A meta-analysis into recidivism by juvenile sexual offenders : individual, interpersonal and socio-contextual risk factors

(1)

RUNNING HEAD: A META-ANALYSIS INTO RECIDIVISM BY JUVENILE SEXUAL OFFENDERS

A Meta-Analysis into Recidivism by Juvenile Sexual Offenders

Individual, interpersonal and socio-contextual risk factors

Master Scriptie Forensische Orthopedagogiek

Graduate School of Child Development and Education Universiteit van Amsterdam

A.L. Carslaw 10787968

Begeleiding: Dr. Daphne van de Bongardt Tweede beoordelaar: Drs. Marita van Langen Amsterdam, 13 december 2015

(2)

A META-ANALYSIS INTO RECIDIVISM BY JUVENILE SEXUAL OFFENDERS

2 Abstract

These meta-analyses determined risk factors for recidivism by juvenile sexual offenders (JSOs), between 12 and 21 years old. The sexual risk factors, peers and community subcategories had the largest effect, whereas individual, personal, general deviancy, family and parental risk factors had the smallest effect on JSOs recidivating. Individual and interpersonal risk factors had a medium effect on recidivism in comparison to the other risk factor categories. The personal and sexual risk factors had a larger effect on Caucasian JSOs, whereas general deviancy risk factors had a greater effect on samples with mixed ethnicities. The older the offender, the more influence the community and general socio-contextual risk factors had on the chance of JSOs recidivating, with the exception of risk factors connected to their family and parents. These findings could be important when developing interventions, choosing an intervention for a certain JSO, and also give insight into possibilities for future research.

Keywords

(3)

3

A Meta-Analysis into Recidivism by Juvenile Sexual Offenders: Individual, Interpersonal and Socio-Contextual Risk Factors

A sexual offence is defined by Terry (2012) as sexual acts enforced on the victim, without their consent or without valid consent (for example with children under the age of 18 or disabled people). This includes hands-on sexual behaviour like rape, assault and the touching of genitalia. It also includes hands-off sexual behaviour like voyeurism or exposing oneself, viewing, possessing or creating child pornography or sexual solicitation or trafficking. Knowledge about risk factors for committing such an offence is still limited. In particular, research specialising in juvenile sexual offenders (JSOs), between the ages of 12 and 21 years old, is scarce. The screening of risk groups could help the prevention of sexual offences. Once an offence has been committed, however, sexual offenders could receive an intervention to decrease the effect of risk factors for recidivating. When a sexual offender commits another sexual offence, this is recidivism. In this meta-analysis, recidivism refers solely to recidivism while still of adolescent age (between 12 and 21 years old), unless specified otherwise. The crime does not necessarily need to be the exact same offence, as long as the offence concerned is of a sexual nature.

Prevalence Different Forms of Sexual Offences in Different Countries

The importance of gathering information about sexual offences becomes clear when one views the prevalence statistics of these crimes. In the United Kingdom, 6% of women fall victim to a sexual offence in their lifetime (Office for National Statistics, 2015). In the United States, one in twenty women and men experience some form of sexual violence besides rape and one in every five women and one in 71 men experience rape (Walters, Chen, & Breiding, 2013).In 2011, 8% of Dutch men and 33% of Dutch women older than 15, have experienced

(4)

4

sexual violence (De Haas, 2012). Every year 1% of men and women in the Netherlands are victims of sexual crime (De Haas, 2012).

Prevalence Sexual Offences and Recidivism by JSOs

In the United States, 13% of all arrests for rape concern arrests of adolescent perpetrators between the age of 12 and 21 (United States Department of Justice, 2009). Out of all arrests following non-rape sexual offences, 14% are arrests of JSOs (United States Department of Justice, 2009).

According to Lussier and Cale (2013), 10% of JSOs will continue committing sexual offences into adulthood. Others discontinue offending naturally as they become older and their lives naturally change, because they make other friends, start a relationship or get a job that keeps them from misbehaving (Lussier and Cale, 2013). Roughly 7% of JSOs recidivate with another sexual offences within five years or before adulthood (Gerhold, Browne, & Beckett, 2007). This proportion increases to 28% in studies where the follow-up time is greater, or when self-reports are used in addition to data on actual convictions (Gerhold, Browne, & Beckett, 2007). However, these recidivating JSOs account for 70% of all rapes commited by JSOs, and therefore are a very significant group to focus research on (Tracy, Wolfgang, & Figlio, 1990).

Consequences of a Sexual Offence for Victims and Offenders

After a sexual offence has been committed, this can have a great impact, on both the victim and the perpetrator. Victims of sexual assault are generally more susceptible to depression, post-traumatic stress disorder and anxiety after they have fallen victim to a sexual offence (Ullman & Filipas, 2001). In addition, according to Mason and Lodrick (2013), victims can experience trust issues, have problems with intimacy or will no longer engage in

(5)

5

sexual acts. They have a higher tendency to develop emotional problems, substance abuse, suicidal thoughts and behaviour and experience low self-esteem (Mason, & Lodrick, 2013).

Sexual offenders also experience negative consequences of committing an offence as well. A short-term consequence is the possible penalty they will receive after committing the offence, for example a fine or community service. Some JSOs face incarceration, which will increase their chance of unemployment, a lower income and future arrests (Freudenberg, et al., 2005). According to Piquero, Farrington, & Blumstein (2007), an early age of onset of sexual offending has been found to be a predictor for long-term general offender behaviour. Up to 30% of JSOs commit a non-sexual offence within six months after their sexual offence (Chui, & Chan, 2012). These non-sexual offences increase their chance of coming into contact with the justice system, and increase the amount and severity of their penalties. When adolescents have been in contact with the justice system, they have an increased risk of suicide as adults (Webb, et al., 2012).

Theoretical Framework for Sexual Recidivism: Generalist and Specialist Perspective Risk factors for recidivism should be determined to be able to provide effective intervention strategies for JSOs and therefore prevent JSOs recidivating. Adolescent sexual offending and recidivism can be explained using two main theoretical perspectives, the generalist perspective and the specialist perspective (Pullman, & Seto, 2012). The generalist perspective proposes that sexual offences are merely a part of general offending behaviour. Sexual offending is an act induced by general delinquent tendencies, similar to stealing or violence. As such, JSOs would share risk factors with adolescent non-sex offenders.

On the other hand, the specialist perspective proposes that the risk factors explaining general adolescent recidivism do not explain adolescent sexual recidivism. This perspective assumes that a JSO recidivates with a sexual crime instead of a general crime, due to specific risk factors.

(6)

6

Both perspectives are backed up by empirical research (Pullman, & Seto, 2012). Some papers find the same risk factors for general offenders as sexual offenders. Other papers conclude that certain risk factors will only predict one kind of offence and not the other (Pullman, & Seto, 2012). If it is assumed that all JSOs, both those recidivating sexually and those recidivating generally, share the same risk factors, papers including all offender types could be analysed to determine the risk factors for sexual offenders. However, if the specialist perspective seems more relevant in this case than the generalist perspective, the risk factors found in this meta-analysis might be wrongly generalised to all offenders. Additionally, the overlapping risk factors could be explained by the fact that some sexual offenders will also recidivate generally, and therefore show the same risk factor patterns as general recidivists (Pullman, & Seto, 2012). Thus, this for meta-analysis, only paper concerning JSOs recidivating sexually will be analysed.

