneficial environmental factors that offer protection from outside the individual (social network, intimate relationship, professional care, living circumstances, external control). All of the SAPROF items are dynamic except for two internal items (intelligence and secure attachment in childhood) and are rated on a 3-point scale: clearly present (2), partially present (1), and not present (0). The final protection judgment is determined by using the SPJ approach (e.g., Douglas & Reeves, 2010) in terms of weighting and prioritizing the protective factors rated by the clinician on a 3-point scale: high, moderate, or low. The SAPROF was developed primarily to assess future violence in adult forensic patients and not for sexual offenders, in particular. Therefore, the coding procedure of some items was adapted to adolescent circumstances (Klein et al., 2012). The item financial management contained dealing with pocket money and the original item work measured school achievement. The SAPROF items life goals and medication were excluded from analysis due to a high level of missing data (92.3%; 90.8%). Both items seem to be inherently inappropriate in the adolescent population. The concept of life goals implies the presence of a differentiated self-concept, which might still be immature in young people due to the dynamic process of adolescence. The poor applicability of the item medication is possibly due to the sample characteristics, because the present sample consisted of non-forensic pre-treatment juveniles.

SAVRY

The SAVRY (Borum et al., 2003) is a well-validated SPJ instrument that is frequently used in the adolescent population and was designed to assess the risk of future violence in young offenders aged between 12 and 18 years (Borum et al., 2003). The instrument includes 10 historical, 6 socio-contextual, and 8 individual risk factors. Like the SAPROF, all the risk factors are rated on a 3-point scale. A special characteristic of the SAVRY is the Protective Factor Scale. This scale includes six protective factors for violent offending (pro-social involvement, strong social support, strong attachments and bonds, positive attitude toward intervention and authority, strong commitment to school or work, resilient personality), which are rated dichotomously as either present or absent. The final risk judgment is made under due consideration of all risk and protective factors in the SPJ categories low, moderate, and high. In the present study, the Protective Factor Scale of the SAVRY was of particular interest. To examine the increment validity of the protective factors beyond the risk factors, the SAVRY risk scales were also integrated into the data analysis, even though the assessment of risk factors and the examination of the predictive validity of the SAVRY were not the primary aim of the present study.

Recidivism Data

Records from the German police information system (POLAS) were provided by the Hamburg Police Department and used to examine the youths’ recidivism. The POLAS is a regularly updated computer-aided database that contains information on the accused individual, the date, and elements of the criminal charges. It is important to note that the POLAS database contains criminal charges that are in the process of being investigated or prosecuted. These charges therefore represent an overestimation rather than an underestimation of the actual reoffense rates, as not all the charges lead to a conviction. No information about the rates of conviction at a later point in time was available. Thus, recidivism in the present study was defined as criminal charges and may therefore not be understood in the proper sense of offenses. However, previous research has repeatedly shown that charges provide a better measure of the actual reoffense rate compared with convictions (e.g., Quinsey, Harris, Rice, & Cormier, 2006; Rice, Harris, Lang, & Cormier, 2006). In addition, the POLAS data do not allow an accurate differentiation of the charges based on the criminal code. Therefore, in the present study, the POLAS information was classified into three recidivism categories: general recidivism, sexual recidivism, and violent recidivism. General recidivism was defined as any further criminal charge in the follow-up period (charges included, beside violent and sexual offenses, other types of delinquent behavior such as drug offenses, forgery, criminal property damage, and larceny). Sexual recidivism included charges for hands-on (sexual abuse, assault, or rape) and hands-off offenses (exhibitionism, voyeuristic acts). Furthermore, charges such as causing bodily harm and taking part in a brawl were added to the category violent recidivism (excluding sexual recidivism). The total follow-up period for the present study ranged from 30 to 68 months with a mean of 47.84 months (SD = 9.55). In total, 78.9% (n = 56) of the juveniles had police contact due to further criminal charges. During the follow-up period n = 10 juveniles (14.1%) were charged with sexual recidivism. The violent recidivism rate was 50.7% (n = 36).

Procedure and Data Analysis

Three psychologists rated the SAVRY and the SAPROF retrospectively, based on written summaries of the German version of the Basis Raads Onderzoek interviews (BARO, Doreleijers, Boonmann, van Loosbroek, & Vermeiren, 2011; Gutschner & Doreleijers, 2007) and additional clinical files from the FIT. The BARO is a semi-structured screening tool for the assessment of adolescent offenders at first contact and captures domains such as psychopathological and environment problems as well as criminal history. The interviews were conducted by social workers during a house call. The FIT files included a comprehensive assessment protocol regarding the criminological, psychological, and medical characteristics of the juvenile. The ratings were blind to outcome. Missing values were corrected by using a pro-rating procedure developed previously for adult risk assessment instruments by Quinsey et al. (2006) who proposed a pro-rating system for the Violence Risk Appraisal Guide (VRAG) and the Sex Offender Risk Appraisal Guide (SORAG), which consisted of the following steps: First, the highest possible score that could have been obtained on all available (i.e., non-missing) items was determined. Then, the proportion of possible points an offender obtained on those items was calculated. In the next step, the highest possible score an offender could have obtained on the missing items was determined. Afterward, the latter missing-items-score was multiplied by the proportion of possible positive points obtained on the items actually scored. In the last step, the resulting number was added to the previously obtained total score, to get the pro-rated final score. Inter-rater reliability for the same sample had been tested in a previous study (Klein et al., 2012) using the single measure interclass correlation coefficient (ICC) with the two-way random effects model absolute agreement type (McGraw & Wong, 1996). The inter-rater reliability of the SAPROF total score was ICC = .92 (internal scale, ICC = .86; motivational scale, ICC = .94; external scale, ICC = .92) and of SAVRY total score was ICC = .94 (historical scale, ICC = .96; individual scale, ICC = .96; social scale, ICC = .80; protective scale, ICC = .56). To determine the association between protective factors and recidivism, in a first step, Pearson’s r was calculated (see Cohen, 1992, for the critical values interpreting the effect sizes; r = .5 = large; r = .3 = moderate; r = .1 = small). Correlation analysis was also conducted to examine the relationship between the SAPROF sum score and the sum score of the SAVRY Protective Factors Scale. Receiver Operating Characteristics (ROC) was calculated to test the predictive accuracy of protective factors for the prediction of non-recidivism. ROC analyses are typically used in forensic research on risk assessment because of their low sensitivity to recidivism base rates (Rice & Harris, 1995). Values for the area under the curve (AUC) are one of the most commonly used measures for diagnostic and predictive accuracy measurements (Swets, Dawes, & Monahan, 2000). AUC values range from 0 to 1, whereby an AUC value of 1 represents a perfect predictive performance and a value of .5 indicates a prediction at chance level. The guidelines proposed by Rice and Harris (2005) were used to interpret the predictive accuracy values (AUC ≤ .63 = small, AUC between .64 and .71 = moderate, and AUC ≥ .72 = large effect size). The total scores and the items of the SAPROF and the SAVRY Protective Factor Scale were used as predictor variables. With regard to the SAPROF, the subscale total scores were also analyzed in their predictive accuracy. Non-recidivism was defined as the main outcome measure instead of recidivism, because, in this study, the protective factors are supposed to predict the absence of recidivism. Furthermore, Bonferroni-type adjustments were used to control for Type I error in multiple comparisons. The corrected significance level (α_corr) is reported.

The incremental validity of the protective factors beyond the SAVRY risk factors was examined using sequential regression models (e.g., Hunsley & Meyer, 2003; Rettenberger & Eher, 2013). A sequential Cox regression model with the SAVRY and SAPROF total and subscale scores as independent variables and recidivism as the dependent variable was selected for the present study. Because of the variability in the follow-up periods due to unequal time-at-risk periods, Cox regression was preferred over the logistic regression model. Cox regression is a particularly appropriate method for estimating the strength between an outcome and one or more predictor variables when using a data set with unequal follow-up periods (Eher, Matthes, Schilling, Haubner-MacLean, & Rettenberger, 2012; Hanson, 2006). SPSS version 18.0 was used for all steps of statistical analysis.

Results

The mean sum score of the SAPROF was 12.9 (SD = 4.46; range = 2-23) and the mean total score of the SAVRY Protective Scale was 2.83 (SD = 1.5; range = 0-6). The SAPROF sum score and the sum score of the SAVRY Protective Scale were positively correlated (r = .85, p < .01).

Table 2 shows the correlations between the SAPROF total score and the SAVRY Protective Factor Scale score with general, violent, and sexual recidivism. The SAPROF total score was negatively correlated with general and violent recidivism. In contrast, the SAVRY Protective Factor Scale showed no significant association with any recidivism categories in the present sample. In Table 3, the AUC values for the SAPROF total score, the SPJ final risk judgment, and all single protective factors included in the SAPROF are presented.

Table 2.

Correlations Between Recidivism and the SAPROF Total Score and the SAVRY Protective Factor Scale.

	The SAPROF total score	The SAVRY Protective Factor Scale
General recidivism	−.23*	−.14
Violent recidivism	−.26*	−.21
Sexual recidivism	−.19	−.14

Note. SAPROF = Structured Assessment of PROtective Factors; SAVRY = Structured Assessment of Violence Risk in Youth.

Correlation is significant at .05 level (one-tailed).

Table 3.

Predictive Validity of Protective Factors Measured by the SAPROF Regarding Absence of Recidivism.