Theoretical Framework for Risk Factors: The Ecological Perspective by Bronfenbrenner

Prior research has provided various risk factors for JSOs recidivating. It is difficult to create interventions that focus on the many risk factors based on detailed characteristics of the JSOs. To increase the practical implementation of the results from this meta-analysis, it is desirable to categorise these risk factors.

According to Bronfenbrenner’s ecological model, the development of children (and JSOs) is influenced both by their individual characteristics as well as their social surroundings and their relations with various social actors in their social environment. The risk factors for JSOs to recidivate can also be categorised into these three groups: individual factors, interpersonal factors and social-contextual factors (Bronfenbrenner, 1979).

(7)

7

Individual factors are variables associated with the perpetrator him/herself, including: gender, personality, psychiatric or psychological issues, age and education. Interpersonal factors are variables describing the interaction between the perpetrator and other people. This includes the perpetrator's connection to and relationship with the victim. Socio-contextual factors are variables within the environment of the perpetrator, such as socio-economic status, peers, neighbourhood and environment supporting opportunities to recidivate sexually (Bronfenbrenner, 1979). As shown in Figure 1, this meta-analysis will focus specifically on individual risk factors, with personality characteristics, sexual tendencies and general deviancy as subcategories. In addition, it will focus on interpersonal risk factors. Finally, it will focus on socio-contextual risk factors, with family and parental risk factors, risk factors in peer characteristics and community risk factors as subcategories.

Meta-Analyses

A meta-analysis uses all known papers on a certain subject as its data sample. It increases the numbers of observations and the statistical power of a risk factor in comparison to individual papers. It therefore improves the estimates of the effect size of the found risk factors (Fagard, Staessen, & Thijs, 1996). In addition, a meta-analysis collects all risk factors into one database, to create as complete a list of risk factors for adolescent sexual recidivism as thus far possible. Hence, in this present study, a meta-analysis will be conducted.

Prior Meta-Analytic Research and Reviews into Risk Factors for General Recidivism Prior meta-analyses have provided risk factors for JSOs to recidvate generally. According to the generalist perspective (Seto, & Pullman, 2010), this information offers an insight into risk factors for JSOs to recidivate sexually. A meta-analysis by Van der Put, et al. (2013) showed the individual risk factors; alcohol- and drugs abuse, attitudes and aggressive

(8)

8

behaviour, as risk factors for general recidivism. Interpersonal risk factors connected to an increased chance of recidivating generally are relationships and family problems. No contextual risk factors were found (Van der Put, et al., 2013).

Prior Research into Risk Factors for Adult Sexual Recidivism

The specialist perspective proposed that prior research into JSOs recidivating sexually, during adulthood, could offer an insight into risk factors for JSOs recidvating sexually during adolescence. Based this prior research, it is hypothesised that individual and interpersonal risk factors increase the chance of adolescent sexual recidivism the most, while socio-contextual risk factors have the smallest effect on the chance of recidivism.

A systematic review by Gerhold, Browne and Beckett (2007) described risk factors for JSOs to recidivate sexually as adults. They determined one individual factor was the most related to recidivism; having been a victim of sexual abuse during childhood. However, individual factors of the victim, a child and male victim, were also connected to an increased chance of sexually recidivating. Interpersonal risk factors were also found; characteristics of the crime (how many prior sexual crimes, location of the crime), the relationship with the victims (how well the offender knows the victim, number of victims), blaming the victim and the use of verbal threats and general inter- and intrapersonal functioning. No social-contextual factors were found (Gerhold, Browne and Beckett, 2007).

Earlier research has also been performed into sexual recidivism, focusing both on adolescent and adult re-offenders (Hanson, & Morton-Bourgon, 2005). They concluded that the individual risk factor antisocial orientation and the sexual risk factor deviant sexual preferences had the greatest effect on JSOs recidivating.

There seems to be a lack of research into sexual recidivism, in particular sexual recidivism during adolescence. This meta-analysis will be, as far as the author is aware, the

(9)

9

first meta-analysis focussing on adolescent sexual recidivism during adolescence. This will produce information that could be valuable for treatment and consequently, the prevention of sexual recidivism. In addition, it would provide insight into options for further research.

Treatment of Sexual Offenders and Prevention of Sexual Recidivism

It is important to provide an intervention for JSOs, to prevent recidivism. This will not only help reduce the chance of people becoming victims, but also prevent possible recidivating offenders from living with the legal and psychological consequences of being a life-long offender. However, according to Hanson and Morton-Bourgon (2005), the treatments most commonly provided for JSOs, do not focus on the risk factors with the largest effect on recidivism.

According to The National Adolescent Perpetrator Network (1988), an effective treatment for JSOs needs to focus on the most significant risk factors for recidivating. They determined that treatment created for JSOs should address individual risk factors, including personality characteristics (accepting responsibility without victimising oneself, feeling empathy, power and control issues), sexual risk factors (decreasing deviant arousal and negative sexual identity) and general deviancy risk factors (addictive and compulsive qualities). It should also address interpersonal risk factors like a lack of social skills. Socio-contextual risk factors should also be included in the treatment of JSOs, including family dysfunction, peer interaction and restitution of the community (National Adolescent Perpetrator Network, 1988).

JSOs can currently receive a number of treatments, to prevent recidivism. They contain ways to reduce a number of risk factors, recommended by the National Adolescent Perpetrator Network (1988). Currently, most treatments focus merely on individual risk factors, mostly those that are of a personal or sexual nature (Efta-Breitbach, & Freeman,

(10)

10

2005). Many additional risk factors that are recommended to be part of the intervention, like certain interpersonal and socio-contextual risk factors, rarely are.

Firstly, an example of a JSO intervention is Cognitive Behavioural Therapy (Maletzky, & Steinhauser, 2002), focussing mostly on individual risk factors. It teaches JSOs skills to change attitudes and cognitive distortions and reducing arousal and impulsive behaviour. In addition, it also teaches the JSOs to feel empathy, which is an interpersonal factor (Efta-Breitbach, & Freeman, 2005). Secondly, psycho education (Todoroska-Gjurchevska, Naumovska, & Kozarev, 2015), an intervention focussing on creating insight for the JSOs on their thoughts and behaviour. It focusses mainly on individual risk factors, including their sexual thoughts, behaviour, cognitive distortions, and anger management. Psycho-education can also be given to the JSOs family, therefore reducing socio-contextual risk factors. One commonly used socio-contextual intervention is Multi Systemic Therapy (Borduin, Schaeffer, & Heiblum, 2009)

In some occasions a third form of intervention is used, focussing only on individual risk factors. This intervention type is the biological and pharmaceutical treatment (Efta-Breitbach, & Freeman, 2005). It changes the hormone balance in the offender by administrating medication. In rare circumstances a castration is performed (Efta-Breitbach, & Freeman, 2005).

Moderating Factors of the Relationship Between Risk Factors and Recidivism by JSOs As shown in Figure 1, the moderating factors in this meta-analysis surrounding the offence will be socio-demographical information about the offender (i.e., age and ethnicity). In addition, offence typology (i.e., hands-on or hands-off, alcohol use during crime), victim typology (i.e., age, gender and relationship to the offender) and treatment (i.e., treatment or penalty) will be described.