	Absence of general recidivism			Absence of sexual recidivism			Absence of violent recidivism
	AUC	p	95% CI	AUC	p	95% CI	AUC	P	95% CI
Internal items	.68*	.03	[0.54, 0.82]	.68	.07	[0.47, 0.88]	.68**	.01	[0.55, 0.80]
Intelligence	.68*	.04	[0.50, 0.85]	.52	.82	[0.35, 0.70]	.64*	.05	[0.51, 0.77]
Secure attachment in childhood	.55	.58	[0.39, 0.70]	.54	.72	[0.34, 0.74]	.57	.33	[0.43, 0.70]
Empathy	.67*	.05	[0.52, 0.82]	.68	.06	[0.50, 0.87]	.56	.38	[0.43, 0.70]
Coping	.51	.93	[0.33, 0.69]	.66	.11	[0.50, 0.82]	.61	.11	[0.48, 0.74]
Self-control	.66	.06	[0.50, 0.82]	.60	.30	[0.39, 0.82]	.70**	.00	[0.58, 0.83]
Motivational items	.63	.12	[0.48, 0.79]	.62	.23	[0.42, 0.82]	.61	.12	[0.48, 0.74]
Work	.59	.31	[0.44, 0.74]	.51	.95	[0.35, 0.67]	.56	.42	[0.42, 0.69]
Leisure activities	.55	.54	[0.38, 0.72]	.56	.55	[0.38, 0.74]	.56	.43	[0.42, 0.69]
Financial management	.60	.27	[0.44, 0.75]	.56	.52	[0.38, 0.75]	.57	.36	[0.43, 0.70]
Motivation for treatment	.51	.92	[0.33, 0.69]	.53	.74	[0.32, 0.75]	.53	.62	[0.40, 0.67]
Attitudes toward authority	.59	.27	[0.45, 0.74]	.65	.14	[0.43, 0.86]	.58	.24	[0.45, 0.71]
External items	.50	.98	[0.33, 0.67]	.50	.96	[0.31, 0.68]	.53	.66	[0.40, 0.67]
Social network	.63	.14	[0.47, 0.78]	.64	.17	[0.48, 0.79]	.61	.10	[0.48, 0.74]
Intimate relationship	.50	1.0	[0.33, 0.67]	.57	.51	[0.39, 0.75]	.49	.91	[0.36, 0.63]
Professional care	.43	.43	[0.27, 0.59]	.43	.49	[0.23, 0.64]	.45	.49	[0.32, 0.59]
Living circumstances	.50	.98	[0.33, 0.66]	.48	.86	[0.29, 0.68]	.49	.85	[0.35, 0.62]
External control	.45	.53	[0.29, 0.60]	.44	.53	[0.24, 0.64]	.45	.43	[0.31, 0.58]
Final protection judgment	.60	.24	[0.45, 0.75]	.63	.18	[0.42, 0.85]	.64*	.04	[0.51, 0.77]
SAPROF sum score	.66	.06	[0.52, 0.80]	.65	.12	[0.45, 0.86]	.65*	.03	[0.52, 0.78]

Note. α_corr = .00083. SAPROF = Structured Assessment of PROtective Factors; AUC = area under the curve; CI = confidence interval.

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

The SAPROF sum score and the SAPROF–SPJ significantly predicted the absence of violent recidivism. Among the subscale scores, only the internal items showed significant predictive accuracy for the violent and general recidivism category. The other subscales showed no predictive validity. However, there were several single items with significant predictive power regarding different recidivism categories. Self-control and intelligence predicted absence of violent recidivism, and intelligence and empathy were the best predictors for general non-recidivism. Neither the SAPROF total score nor any single item reached statistical significance concerning the corrected significance level based on Bonferroni-type adjustments.

As demonstrated in Table 4, the SAVRY sum score, the SAVRY Historical Scale, and the SAVRY Individual Scale significantly predicted general and violent recidivism. To test the hypothesis that the protective factors measured by the SAPROF and the SAVRY provide incremental predictive accuracy beyond the SAVRY risk factors, Cox regression analyses were conducted. In the first analysis, the SAVRY risk factor scores were entered first, followed by the SAVRY protective factors, and in the second analysis, the SAPROF total scores were entered in the second block. Table 5 shows the results for sexual recidivism which indicate that the protective factors failed to reach any significant incremental predictive accuracy beyond what was captured by the SAVRY risk factors alone.¹

Table 4.

Predictive Validity of SAVRY Protective Factors and Risk Domains.

	Absence of recidivism			Absence of sexual recidivism			Absence of violent recidivism
	AUC	p	95% CI	AUC	p	95% CI	AUC	p	95% CI
Prosocial involvement	.53	.69	[0.37, 0.70]	.57	.50	[0.38, 0.76]	.51	.84	[0.38, 0.65]
Strong social support	.61	.20	[0.45, 0.77]	.55	.59	[0.36, 0.75]	.61	.13	[0.47, 0.74]
Strong attachments and bonds	.54	.67	[0.37, 0.70]	.49	.89	[0.29, 0.68]	.54	.59	[0.40, 0.67]
Positive attitude toward intervention and authority	.52	.79	[0.36, 0.69]	.62	.23	[0.42, 0.82]	.55	.43	[0.42, 0.69]
Strong commitment to school or work	.54	.60	[0.37, 0.71]	.56	.56	[0.38, 0.74]	.53	.65	[0.40, 0.67]
Resilient personality	.52	.84	[0.35, 0.68]	.52	.82	[0.33, 0.71]	.57	.29	[0.44, 0.71]
SAVRY Protective Factor Scale	.60	.22	[0.44, 0.77]	.63	.19	[0.44, 0.83]	.62	.08	[0.49, 0.76]
	General recidivism			Sexual recidivism			Violent recidivism
SAVRY Historical Scale	.74**	.00	[0.59, 0.89]	.66	.11	[0.50, 0.82]	.73***	.00	[0.61, 0.85]
SAVRY Social Scale	.56	.45	[0.38, 0.75]	.59	.37	[0.40, 0.78]	.61	.13	[0.47, 0.74]
SAVRY Individual Scale	.71**	.01	[0.57, 0.85]	.66	.12	[0.45, 0.86]	.71**	.00	[0.59, 0.83]
SAVRY sum score	.72**	.01	[0.58, 0.87]	.68	.06	[0.50, 0.86]	.73***	.00	[0.61, 0.85]

Note. α_corr SAVRY Protective Factors = .0024, α_corr SAVRY Risk Factors = .0041. SAVRY = Structured Assessment of Violence Risk in Youth; AUC = area under the curve; CI = confidence interval.