(11)

11

JSOs, although similar to each other in the crimes committed, have a very diverse array of characteristics. JSOs can be male or female and have different ages. Young offenders are influenced more by their family, whereas older offenders are influenced more by their friends, school and community (Van der Put, Deković, Stams, Hoeve, Laan, 2012). It is therefore hypothesised that the age of the offender could have a moderating ability in the relationship between socio-contextual risk factors and the chance of sexually recidivating.

Van der Put, et al. (2013), conducted a meta-analysis into adolescent general recidivism after a sexual crime, providing an insight into possible moderating factors. They concluded that interpersonal risk factors have a bigger influence on JSOs recidivating for hands-off offenders than hands-on offenders. In addition, the age of the victim has an influence on the relationship between personal risk factors and the type of crime and JSOs recidivating.

This meta-analysis will provide information on risk factors for JSOs to recidivate sexually during adolescence, by answering the following research question: which individual, interpersonal, and socio-contextual risk factors contribute to the chances of recidivism of juvenile sexual delinquency?

Methods Selection of Papers

Relevant articles were collected by searching through three online databases; PsychInfo, Web of Science and PubMed. Various combinations of the following keywords were used: sexual crime, sex crime, rape, molestation, sexual assault; victim, perpetrator, offender, offender; recidivism, repeat crime, recurrence, repeat, multiple; risk(s), risk factors and adolescent, teen, juvenile, youth, young, child. This resulted in 15,763 hits, which were evaluated. In addition, the ancestry method was applied.

(12)

12

When, based on the title of an article, it seemed the research entailed JSOs, the abstract was read to determine whether the paper met the criteria of this meta-analysis. If the article was deemed to be relevant, namely when containing information about risk factors for recidivism in sexual crimes committed by JSOs, the full article was read to further determine its significance of aforementioned article to this meta-analysis. The reference lists of all papers with relevant abstracts were also viewed in search of eligible articles. This provided in two articles.

Five inclusion criteria were used to determine if the articles were eligible for the meta-analysis. Based on these criteria, a selection of ten articles was made, bringing the definitive selection of articles to twelve, as shown in Table 1. Firstly, the article had to research a relationship between a risk factor and recidivism. Secondly, the study had to provide a valid statistic. Thirdly, JSOs included in the studies had to be aged between 12 and 21 years of age during their crime and the included papers focused on recidivism while the sexual offender was still considered an adolescent (between 12 and 21 years of age). When an article offered insufficient information regarding the age at the time of recidivism, an estimation was made based on the average age at the first crime and the average follow up period. Fourthly, only papers that focused on perpetrators were to be included, no victim reports were used. The sexual offenders had to have committed a sexual crime towards another person. Finally, the individual papers had to be statistically independent samples.

Only published articles were included, as non-published articles were not all possible to find. In general, only studies with positive and/or significant results tend to be published. It was noted that this could have led to a less objective selection of studies; the publication bias. Articles published before the first of January 2015 were included in the search.

(13)

13 Coding

When an article was included, the article was structurally coded by the author. General information about the paper (i.e., sample size and sample description, measures, and analysis methods), results and moderators were coded using a structured coding scheme written prior to the data selection.

Concepts

Juvenile sexual offenders recidivating sexually during adolescence. The following measures regarding the JSOs were reported: (a) type of crime, either a hands-on crime (i.e., rape, assault, touching of genitalia and forcing the victim to touch the offender), or a hands-off crime (i.e., voyeurism, exhibitionism and possession of child pornography), (b) hands-offender characteristics (i.e., age of offender – between 14.0 and 21.0 years old - and possible alcohol use during the offence) and (c) victim characteristics; the age of the victim (i.e., under or over the age of twelve) and the type of relationship between the victim and the sexual offender (i.e., known or unknown). The following measures regarding the recidivism offence were reported: (a) the time between the offences and (b) the type of recidivism (i.e., same crime, same victim, or different crime or different victim).

Risk Factors for JSOs Recidivating

The twelve studies described 74 risk factors as separate variables. Categorising them into three subcategories made the results more practically applicable. The risk factors analysed were variables influencing the chance of JSOs recidivating with a sexual offence.

Individual risk factors were variables associated with the perpetrator him/herself. These factors were divided into three subcategories; personality characteristics, general deviancy and sexual tendencies.

(14)

14

Firstly, risk factors in personality characteristics consisted of an impaired intellectual or language ability, psychological or psychiatric problems or diagnoses (i.e., impulsivity, hyperactivity, concentration problems, conduct disorder or a high score on the YLS/CMI or PCL-R questionnaire). In addition, it consisted of somatic problems, not experiencing remorse or guilt, and a lack of motivation to change.

Secondly, sexual risk factors included being a victim of sexual abuse or victimization, deviant sexual arousal, preoccupation or drive. It also included inappropriate sexual behaviour, including deviant sexual interests (i.e., children or violence), having a supportive attitude towards sexual offending and an unwillingness to alter these deviant sexual interests and attitudes.

Thirdly, general deviancy risk factors were school rebelliousness or truancy, aggressive or anti-social behaviour and the abuse of alcohol and drugs. Previous convictions for non-sexual crimes were also included as general deviancy risk factors.

Interpersonal risk factors were variables describing the relationship between the perpetrator and other people. In this meta-analyses it included an anti-social or aggressive interpersonal orientation and a social skills deficiency.

Socio-contextual factors were variables within the social environment of the perpetrator. These were divided into three subcategories; family/parents, peers and community.

Firstly, family and parental risk factors included non-cohabiting or divorced parents, and separation of biological parents, or out-of-home placement. It also included parental behavioural problems, inadequate parenting and a problematic parent-child relationship and a non-supportive or abusive parent. In addition, a disposition of a hereditary psychiatric disorder was also a family and parental risk factor.

(15)

15

Secondly, peer risk factors consisted of a lack of friends, a history of being bullied and negative peer relationships and influences.

Thirdly, community risk factors included a lack of community stability and adjustment, negative environmental characteristics and the environment supporting opportunities to recidivate sexually.

Moderators

Socio-demographic characteristics (of the offender). The socio-demographic characteristics of the offender that were described in the studies, were the age and the ethnicity of the offender. The ethnicity of the offender was scored as either Caucasian, or non-Caucasian. If the sample included more than 67% Caucasian JSOs, the whole sample was seen as Caucasian (k = 4). Other samples were scored as mixed (k = 4); no samples had a majority of non-Caucasian JSOs. The age of the offender at the time of the first offence was scored as a continuous variable. The age of the offender during the study could not be scored, as the studies mostly reported on retrospective case-studies. All studies reported on mostly hands-on offences; over 67%, so moderator analyses could not be performed into the type of offence.

If more than 67% of the victims in a study were of pre-pubescent age (under twelve years old), the sample was viewed as having pre-pubescent victims (k = 5). If more than 67% of the victims were over the age of twelve, it was viewed as post-pubescent (k = 1). Other samples were considered mixed (k = 2).

When more than 67% of JSOs in a sample received therapy as an intervention after their first offence, the sample was viewed as having received therapy (k = 1). When more than 67% of JSOs in a sample received a penalty after their first offence (i.e., incarceration, fine or community service order), the sample was viewed as having received a penalty (k = 0).

(16)

16

Two samples included JSOs receiving both therapy and a penalty (more than 67%). Other samples were viewed as mixed (k = 2).

As all samples included more than 67% female victims, moderator analyses could not be performed into the gender of the victim. When more than 67% of the victims in a sample knew the offender, the sample was viewed as known (k = 2). The same categorisation is used for non-known samples (k = 1). As only one study reported on victims who did not know the offender, moderator analyses could not be performed.

Data-Analysis

In order to make the individual papers comparable, the reported effect sizes had to be transformed into equal metrics. The chosen effect size was the Pearson’s correlation, r, as this was a commonly used effect size for reporting in meta-analyses. Papers which used the metrics chi-squared or frequencies to report, were transformed using an online transformation program developed by Wilson (2001).

Individual Study Effect Size

The individual study effect size was determined for all twelve samples, for every risk factor subcategory. It was determined to which subcategory every risk factor belonged (e.g., individual and sexual). All reported effect sizes of risk factors belonging to one subcategory were averaged within the studies, providing the individual study effect size for a subcategory of risk factors. The average was calculated using Wilson's (2011) online transformation program.

(17)

17 Mean Effect Size

Using a macro developed by Wilson (2005a), the mean effect size for a risk factor subcategory was determined, by comparing all individual study effect sizes for this subcategory of risk factors. To reliably generalise the effect of the risk factors for the population, the size of a sample group had to be considered when calculating the mean effect size. Therefore, the weighted sample size was determined by using the formula n-3, for each study (Field & Gillett, 2010). The resulting mean overall effect size was interpreted as r = 0 equals no effect and r = 1 equals a perfect effect. In addition, a Pearson's correlation r between .00 and .10 had no effect, an r between .11 and .30 a small effect, an r between .31 and .50 a medium effect and an r larger than .51 a large effect (Cohen, 1992).

Moderator Analyses

After computing the overall mean effect size, a homogeneity test was conducted to determine whether variance in results within in the data were due to statistical variance, or due to a difference in the sample characteristics of the separate studies. A significant result in the homogeneity test therefore gave reason to perform moderator analyses (Field, & Gillett, 2010).

If moderator analyses were deemed appropriate, they were conducted to establish whether the moderator variables had an influence on the found effect the risk factor subcategory had on the chance to recidivate. Continuous moderators were tested by a regression analysis using a macro developed by Wilson (2005b) and the categorical moderators were tested with a one-way ANOVA, using a macro developed by Wilson (2005c).

To determine if the results of this meta-analysis were invalidated by the publication bias, Rosenthal’s fail safe number N was calculated. A publication bias may have occurred, as

(18)

18

it was not possible to be sure that all research regarding this subject was included in this meta-analysis. Rosenthal's fail safe number N is the number of non-published studies required to not be included in the meta-analysis to make the results of the meta-analysis non-significant (Field & Gillett, 2010). When N was larger than (5 x k + 10), it was assumed that the publication bias did not significantly influence the results of the meta-analysis (Fragkos, Tsagris, & Frangos, 2014).

Results Sample of Studies

Twelve studies, representing twelve independent samples, were included in this meta-analysis, as shown in Table 1. The studies were published between 1991 and 2013, using data collected between 1978 and 2008. The studies reported on a total of 2,740 JSOs (M = 228.33, SD = 180.25). The sample sizes ranged from 9 to 636 JSOs. They were mostly male JSOs (99%, range: 96 % – 100 %) between the ages of 14.0 and 18.1 years old (M = 15.66, SD = 1.35) at the time of their first sexual offence. The age of the offenders during the study was not reported, as the studies consisted of retrospective file studies. Six of the articles were based on American data, three on Canadian data, one on Swiss data, one on Swedish data and one on Dutch data. Four samples included more than 67% Caucasian JSOs, four samples were mostly non-Caucasian (i.e., African American, Native American, Hispanic, Asian, Moroccan or Antillean). The other four samples did not report on the ethnicity of the JSOs. The articles were all based on a case-control study, using control groups including JSOs that did not recidivate, JSOs that only recidivated with a non-sexual offence, or a combination of both, to determine risk factors for sexual recidivism. Together, the articles provided a total of 67 effect sizes for the subcategories.

(19)

19 First Offence Description

Of the 2,740 JSOs, 1,096 (range: 77% - 100%) committed a hands-on first time sexual offence. Three studies reported between 67% and 74% female victims and four studies reported between 14% and 26% male victims. Most offences (82%) were directed at a victim known to the offender, with 76% of the victims being of pre-pubescent age (under 12 years old). Two studies reported on either 71% or 58% of the offenders using violence during the offence. During the follow-up period (M = 45.3 months, range: 3.7 – 79.2 months), 873 JSOs (M = 72.83, SD = 94.83) recidivated sexually, between the ages of 14.6 and 21.0 years old (M = 18.52, SD = 2.46).

Meta-Analysis 1: Individual Risk Factors for Adolescent Sexual Recidivism

Eleven studies, reporting on 2626 JSOs, were included in the meta-analysis regarding individual risk factors (i.e., personality characteristics, sexual risk factors and risk factors in general deviancy). The studies reported individual study effect sizes between .00 and .49. This resulted in overall mean effect sizes of Esrfixed = .10 and Esrrandom = .12. Thus, the more

individual risk factors, the higher the chance of sexual recidivism by JSOs. The fail safe number N (89.27), was larger than (5 x k + 10) = 65, therefore the publication bias probably did not influence the results of this meta-analysis. The heterogeneity test revealed a significant result (Q = 36.41, df = 10, p < .001), so moderator analyses were conducted.

Socio-demographic characteristics. The ethnicity of the JSOs was not a significant moderator (Q = .56, df = 1, p = .451). The samples with mostly Caucasian JSOs (k = 3) and mixed samples (k = 4), had the dame relationship between individual risk factors and JSOs recidivating. The age of the JSOs (k = 9) was also not a significant moderator (ß = -.33, df = 1,

(20)

20

p = .351). The age and ethnicity of the JSOs did not significantly alter the relationship between individual risk factors en JSOs recidivating.

Victim typology. The age of the victim at the time of the first offence did not provide a significant result as a moderator (Q = .09, df = 1, p = .765). The difference between the samples with victims under twelve years old (k = 5) and the samples with mixed samples (k = 2) did not significantly influence the relationship between individual risk factors en JSOs recidivating.

Intervention characteristics. Only one sample reported on JSOs that had received only therapy, and only one has reported on JSOs receiving both therapy and a penalty. Therefore, a valid moderator analyses could not be performed.

Meta-Analysis 1.1: Risk Factors in Personality Characteristics for Adolescent Sexual Recidivism

Risk factors in personality characteristics (e.g., psychological, psychiatric problems and a lack of motivation to change), were described in nine studies, reporting on 1767 JSOs. The studies offered individual study effect sizes between -.26 and .22. The overall mean effect sizes for both the fixed and random model were Esrfixed = -.02 and Esrrandom = .00. Therefore,

having more personality risk factors had a little effect on the chance of recidivism. The fail safe number N (-7.54), was smaller than (5 x k + 10) = 55, therefore the publication bias probably did not influence the results of this meta-analysis. The heterogeneity test revealed a significant result (Q = 31.08, df = 8, p < .001), thus moderator analyses were conducted.

Socio-demographical characteristics. The ethnicity of the JSOs significantly influenced the relationship between personality risk factors and JSOs recidivating (Q = 7.32,

(21)

21

df = 1, p = .007). For samples with Caucasian JSOs (k = 2), the relationship was positive (Esrfixed = .08) and for mixed samples (k = 4) it was a negative relationship (Esrfixed = -.08).