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

Table 5.

The Incremental Contribution of the SAVRY and SAPROF Protective Factors Beyond the SAVRY Risk Factors for the Prediction of Sexual Recidivism Using Cox Regression Analyses.

	χ² change			Regression coefficient				Rate ratio
	Change	df	p	B	SE	Wald	p	Exp(B)	95% CI
Step 1: SAVRY risk
SAVRY risk	2.33	1	.127	.065	.043	2.33	.127	1.07	[0.98, 1.16]
Step 2: SAVRY risk and SAVRY protection
SAVRY risk				.058	.051	1.27	.260	1.06	[0.96, 1.17]
SAVRY protection	0.08	1	.784	−.071	.259	0.08	.785	0.93	[0.56, 1.55]
Step 2: SAVRY risk and SAPROF
SAVRY risk				.040	.061	0.42	.518	1.04	[0.92, 1.17]
SAPROF	0.31	1	.580	−.057	.103	0.31	.581	0.95	[0.77, 1.16]

Note. N = 71. The values show incremental contribution of the protective factor scores after controlling for the previously entered risk factor scores. SAVRY = Structured Assessment of Violence Risk in Youth; SAPROF = Structured Assessment of PROtective Factors; CI = confidence interval.

Discussion

The findings of the current study support the results previously reported by Spice et al. (2013). In their study, no protective factor of the SAVRY was associated with sexual recidivism. In the present study, protective factors measured by the SAVRY did not reveal predictive accuracy regarding various types of recidivism. In contrast, the SAPROF protective factors in total showed moderate predictive validity. The internal scale of the SAPROF, in particular, yielded moderate predictive accuracy for the absence of violent and general recidivism, though not for the absence of sexual recidivism. It is important to note that neither the SAPROF total score nor any single item reached statistical significance related to the corrected significance level based on Bonferroni-type adjustments.

Furthermore, protective factors are not able to provide incremental predictive accuracy beyond the SAVRY risk factors in the present sample. This result might bring one to assume reduced clinical relevance for protective factors as defined in the present study. It could thus challenge the concept that the assessment of protective factors is particularly important with regard to young offenders to counteract the established risk-related and deficit-oriented evaluations (de Vogel et al., 2009; Miller, 2006; Rogers, 2000). Concerning the potential utility of the SAPROF for assessing protective factors, the present results suggest that more research is needed before using this instrument in juveniles who have sexually offended. In addition, one important practical implication of the present study is that further studies should determine the reliability and validity of the SAPROF items that require alterations for developmental reasons.

There is, therefore, a definite need for further studies on protective factors and its possible impact on treatment changes and recidivism among juveniles who sexually offended. Nevertheless, the consideration of protective factors in supervision- and treatment-related circumstances is certainly not only important for the prediction of recidivism but could also be relevant for assessing which factors are important enough to be incorporated into a case model and a supervision or treatment plan, even if they do not have incremental predictive accuracy beyond risk factors. A shift to a case model with protective factors in mind could improve not only the relationship between client and treatment provider but also the satisfaction of the treatment provider herself or himself. This has been shown to be the case for good lives model (GLM)–orientated treatment models (Harkins, Flak, Beech, & Woodhams, 2012).

Even if the results of the present study do not support the influence of protective factors on non-recidivism in accused juvenile who sexually offended, further protective factors such as intelligence and self-control deserve more attention in future research. Intelligence as a protective factor in a broader sense, which is also related to moral judgment and inhibition, has been suggested in previous research even before the importance of protective factors was suggested (Koolhof, Loeber, Wei, Pardini, & D’Escury, 2007; Stams et al., 2006; White, Moffitt, & Silva, 1989). In addition, the relevance of self-control as one of the decisive factors influencing criminal behavior has already been studied intensively and is theoretically well-founded, for example, by the General Theory of Crime proposed by Gottfredson and Hirschi (1990). Studies in the area have, for example, supported the assumption that a poor level of self-control in children predicts criminal offenses (Moffitt et al., 2011).

The present sample exhibits several characteristics consistent with numerous studies on recidivism among juveniles who sexually offended (Caldwell, 2010; Carpentier & Proulx, 2011; McCann & Lussier, 2008; Zimring, Piquero, & Jennings, 2007). Most of the accused juveniles in the present study were charged again because of a general or a violent crime in the follow-up period, whereas only a much smaller proportion of juveniles were reported due to a sexual offense. Thus, in alignment with the previous suggestions regarding a distinction between generalist and specialist in the population of sexual offenders (Pullman & Seto, 2012; Seto & Lalumière, 2010), the juveniles accused for sexual offending seem more likely to persist with non-sexual than with sexual problematic behavior. In the consequence, one further practical implication suggested by this study is that the SAVRY constitutes a valuable measure for examining the risk for violent and general criminal behavior in juveniles who are at risk of sexual offending.