The age of the JSOs at the time of their first crime (k = 7) did not have a significant influence on the relationship between personality risk factors and JSOs recidivating (ß = -.06, p = .883).

Victim typology. The moderator analysis regarding the age of the victim could not be performed, as only one study reported on mixed samples, no studies reported on victims under the age of 12. Most samples reported on victims over the age of 12 (k = 5).

Intervention characteristics. As the type of therapy was not varied between the studies, all intervention types were described in only on study, it could not be analysed as a moderator.

Meta-Analysis 1.2: Sexual Tendency Risk Factors for Adolescent Sexual Recidivism Sexual risk factors (e.g., inappropriate or deviant sexual arousal and being a victim of sexual abuse) were described in ten studies, including 2106 JSOs. These studies provided individual study effect sizes between .04 and .91. The fixed model resulted in an overall mean effect size of Esrfixed = .21, the random model resulted in an overall mean effect size of

Esrrandom = .26. Therefore, the more sexual tendency risk factors, the higher the chance of

recidivating. The fail safe number N (159.25), was larger than (5 x k + 10) = 60, therefore the publication bias probably did not influence the results of this meta-analysis. Moderator analyses were conducted, as the heterogeneity test revealed a significant result (Q = 317.81,

df = 9, p < .001).

Socio-demographical characteristics. The ethnicity of the JSOs significantly influenced the relationship between sexual risk factors and JSOs recidivating (Q = 5.88, df =

(22)

22

1, p = .015). Samples with Caucasian JSOs (k = 3) showed a stronger relationship (Esrfixed =

.291) than mixed samples (k = 3) (Esrfixed = .162). The age of the JSOs at the time of their first

offence did not significantly influence the relationship between sexual risk factors and JSOs recidivating (ß = -.45, p = .222)

Victim typology. The difference between samples with victims under the age of 12 (k = 4) and mixed samples (k = 2), was not significant (Q = 2.61, df = 1, p = .107). The age of the victim did not significantly alter the relationship between sexual risk factors and JSOs recidivating.

Intervention characteristics. As only one study reported on JSOs receiving only treatment, and only one study reported on JSOs receiving both treatment and a penalty, moderator analysis could not be performed.

Meta-Analysis 1.3: General Deviancy Risk Factors for Adolescent Sexual Recidivism Eleven studies, including 2626 JSOs, were included in the meta-analysis describing the effect of general deviancy risk factors (e.g., previous convictions and substance abuse) on adolescent sexual recidivism. The individual study effect sizes provided by the studies lay between -.13 and .21. A small overall mean effect size was found using the fixed model Esrfixed = .08 and the random model Esrrandom = .07. Thus, a higher general deviancy risk

increased the chance of JSOs recidivating. The fail safe number N (17.17), was smaller than (5 x k + 10) = 65, therefore the publication bias probably did influence the results of this meta-analysis. The homogeneity analysis gave a significant outcome (Q = 23.93, df = 10, p = .008), so moderator analyses were conducted.

(23)

23 Moderator Analyses

Socio-demographical characteristics. The ethnicity of the JSOs significantly influenced the relationship between general deviancy risk factors and JSOs recidivating (Q = 7.80, df = 1, p = .005). The relationship was stronger (Esrfixed = .16) for mixed samples (k =

4), but than for Caucasian samples (Esrfixed = .03) (k = 3). There was no significant influence

found on the relationship between general deviancy risk factors and JSOs recidivating (k = 9), by the age of the JSOs at the time of their first offence (ß = .36, p = .302). Thus, the age of the offender had no influence on the relationship between general deviancy risk factors and JSOs recidivating.

Victim typology. The age of the victim did not significantly influence the relationship between general deviancy risk factors and JSOs recidivating (Q = .98, df = 1, p = .322). The relationship between general deviancy risk factors and JSOs recidivating was the same for samples with victims under the age of twelve (k = 5) and mixed samples (k = 2).

Intervention characteristics. The type of treatment the JSOs received was not varied between the studies and therefore it moderator analyses could not be performed. Only one study reported on JSOs receiving only therapy and only one reported on JSOs receiving both therapy and a penalty.

Meta-Analysis 2: Interpersonal Risk Factors for Adolescent Sexual Recidivism

Interpersonal risk factors (i.e., an anti-social or aggressive interpersonal orientation or a social skills deficiency), were described in three studies, including 421 JSOs. They provided individual study effect sizes between -.01 and .21. These resulted in an overall mean effect size of Esrfixed = .09 and Esrrandom = .10. Thus, the higher the interpersonal risk, the higher the

chance of JSOs recidivating. The fail safe number N (.77), was smaller than (5 x k + 10) = 20, therefore the publication bias probably did influence the results of this meta-analysis. As the

(24)

24

homogeneity test was not significant (Q = .00, df = 1, p = .979), moderator analyses were not performed.

Meta-Analysis 3: Socio-Contextual Risk Factors for Adolescent Sexual Recidivism

The meta-analysis into socio-contextual risk factors for adolescent sexual recidivism (i.e., family and parental risk factors, risk factors in peers and community risk factors.) included eight studies, reporting on 2142 JSOs. The individual study effect sizes lay between -.15 and .30. The overall mean effect size found was Esrfixed = .03 and Esrrandom = .06. Thus,

the higher the socio-contextual risk, the higher the chance of JSOs recidivating. The fail safe number N (-.97), was smaller than (5 x k + 10) = 50, therefore the publication bias probably did influence the results of this meta-analysis. The homogeneity test revealed a significant result (Q = 42.32, df = 7, p < .001), therefore moderator analyses were performed.

Socio-demographical characteristics. The age of the offender at the time of the first offence (k = 6), had a significant influence on the relationship between socio-contextual risk factors and JSOs recidivating (Mean ES = .04, ß = .82, p < .001). Thus, the older the offender at the time of the first offence, the larger the effect of socio-contextual risk factors on them recidivating. Only one study reported on Caucasian samples, and therefore moderator analyses was not performed.

Victim typology. The age of the victim does not significantly alter the relationship between socio-contextual risk factors and JSOs recidivating (Q = 1.12, df = 1, p = .291). Samples with victims under twelve years old (k = 3) and mixed samples (k = 2) showed the same relationship between socio-contextual risk factors and JSOs recidivating.

(25)

25

Intervention characteristics. The characteristics of the treatment the JSOs received were not varied between the studies and moderator analyses could therefore not be performed. Only one study reported on JSOs receiving therapy.

Meta-Analysis 3.1: Family and Parental Risk Factors for Adolescent Sexual Recidivism A total of 1283 JSOs were included in the six studies included in the meta-analysis regarding family and parental risk factors for adolescent sexual recidivism (e.g., non-cohabiting or divorced parents and disposition for a hereditary psychiatric disorder). The studies reported on individual study effect sizes between -.15 and .28. The overall mean effect size was Esrfixed = .02 and Esrrandom = .05. Thus, the higher the risk factors associated with the

family and parents of the JSOs, the higher the chance of the JSOs recidivating. The fail safe number N (-5.50), was smaller than (5 x k + 10) = 40, therefore the publication bias probably did influence the results of this meta-analysis. As the heterogeneity test revealed a significant result (Q = 38,49, df = 5, p < .001), moderator analyses were conducted.