There are several limitations to the current study. First of all, the sample included juveniles who were accused for sexual offending, who had merely been criminally charged by the police or prosecution office but not necessarily convicted. Furthermore, no youth in the present sample had a prior conviction for a sexual offense. Similar to the index crime, the recidivism criteria were operationalized as new criminal charges instead of convictions within a follow-up period of approximately 4 years. This could, however, be an advantage of this study, because it delivers an insight into young accused juveniles—the mean age of the present sample is about 14 years ranging from 12 to 17—who have been identified in an early stage of deviant behavior as Butler and Seto (2002) recommend. In addition, even those adolescents who are known to the authorities would generally be sentenced in only the rarest cases (Caldwell, 2010; Vizard, 2007). Hence, research with recidivism data on a low threshold level such as the police records, in this study, could demonstrate a methodological strength as it enables early detection and possible intervention in problem behaviors before they result in more severe legal consequences. However, the juveniles in this sample might not have been convicted eventually or might have desisted after adolescence. This could not be verified due to the rather short follow-up period in relation to the young initial age. The need for prospective study designs in research on risk assessment tools was strongly emphasized by Worling et al. (2012). The retrospective rating of the instruments based on file information is therefore a further methodical limitation of the present study. Another relevant limitation of the present study is the generally small sample size and the precision of ROC-related indices of accuracy in such small samples (Hanczar et al., 2010; Hill et al., 2012). Even if the use of the AUC metric is still referred to as the gold standard method for assessing predictive accuracy (e.g., Lobo, Jiménez-Valverde, & Real, 2008; Mossman, 1994; Rice & Harris, 1995, 2005), its accuracy has to be interpreted cautiously, especially in relatively small samples with considerably low base rates (e.g., Eher, Rettenberger, Schilling, & Pfäfflin, 2008; Hill et al., 2012). Nevertheless, until now, no solution has been made available for this problem (Hanczar et al., 2010). Furthermore, none of the indices of accuracy commonly used in the past provide an alternative means of evaluating the predictive validity of risk assessment instruments (Rice & Harris, 1995). In consideration of the effect sizes reported in the present study, one could anticipate that the results would reach statistical significance by using a larger sample size: For example, the SAVRY sum score (AUC = .68, p = .06) as well as the sum score of the SAPROF internal (AUC = .68, p = .07) items yielded both moderate effect sizes (Rice & Harris, 2005), which failed marginally the threshold for statistical significance. In this context, the present study could probably be regarded as an example for an underpowered study due to a sample size (Cohen, 1962).

In the present study, protective factors were treated according to the definition of the SAPROF and the SAVRY as well as to Spice et al. (2013) as positive and distinct entities and therefore not simply as the absence of risk factors. The SAPROF was developed as a positive, dynamic, and treatment-focused SPJ-assessment instrument. Positive in this sense means counterbalancing a risk-oriented approach by the use of theory, research, and interventions focusing on helpful institutions, positive individual traits, emotions, and behaviors to prevent reoffending. Nevertheless, risk and protective factors could also be associated independently to recidivism in opposite directions, in which case, protective factors could be present or absent at any estimated level of risk of reoffending and thus buffer a person against risk. The present study focused on analyses concerned direct effects of protective factors and not on interaction effects. Future research efforts should address interaction effects, which could indicate the “buffering” effect that protective factors might have on the association between risk factors and types of recidivism. In conclusion, it would appear that there is a critical need for further studies to clarify the function of protective factors regarding recidivism risk. Nevertheless, protective factors may be especially important in changing attitudes and strengthening the relationships between clients, institutions, supervisors, and treatment providers, as well as in preserving their motivation and engagement.

Footnotes

Authors’ Note

Verena Klein and Martin Rettenberger contributed equally to this article.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a grant from the Behörde für Soziales, Familie, Gesundheit und Verbraucherschutz, City of Hamburg, Germany.

Notes

References

Blum

R. W.

Ireland

(2004). Reducing risk, increasing protective factors: Findings from the Caribbean Youth Health Survey. Journal of Adolescent Health, 35, 493-500. doi:10.1016/j.jadohealth.2004.01.009

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. Burnaby, British Columbia, Canada: Mental Health, Law, and Policy Institute Simon Fraser University.

Borowsky

I. W.

Hogan

Ireland

(1997). Adolescent sexual aggression: Risk and protective factors. Pediatrics, 100, Article e7, 1-8. doi:10.1542/peds.100.6.e7

Borum

Bartel

P. A.

Forth

A. E.

(2003). SAVRY: Structured Assessment of Violence Risk in Youth (Version 1.1). Tampa: University of South Florida.