Moderator analyses

Socio-demographical characteristics. The ethnicity of the JSOs was not varied between the studies and therefore moderator analyses could not be performed. Only one study reported on Caucasian samples and two studies on mixed samples. The age of the JSOs at the time of the first offence had a significant influence on the relationship between family and parental risk factors and recidivism (k = 7). The older the offender, the less influence family and parental risk factors had on JSOs recidivating (Mean ES = -.01, ß = .92, p < .001).

Victim typology. The age of the victim was not varied, most samples (k = 3) were mixed, and only one had victims under the age of twelve. Therefore moderator analyses could not be performed.

(26)

26

Intervention characteristics. The type of intervention received by the JSOs were not sufficiently reported on to perform moderator analysis. One sample reported on JSOs receiving only treatment, and one sample reported on mixed interventions.

Meta-analysis 3.2: Risk factors in peer characteristics for Adolescent Sexual Recidivism Risk factors in peer characteristics (e.g., a lack of friends, or negative peer relationships and influences or having been bullied) were reported on by two studies, including 200 JSOs. They reported individual study effect sizes of r = .13 and r = .28. This resulted in a mean overall effect size of Esrfixed = .13 and Esrrandom = .13. Thus, the more risk

factors in peer characteristics, the higher the chance of JSOs recidivating.

As this meta-analyses was based on the report of two studies, moderator analyses could not be conducted.

Meta-Analysis 3.3: Community Risk Factors for Adolescent Sexual Recidivism

Four studies, reporting on 1336 JSOs, were used in the meta-analysis describing community risk factors for adolescent sexual recidivism (e.g., community stability). Individual study effect sizes between .03 and .32 were reported. This resulted in an overall mean effect size of Esrfixed = .12 and Esrrandom = .12. Therefore, the larger the community risk,

the higher the chance of the JSOs recidivating. The fail safe number N (7.25), was smaller than (5 x k + 10) = 30, therefore the publication bias probably did influence the results of this meta-analysis. The heterogeneity test revealed a significant result (Q = 17,84, df = 3, p = .001), thus moderator analyses were conducted.

(27)

27 Moderator analyses

Socio-demographical characteristics. One study reported on a Caucasian sample, one on a mixed sample. These groups were too small to perform moderator analyses. The age of the JSOs at the time of their first offence had a significant influence on the relationship between community risk factors and JSOs recidivating. The older the JSOs at the time of their first crime, the larger the effect of community risk factors on the JSOs recidivating (Mean ES = .119, ß = .86, p < .001).

Victim typology. Only one study reported on JSOs with victims under the age of twelve, and one study reported on victims with mixed ages, therefore moderator analysis could not be conducted.

Intervention characteristics. Moderator analysis into intervention characteristics could not performed, as only on study reported on mixed samples. No other studies reported on the influence of intervention characteristics on the relationship between community risk factors and JSOs recidivating.

Discussion

Prior research has been conducted to into risk factors for JSOs recidivating. However, most of this research was aimed at either adult sexual offenders, or non-sexual recidivism. Few studies focussed on adolescent sexual recidivism. In this study, three main meta-analyses, regarding individual, interpersonal and socio-contextual risk factors, were conducted to combine all research on this topic and compare risk factors for adolescent recidivism. In addition, three sub meta-analyses were performed into more specific individual risk factors (i.e., personal, sexual and general deviancy), and three into more specific socio-contextual factors (i.e., family/parents, peers and community). They included the data of twelve studies, reporting on 2,740 JSOs, describing 67 independent study effect sizes. These

(28)

28

meta-analyses concluded that individual risk factors had the highest effect, with the sexual risk factors subcategory having the highest effect of all subcategories. The smallest effect was that of another individual risk factor subcategory; the personality characteristics. Socio-contextual risk factors, specifically the family and parental risk factors had some of the smallest effects. Peers and community risk factors had some of the greatest effects on JSOs recidivating

Individual Risk Factors and Adolescent Sexual Recidivism

Prior research into general recidivism and adult sexual recidivism showed that individual risk factors had the biggest effect on JSOs recidivating, either with non-sexual offences, or during adulthood (Van der Put, et al., 2013; Seabloom et al., 2003). As hypothesised, individual risk factors also had the largest effect on adolescent sexual recidivism, of the three main meta-analyses. When the individual risk factors were sub categorised, risk factors in personality characteristics (e.g., academic skills, ADHD diagnoses) had the smallest effect, of all subcategories of risk factors, on JSOs recidivating. Sexual risk factors (e.g., deviant sexual interests and attitudes), had the biggest effect, of all subcategories of risk factors, on JSOs recidivating. This suggested that sexual risk factors were the most important risk factors of all subcategories. General deviancy risk factors (e.g., prior convictions, substance abuse), had an intermediate effect on JSOs recidivating in comparison to other risk factor subcategories.

Theoretical Framework Compared to the Results of these Meta-Analyses: Generalist and Specialist Perspective

In earlier research, risk factors had been tied specifically to sexual recidivism (specialist perspective), or to both sexual and general recidivism (generalist perspective)

(29)

29

(Pullman, & Seto, 2012). General deviancy risk factors (e.g., previous convictions, deviant behaviour) had one of the smallest effects on the chance of JSOs recidivating. In addition, sexual risk factors had the largest influence on JSO recidivating sexually. General deviant behaviour did not seem to have a connection to sexual recidivism. These results seemed to confirm the theory of a specialist perspective, that general recidivism had different risk factors than sexual recidivism.

Interpersonal Risk Factors and Adolescent Sexual Recidivism

Marshall (2010) found that when JSOs do not have social skills and have problems initiating and maintaining contact with others, they might seek less appropriate ways of contacting other people. As hypothesised, interpersonal risk factors had one of the greater effects on adolescent sexual recidivism, compared to the other risk factors. It had a smaller effect than individual, sexual, peers and community risk factors. This suggested that the relationship that the JSOs had with others, was an important risk factor for them recidivating.

Socio-Contextual Risk Factors and Adolescent Sexual Recidivism

It was hypothesised that socio-contextual risk factors did not influence the chance of adolescent sexual recidivism, as prior research into general, or adult recidivism by JSOs provided hardly any evidence for socio-contextual risk factors (Van der Put, et al., 2013; Seabloom et al., 2003). This was supported by the findings of this meta-analysis. When the subcategories were viewed however, only the family and parental risk factors seem to have little effect on the chance of JSOs recidivating. The community and peers risk factors seemed to have some of the largest effects. However, only two studies reported on risk factors surrounding peers. As it did seem to have an important effect, one of the greater effects of all risk factor categories, it seemed that it would be an important subject for further research.

(30)

30 Moderators

As hypothesised, sexual risk factors had the biggest effect on the chance of JSOs recidivating. As the offence they commit was sexual, it was expected that general sexual problems were the greatest predictor. This is however only the case for Caucasian samples, not for mixed samples. On the contrary, general deviancy risk factors only have an effect when the sample is non-Caucasian.

No significant moderators were found for interpersonal risk factors, which was not as expected for the full population. In the case of this meta-analysis however, it was as expected. Only three of the twelve studies provided information about interpersonal risk factors. Therefore, the variety of characteristics was minimal, decreasing the chance of moderating factors. Future research into interpersonal risk factors would be advised, as they did have an effect on adolescent sexual recidivism, but still very little research has been conducted into this topic.

The effect socio-contextual and community risk factors had on JSOs recidivating, were mostly significant for older JSOs. As hypothesised, younger JSOs were influenced more by their families and parents, than older JSOs (Van der Put, Deković, Stams, Hoeve, & van der Laan, 2012).