Busch

K. G.

Zagar

R. J.

Grove

W. M.

Hughes

J. R.

Arbit

Bussell

R. E.

Bartikowski

(2009). Looking forward in records of young adults convicted of sexual homicide, rape, or molestation as youth: Risks for reoffending. Psychological Reports, 104, 155-184. doi:10.2466/pr0.104.1.155-184

Butler

S. M.

Seto

M. C.

(2002). Distinguishing two types of adolescent sex offenders. Journal of the American Academy of Child & Adolescent Psychiatry, 41, 83-90. doi:10.1097/00004583-200201000-00015

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. doi:10.1177/0306624X08330016

Caldwell

M. F.

(2013). Accuracy of sexually violent person assessments of juveniles adjudicated for sexual offenses. Sexual Abuse: A Journal of Research and Treatment, 25, 516-526. doi:10.1177/1079063213480818

Carpentier

Proulx

(2011). Correlates of recidivism among adolescents who have sexually offended. Sexual Abuse: A Journal of Research and Treatment, 23, 434-455. doi:10.1177/1079063211409950

10.

Chu

C. M.

Fong

Teoh

(2012). Assessing youth who sexually offended: The predictive validity of the ERASOR, J-SOAP-II, and YLS/CMI in a non-Western context. Sexual Abuse: A Journal of Research and Treatment, 24, 153-174. doi:10.1177/1079063211404250

11.

Cohen

(1962). The statistical power of abnormal-social psychological research: A review. The Journal of Abnormal and Social Psychology, 65, 145-153. doi:10.1037/h0045186

12.

Cohen

(1992). A power primer. Psychological Bulletin, 112, 155-159. doi:10.1037/0033-2909.112.1.155

13.

Costa

F. M.

Jessor

Turbin

M. S.

(1999). Transition into adolescent problem drinking: The role of psychosocial risk and protective factors. Journal of Studies on Alcohol, 60, 480-490.

14.

de Vogel

de Ruiter

Bouman

Y. H. A.

de Vries Robbé

(2009). SAPROF: Guidelines for the assessment of protective factors for violence risk. Utrecht, The Netherlands: Forum Educatief.

15.

Doreleijers

T. A.

Boonmann

van Loosbroek

Vermeiren

(2011). Assessing the psychometric properties and the perceived usefulness of the BasisRaadsOnderzoek (BARO) as a first-line screening instrument for juvenile offenders. Child & Adolescent Psychiatry & Mental Health, 5, Article 24. doi:10.1186/1753-2000-5-24

16.

Douglas

K. S.

Reeves

K. A.

(2010). Historical-Clinical-Risk Management-20 (HCR-20) Violence Risk Assessment Scheme: Rationale, application, and empirical overview. In Otto

R. K.

Douglas

K. S.

(Eds.), Handbook of violence risk assessment (pp. 147-185). Oxford, UK: Routledge.

17.

Driemeyer

Spehr

Yoon

Richter-Appelt

Briken

(2013). Comparing sexuality, aggressiveness, and antisocial behavior of alleged juvenile sexual and violent offenders. Journal of Forensic Sciences, 58, 711-718. doi:10.1111/1556-4029.12086

18.

Eher

Matthes

Schilling

Haubner-MacLean

Rettenberger

(2012). Dynamic risk assessment in sexual offenders using STABLE-2000 and the STABLE-2007: An investigation of predictive and incremental validity. Sexual Abuse: A Journal of Research and Treatment, 24, 5-28. doi:10.1177/1079063211403164

19.

Eher

Rettenberger

Schilling

Pfäfflin

(2008). Failure of Static-99 and SORAG to predict relevant reoffense categories in relevant sexual offender subtypes: A prospective study. Sexual Offender Treatment, 3, 1-14. Retrieved from http://www.sexual-offender-treatment.org/1-2008_02.html

20.

Fanniff

A. M.

Letourneau

E. J.

(2012). Another piece of the puzzle: Psychometric properties of the J-SOAP-II. Sexual Abuse: A Journal of Research and Treatment, 24, 378-408. doi:10.1177/1079063211431842

21.

Fitzpatrick

K. M.

(1997). Fighting among America’s youth: A risk and protective factors approach. Journal of Health and Social Behavior, 38, 131-148. doi:10.2307/2955421

22.

Gottfredson

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

23.

Gutschner

Doreleijers

T. A. H.

(2007). BARO.de. Screeninginstrument zur Erstbeurteilung von jugendlichen Straftätern [Initial evaluation of juvenile offenders. The screening instrument BARO]. Bern, Switzerland: Institut für forensische Kinder- und Jugendpsychologie.

24.

Hanczar

Hua

Sima

Weinstein

Bittner

Dougherty

E. R.

(2010). Small-sample precision of ROC-related estimates. Bioinformatics, 26, 822-830. doi:10.1093/bioinformatics/btq037

25.

Hanson

R. K.

(2006). Does STATIC-99 predict recidivism among older sexual offenders? Sexual Abuse: A Journal of Research and Treatment, 18, 343-355. doi:10.1007/s11194-006-9027-y

26.

Harkins

Flak

Beech

A. R.

Woodhams

(2012). Evaluation of a community-based sex offender treatment program using a good lives model approach. Sexual Abuse: A Journal of Research and Treatment, 24, 519-543. doi:10.1177/1079063211429469

27.

Hempel

Buck

Cima

van Marle

(2013). Review of risk assessment instruments for juvenile sex offenders: What is next? International Journal of Offender Therapy and Comparative Criminology, 57, 208-228. doi:10.1177/0306624X11428315

28.

Hill

Rettenberger

Habermann

Berner

Eher

Briken

(2012). The utility of risk assessment instruments for the prediction of recidivism in sexual homicide perpetrators. Journal of Interpersonal Violence, 27, 3553-3578. doi:10.1177/0886260512447570

29.

Hoge

R. D.

Andrews

D. A.

Leschied

L. W.

(1996). An investigation of risk and protective factors in a sample of youthful offenders. Journal of Child Psychology and Psychiatry and Allied Disciplines, 37, 419-424. doi:10.1111/j.1469-7610.1996.tb01422.x

30.

Hunsley

Meyer

G. J.

(2003). The incremental validity of psychological testing and assessment: Conceptual, methodological, and statistical issues. Psychological Assessment, 15, 446-455. doi:10.1037/1040-3590.15.4.446

31.

Jessor

(1991). Risk behavior in adolescence: A psychosocial framework for understanding and action. Journal of Adolescent Health, 12, 597-605.

32.

Jessor

(2013, January). Resilience. Paper presented at the European Winterschool Novel Insights into Antisocial Behavior, Monastery Seeon, Germany.

33.

Klein

Yoon

Briken

Turner

Spehr

Rettenberger

(2012). Assessment of accused juvenile sex offenders in Germany: A comparison of five different measures. Behavioral Sciences & the Law, 30, 181-195. doi:10.1002/bsl.2006

34.

Koolhof

Loeber

Wei

E. H.

Pardini

D’Escury

A. C.

(2007). Inhibition deficits of serious delinquent boys of low intelligence. Criminal Behavior and Mental Health, 17, 274-292. doi:10.1002/cbm.661

35.

Laub

J. H.

Sampson

R. J.

(2001). Understanding desistance from crime. In Tonry

(Ed.), Crime and justice: A review of research (Vol. 28, pp. 1-70). Chicago, IL: University of Chicago Press.

36.

Lobo

J. M.

Jiménez-Valverde

Real

(2008). AUC: A misleading measure of the performance of predictive distribution models. Global Ecology and Biogeography, 17, 145-151. doi:10.1111/j.1466-8238.2007.00358.x

37.

Lodewijks

H. P. B.

de Ruiter

Doreleijers

T. A. H.

(2010). The impact of protective factors in desistance from violent reoffending: A study in three samples of adolescent offenders. Journal of Interpersonal Violence, 25, 568-587. doi:10.1177/0886260509334403

38.

Loeber

Farrington

D. P.

(2012). Advancing knowledge about direct protective factors that may reduce youth violence. American Journal of Preventive Medicine, 43, 24-27. doi:10.1016/j.amepre.2012.04.031

39.

Masten

Garmezy

(1985). Risk, vulnerability, and protective factors in developmental psychopathology. In Lahey

Kazdin

(Eds.), Advances in clinical child psychology (Vol. 8, pp. 1-52). New York, NY: Plenum Press.

40.

McCann

Lussier

(2008). Antisociality, sexual deviance, and sexual reoffending in juvenile sex offenders: A meta-analytic investigation. Youth Violence and Juvenile Justice, 6, 363-385. doi:10.1177/1541204008320260

41.

McGraw

K. O.

Wong

S. P.

(1996). Forming inferences about some intraclass correlation coefficients. Psychological Methods, 1, 30-46. doi:10.1037/1082-989X.1.4.390

42.

Miller

H. A.

(2006). A dynamic assessment of offender risk, needs, and strengths in a sample of pre-release general offenders. Behavioral Sciences & the Law, 24, 767-782. doi:10.1002/bsl.728

43.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. doi:10.1037/0033-295X.100.4.674

44.

Moffitt

T. E.

(2003). Life-course persistent and adolescent limited antisocial behavior: A ten year research review and research agenda. In Lahey

B. B.

Moffitt

T. E.

Caspi

(Eds.), Causes of conduct disorder and juvenile delinquency (pp. 49-75). New York, NY: Guilford Press.

45.

Moffitt

T. E.

Arseneault

Belsky

Dickson

Hancox

R. J.

Harrington

. . .Caspi

(2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences, 108, 2693-2698. doi:10.1073/pnas.1010076108

46.

Mossman

(1994). Assessing predictions of violence: Being accurate about accuracy. Journal of Consulting and Clinical Psychology, 62, 783-792. doi:10.1037/0022-006X.62.4.783

47.

Pullman

Seto

M. C.

(2012). Assessment and treatment of adolescent sexual offenders: Implications of recent research on generalist versus specialist explanations. Child Abuse & Neglect, 36, 203-209. doi:10.1016/j.chiabu.2011.11.003

48.

Quinsey

V. L.

Harris

G. T.

Rice

M. E.

Cormier

(2006). Violent offenders: Appraising and managing risk. Washington, DC: American Psychological Association.

49.

Rennie

C. E.

Dolan

M. C.

(2010). The significance of protective factors in the assessment of risk. Criminal Behavior and Mental Health, 20, 8-22. doi:10.1002/cbm.750

50.

Resnick

M. D.

Ireland

Borowsky

(2004). Youth violence perpetration: What protects? What predicts? Findings from the national longitudinal study of adolescent health. Journal of Adolescent Health, 35, 424.e1-424.e10. doi:10.1016/j.jadohealth.2004.01.011

51.

Rettenberger

Eher

(2013). Actuarial risk assessment in sexually motivated intimate-partner violence. Law and Human Behavior, 37, 75-86. doi:10.1037/b0000001

52.