Questionnaires

Psychologists, psychiatrists and courts use questionnaires designed to examine the risk of recidivism in JSOs. Two of these questionnaires are the Juvenile Sex Offender Assessment Protocol–II, or J-SOAP-II (Prentky & Righthand, 2003) and the Estimate of Risk of Adolescent Sexual Offence Recidivism, or ERASOR (Worling & Curwen, 2001). For example, the total score, statistic summary score and sexual offence severity scale of the

(31)

J-A METJ-A-J-ANJ-ALYSIS INTO RECIDIVISM BY JUVENILE SEXUJ-AL OFFENDERS

31

SOAP-II and subscales 2 and 4, are predictive of the risk to sexual recidivism (Aebi et al, 2011).

The questions in the ERASOR focussed mainly on personal and parental risk factors. The J-SOAP-II focussed mainly on personal, sexual and community risk factors, and the type of intervention. The risk factors with the largest effect on JSOs recidivating were sexual, community and peers risk factors. Therefore, these commonly used questionnaires did not seem to be fully in line with the effect of certain risk factors. A new questionnaire, focussing mainly on sexual, interpersonal, peer and community risk factors could be more effective in predicting recidivism in JSOs.

Practical Uses of the Findings of these Meta-Analyses in the Intervention of JSOs

The conclusions drawn from this meta-analysis can be used in the intervention provided for JSOs. The intervention should focus mainly on the reduction of sexual, interpersonal, community and peers risk factors. Personal, general deviancy, socio-contextual and family and parental risk factors should possibly have less focus, if the aim is to decrease the chance of JSOs recidivating.

According to Efta-Breitbach and Freeman (2005), most treatments for JSOs focussed on individual risk factors, specifically on personality characteristics, like Cognitive Behavioral Therapy. In addition to explaining their personality characteristics to JSOs, psycho-education also explains sexual risk factors to JSOs and their families. Multi Systemic Therapy focusses mainly on the family and parents of the JSOs. Thus, the current treatment options seem to target the risk factor categories that had the littlest effect on recidivism (Hanson, & Morton-Bourgon, 2005). Future treatments should target sexual risk factors, for example decreasing or desensitising deviant sexual arousal, thoughts or interests. It should

(32)

32

also include the peers and community, or at least teach the JSOs how to deal with certain risk factors associated with their peers and community.

Theoretical Framework Compared to the Results of these Meta-Analyses: Bronfenbrenner

As Bronfenbrenner (1979) described, most risk factors were categorised into either individual, interpersonal or socio-contextual risk factors. However, the type of crime, victim and treatment were also viewed as risk factors in some studies, and could not be categorised into Bronfenbrenners ecological model. Most risk factors; 45, were individual, four were interpersonal and 21 were of a socio-contextual nature. This shows that very little research has been conducted into interpersonal risk factors, even though they seem to have one of the greatest effects on JSOs recidivating. Future research could provide more insight into risk factors associated with interpersonal skills and problems.

Limitations

This paper included meta-analyses, with the aim to improve the estimates of the effect size of the found risk factors by increasing the numbers of observations and the statistical power of a risk factor in comparison to individual papers (Fagard, Staessen, & Thijs, 1996). The statistical power of the risk factors could have been increased even more, than what was possible in these meta-analyses, if these meta-analyses had not been limited by the following factors.

Firstly, these meta-analyses were based on twelve studies, some of them based on as little as two. The power of the found effect sizes could be greatly increased if more studies were added. In addition, due to the small number of studies, the risk of publication bias was great, as most fail safe number showed.

(33)

33

Secondly, if the samples that were used in the studies were more varied, more moderator analyses could have been performed. Most samples included only, or more than 95% boys and they had committed mostly hands-on offences. Not enough samples included victims not known to the JSOs, or male victims, to include moderator analyses for these variables.

Finally, the studies were not all sufficient in reporting all the facts and results. Most of them only reported significant findings, while non-significant findings could have been of interest when combined with findings from other samples. Many moderating factors were not described in the sample description, resulting in certain moderator analyses not being performed, as there was to little information, or not enough studies to compare with each other. They would be interesting to include in future research, as they were hypothesised to have an effect.

Future Research into JSOs Recidivating

The limitations of these meta-analyses exposed several opportunities for future research. In comparison to other subject matters, adolescent sexual recidivism was a very little researched subject. While research suggested peers had an influence on the chance of recidivism, not enough research had been conducted to form valid meta-analytical conclusions. Moreover, interpersonal risk factors should also be researched further, as only three studies reported on these risk factors and provided one of the larger effects of JSOs recidivating. In addition, future research could benefit by broadening sample groups, describing the sample characteristics to benefit moderator analyses and including non-significant findings in the results.

(34)

34 Conclusions

This meta-analysis aimed to collect the current data on adolescent sexual recidivism, which offered information that could be incorporated into new interventions or when assigning as intervention to a JSO. Sexual, interpersonal, peers and community risk factors had the largest effect on JSOs recidivating. Personality characteristics, family and parental risk factors and socio-contextual risk factors had the smallest effect.

The effects of the risk factors found in these meta-analyses were subjected to the age and the ethnicity of the offender as moderating factors. The older the JSO at the time of their first offence, the greater the relationship between socio-contextual, specifically community risk factors and recidivism. Family and parental risk factors had a greater effect on younger JSOs recidivating, than older JSOs. General deviancy risk factors had a larger effect on recidivism for Caucasian JSOs, and sexual tendency risk factors had a larger effect on groups of JSOs with mixed ethnicity.

These meta-analyses provided opportunities for future research into more risk factors, advising to focus on more varied samples, which could add valuable information about this important group of young offenders.

(35)

35 References

*Aebi, M., Plattner, B., Steinhausen, H. C., & Bessler, C. (2011). Predicting sexual and

nonsexual recidivism in a consecutive sample of juveniles convicted of sexual offences. Sexual abuse: a journal of research and treatment, 23, 456-473. DOI: 10.1177/1079063210384634

Aebi, M., Vogt, G., Plattner, B., Steinhausen, H. C., & Bessler, C. (2012). Offender types and offenderity dimensions in male juveniles convicted of sexual offenses. Sexual abuse: a journal of research and treatment, 24, 265-288. DOI: 10.1177/1079063211420449

Borduin, C. M., Schaeffer, C. M., & Heiblum, N. (2009). A randomized clinical trial of multisystemic therapy with juvenile sexual offenders: effects on youth social ecology and criminal activity. Journal of consulting and clinical psychology, 77(1), 26. DOI: 10.1037/a0013035

Bronfenbrenner, U. (1979). Contexts of child rearing: Problems and prospects. American Psychologist, 34, 844. DOI: 10.1037/0003-066X.34.10.844

*Caldwell, M. F., & Dickinson, C. (2009). Sex offender registration and recidivism risk in juvenile sexual offenders. Behavioral Sciences & the Law, 27, 941-956.