Rice

M. E.

Harris

G. T.

(1995). Violent recidivism: Assessing predictive accuracy. Journal of Consulting and Clinical Psychology, 63, 737-748. doi:10.1037/0022-006X.63.5.737

53.

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. doi:10.1007/s10979-005-6832-7

54.

Rice

M. E.

Harris

G. T.

Lang

Cormier

C. A.

(2006). Violent sex offenses: How are they best measured from official records? Law and Human Behavior, 30, 525-541. doi:10.1007/s10979-006-9022-3

55.

Rogers

(2000). The uncritical acceptance of risk assessment in forensic practice. Law and Human Behavior, 24, 595-605. doi:10.1023/A:1005575113507

56.

Rutter

(1985). Resilience in the face of adversity: Protective factors and resistance to psychiatric disorder. British Journal of Psychiatry, 147, 598-611. doi:10.1192/bjp.147.6.598

57.

Salekin

R. T.

Lee

Schrum-Dillard

C. L.

Kubak

F. A.

(2010). Child psychopathy and protective factors: IQ and motivation to change. Psychology, Public Policy, and Law, 16, 158-176. doi:10.1037/a0019233

58.

Sampson

R. J.

Laub

J. H.

(1993). Crime in the making: Pathways and turnings points through life. Cambridge, UK: Harvard University Press.

59.

Seto

M. C.

Lalumière

M. L.

(2010). What is so special about male adolescent sexual offending? A review and test of explanations using meta-analysis. Psychological Bulletin, 136, 526-575. doi:10.1037/a0019700

60.

Spehr

Yoon

Briken

(2010). Sexuell auffällige Minderjährige [Sexually striking minors]. Zeitschrift für Sexualforschung, 23, 139-154. doi:10.1055/s-0030-1247412

61.

Spice

Viljoen

J. L.

Latzman

N. E.

Scalora

M. J.

Ullman

(2013). Risk and protective factors for recidivism among juveniles who have offended sexually. Sexual Abuse: A Journal of Research and Treatment, 25, 347-369. doi:10.1177/1079063212459086

62.

Stams

G. J.

Brugman

Deković

van Rosmalen

van der Laan

Gibbs

J. C.

(2006). The moral judgment of juvenile delinquents: A meta-analysis. Journal of Abnormal Child Psychology, 34, 697-713. doi:10.1007/s10802-006-9056-5

63.

Stouthamer-Loeber

Loeber

Stallings

Lacourse

(2008). Desistance from and persistence in offending. In Loeber

Farrington

D. P.

Stouthamer-Loeber

Raskin White

(Eds.), Violence and serious theft: Development and prediction from childhood to adulthood (pp. 269-398). New York, NY: Routledge.

64.

Stouthamer-Loeber

Wei

Loeber

Masten

A. S.

(2004). Desistance from persistent serious delinquency in the transition to adulthood. Development and Psychopathology, 16, 897-918. doi:10.1017/S0954579404040064

65.

Swets

J. A.

Dawes

R. M.

Monahan

(2000). Psychological science can improve diagnostic decisions. Psychological Science in the Public Interest: A Journal of the American Psychological Society, 1, 1-26. doi:10.1111/1529-1006.001

66.

Ullrich

Coid

(2011). Protective factors for violence among released prisoners—Effects over time and interactions with static risk. Journal of Consulting and Clinical Psychology, 79, 381-390. doi:10.1037/a0023613

67.

Viljoen

J. L.

Mordell

Beneteau

J. L.

(2012). Prediction of adolescent sexual reoffending: A meta-analysis of the J-SOAP-II, ERASOR, J-SORRAT-II, and Static-99. Law and Human Behavior, 36, 423-438. doi:10.1037/h0093938

68.

Vizard

(2007). Adolescent sexual offenders. Psychiatry, 6, 433-437.

69.

Webster

C. D.

Douglas

Eaves

Hart

(1997). HCR-20: Assessing risk for violence (Version 2). Burnaby, British Columbia, Canada: Simon Fraser University.

70.

White

J. L.

Moffitt

T. E.

Silva

P. A.

(1989). A prospective replication of the protective effects of IQ in subjects at high risk for juvenile delinquency. Journal of Consulting and Clinical Psychology, 57, 719-724.

71.

Worling

J. R.

Bookalam

Litteljohn

(2012). Prospective validity of the Estimate of Risk of Adolescent Sexual Offense Recidivism (ERASOR). Sexual Abuse: A Journal of Research and Treatment, 24, 203-223. doi:10.1177/1079063211407080

72.

Zagar

R. J.

Busch

K. G.

Grove

W. M.

Hughes

J. R.

(2009). Summary of studies of abused infants and children later homicidal, and homicidal, assaulting later homicidal, and sexual homicidal youth and adults. Psychological Reports, 104, 17-45. doi:10.2466/pr0.104.1.129-154

73.

Zimring

F. E.

Piquero

A. R.

Jennings

W. G.

(2007). Sexual delinquency in racine: Does early sex offending predict later sex offending in youth and young adulthood? Criminology & Public Policy, 6, 507-534. doi:10.1111/j.1745-9133.2007.00451.x