DOI: 10.1002/bsl.907

*Caldwell, M. F., Ziemke, M. H., & Vitacco, M. J. (2008). An examination of the Sex

Offender Registration and Notification Act as applied to juveniles: Evaluating the ability to predict sexual recidivism. Psychology, Public Policy, and Law, 14, 89. DOI: 10.1037/a0013241

*Christiansen, A. K., & Vincent, J. P. (2013). Characterization and Prediction of Sexual and Nonsexual Recidivism Among Adjudicated Juvenile Sex Offenders. Behavioral sciences & the law, 31, 506-529. DOI: 10.1002/bsl.2070

(36)

36

Chui, W. H., & Chan, H. C. O. (2012). Offender recidivism among Hong Kong male juvenile probationers. Journal of Child and Family Studies, 21, 857-868. DOI: 10.1007/s10826-011-9546-0

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

De Haas, S. (2012). Seksueel grensoverschrijdend gedrag onder jongeren en volwassenen in Nederland. Tijdschrift voor seksuologie, 36, 136-145. Retrieved from

http://www.rutgers.nl/sites/rutgersnl/files/PDF-onderzoek/Seksueelgrensoverschrijdendgedrag3.pdf

*Dennison, S., & Leclerc, B. (2011). Developmental factors in adolescent child sexual

offenders: A comparison of nonrepeat and repeat sexual offenders. Offender justice and behavior, 38, 1089-1102. DOI: 10.1177/0093854811417076.

Efta-Breitbach, J., & Freeman, K. A. (2005). Treatment of juveniles who sexually offend: An overview. Journal of Child Sexual Abuse, 13, 125-138. DOI: 10.1300/J070v13n03_07 Fagard, R. H., Staessen, J. A., & Thijs, L. (1996). Advantages and disadvantages of the meta-

analysis approach. Journal of Hypertension, 14, S9-S13. DOI 10.1097/00004872-199609002-00004

Field, A. P. & Gillett, R. (2010). How to do a meta-analysis. British Journal of Mathematical and Statistical Psychology, 63, 665-694. DOI: 10.1348/000711010X502733

Fragkos, K. C., Tsagris, M., & Frangos, C. C. (2014). Publication bias in meta-analysis: Confidence intervals for Rosenthal’s fail-safe number. International Scholarly

Research Notices, 2014.

Freudenberg, N., Daniels, J., Crum, M., Perkins, T., & Richie, B. E. (2005). Coming home from jail: the social and health consequences of community re-entry for women, male

(37)

37

JSOs, and their families and communities. American Journal of Public Health, 95, 1725-1736. DOI:10.2105/AJPH.2004.056325

Gerhold, C. K., Browne, K. D., & Beckett, R. (2007). Predicting recidivism in adolescent sexual offenders. Aggression and Violent Behavior, 12, 427-438.

DOI: 10.1016/j.avb.2006.10.004

Hanson, R. K., & Morton-Bourgon, K. E. (2005). The characteristics of persistent sexual offenders: a meta-analysis of recidivism studies. Journal of consulting and clinical psychology, 73, 1154. DOI: 10.1037/0022-006X.73.6.1154

*Hendriks, J., & Bijleveld, C. (2008). Recidivism among juvenile sex offenders after residential treatment. Journal of sexual aggression, 14, 19-32.

DOI: 10.1080/13552600802133852

*Kahn, T. J., & Chambers, H. J. (1991). Assessing reoffense risk with juvenile sexual offenders. Child Welfare: Journal of Policy, Practice, and Program, 70, 333-345. Retrieved from

http://web.a.ebscohost.com.proxy.uba.uva.nl:2048/ehost/detail/detail? sid=32502b00-004c-4189-8482-5b3a9041ceaf

%40sessionmgr4004&vid=0&hid=4106&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d %3d#AN=106099229&db=rzh

*Långström, N., Grann, M., & Lindblad, F. (2000). A preliminary typology of young sex offenders. Journal of adolescence, 23, 319-329. DOI: 10.1006/jado.2000.0317 Lussier, P., Bouchard, M., & Beauregard, E. (2011). Patterns of offender achievement in

sexual offending: Unravelling the “successful” sex offender. Journal of Offender Justice, 39, 433-444. DOI: 10.1016/j.jcrimjus.2011.08.001

(38)

38

career parameters of adult sex offenders. Aggression and violent behavior, 18, 445-457. DOI: 10.1080/07418825.2015.1028966

Maletzky, B. M., & Steinhauser, C. (2002). A 25-year follow-up of cognitive/behavioral therapy with 7,275 sexual offenders. Behavior Modification, 26(2), 123-147. DOI: 10.1177/0145445502262001

Marshall, W. L. (2010). The role of attachments, intimacy, and loneliness in the etiology and maintenance of sexual offending. Sexual and Relationship Therapy, 25(1), 73-85. DOI: 10.1080/02674659308408187

Mason, F., & Lodrick, Z. (2013). Psychological consequences of sexual assault. Best Practice & Research Clinical Obstetrics & Gynaecology, 27, 27-37. DOI: 10.1016/j.bpobgyn.2012.08.015

National Adolescent Perpetrator Network (1988). Preliminary report from the National Task Force on Juvenile Sexual Offending 1988. Juvenile & Family Court Journal, 39, 1-67. DOI: 10.1111/j.1755-6988.1988.tb00614.x

Office for National Statistics (2015). Crime Statistics - Pre-release access, Focus on Violent

Crime and Sexual Offences, 2013/14. Retrieved from

http://www.ons.gov.uk/ons/rel/crime-stats/crime-statistics/focus-on-violent-crime-and-sexual-offences—2013-14/pra-list-focus-on-violent-crime-2013-14.html

Parks, G. A., & Bard, D. E. (2006). Risk factors for JSO recidivism: Evaluation of predictive factors and comparison of three groups based upon victim type. Sexual abuse: a journal of research and treatment, 18, 319-342. DOI: 10.1007/s11194-006-9028-x Piquero, A., Farrington, D., & Blumstein, A. (2007). Key issues in offender career research:

New analysis of the Cambridge study in offender development. New York, NY: Cambridge University Press. ISBN: 0-521-84865-2.

(39)

39

II) manual. Washington, DC: US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Chicago. Retrieved from https://www.ncjrs.gov/pdffiles1/ojjdp/202316.pdf

Pullman, L., & Seto, M. C. (2012). Assessment and treatment of JSOs:

Implications of recent research on generalist versus specialist explanations. Child abuse & neglect, 36, 203-209. DOI: 10.1016/j.chiabu.2011.11.003

*Rajlic, G., & Gretton, H. M. (2010). An Examination of Two Sexual Recidivism Risk Measures in Adolescent Offenders The Moderating Effect of Offender Type. Offender Justice and Behavior, 37, 1066-1085. DOI: 10.1177/0093854810376354

*Ralston, C. A., & Epperson, D. L. (2013). Predictive validity of adult risk assessment tools with juveniles who offended sexually. Psychological assessment, 25, 905.

DOI: 10.1037/a0032683

*Rasmussen, L. A. (1999). Factors related to recidivism among juvenile sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 11(1), 69-85. DOI: 10.1023/A:1021332929575

Riser, D. K., Pegram, S. E., & Farley, J. P. (2013). Adolescent and young adult male sex offenders: Understanding the role of recidivism. Journal of child sexual abuse, 22(1), 9-31. DOI: 10.1080/10538712.2013.735355

Rothman, K. J., Greenland, S., & Walker, A. M. (1980). Concepts of interaction. American

journal of epidemiology, 112, 467-470. Retrieved from

http://aje.oxfordjournals.org/content/112/4/467.extract

Seabloom, W., Seabloom, M. E., Seabloom, E., Barron, R., & Hendrickson, S. (2003). A 14- to 24-year longitudinal study of a comprehensive sexual health model treatment program for JSOs: Predictors of successful completion and subsequent offender