Offense Trajectories,the Unfolding of Sexual and Non-Sexual Criminal Activity,and Sex Offense Characteristics of Adolescent Sex Offenders

Abstract

The current study examines offending trajectories of adolescent sexual offenders (ASOs). Until recently, classification frameworks have not been designed to account for the heterogeneity of offending patterns in adolescence, how these are associated with the unfolding of sexual and non-sexual criminal activity, and whether and to what extent they are related to the characteristics of sex offenses in adolescence. The current study takes a longitudinal view of offending in adolescence by examining retrospective longitudinal data of 217 ASOs referred for treatment to a clinical service between 2001 and 2009 in Australia. General offending trajectories in adolescence were examined using semi-parametric group-based modeling, and compared according to non-violent non-sexual, violent-non-sexual, and sex offending criminal activity parameters (e.g., participation, onset, frequency, specialization/versatility) and the characteristics of the referral sexual offense. The results show distinct differences in the unfolding of sexual and non-sexual criminal activity along different offending trajectories of ASOs, and further, that these trajectories were differentially associated with the characteristics of the sexual offenses they committed.

Keywords

juvenile sex offenders criminal careers sex offense characteristics criminal behavior offending trajectories

Introduction

Offending trajectories of adolescent sex offenders (ASOs) are receiving increasing attention in the research literature (e.g., Lussier, van den Berg, Bijleveld, & Hendriks, 2012; McCuish, Corrado, Lussier, & Hart, 2014). This is an important innovation because they describe a longitudinal pattern of offending over time (i.e., onset, progression, variety, desistence; Blumstein, Cohen, Roth, & Visher, 1986) and allow for the identification of factors and outcomes associated with different courses of offending (Loeber & Le Blanc, 1990). Until recently, in the clinical context with ASOs, offending trajectories have been interpreted in a very limited scope, namely, onset of sex offending, the extent of criminal involvement, and recidivism (e.g., Butler & Seto, 2002; Miner, 2012; Vizard, Hickey, & McCrory, 2007). These studies have shown that early-onset and criminally involved ASOs most closely resemble adolescent non-sexual offenders (ANSOs) in terms of risk factor and offending profiles and that a sizable proportion of these youth commit a single sexual offense and desist from offending.

Although clinical research has long suggested the presence of an “antisocial type” of ASO (e.g., Becker, 1998; Becker, Kaplan, Cunningham-Rathner, & Kavoussi, 1986), increasing evidence indicates this generic class is much too broad to describe the underlying heterogeneity that characterizes their criminal behavior. Several studies have presented evidence to suggest that ASOs are characterized by multiple offending trajectories (Burton, 2000; Carpentier, Leclerc, & Proulx, 2011; Lussier et al., 2012; McCuish et al., 2014; Vizard et al., 2007). However, it is currently unclear as to how offending trajectories may be related to the unfolding of both sexual and non-sexual criminal activity of ASOs, and the nature of sex crimes they commit. Therefore, the current study consisted of three main aims to take a step forward in this direction by examining the (retrospective) longitudinal sequence of offending in a clinical sample of ASOs. First, we identified different general offending trajectories in adolescence among clinically referred ASOs. Second, we describe demographic and offending characteristics (i.e., participation, age of onset, frequency, and variety/specialization in non-violent, violent non-sexual and sexual offending) associated with different offending trajectories. Finally, we examined whether and how offending trajectories were associated characteristics of referral sexual offenses in the sample (i.e., victim and offense characteristics).

Criminal Involvement of ASOs

The heterogeneity of ASOs across their individual and sex offense characteristics has been long recognized by clinicians working with this population. In fact, although there is no commonly agreed upon optimal method for classifying ASOs, distinctions based on the extent of their criminal involvement (those with criminal histories vs. those without; Butler & Seto, 2002) and sexual offense characteristics such as group versus solo offenses or child versus peer/adult victims (e.g., Barbaree, Hudson, & Seto, 1993; Bijleveld & Hendriks, 2003) have received the most empirical support. Butler and Seto (2002) distinguished between sex-only (adolescents with only sexual offenses in adolescence) and sex-plus (adolescents with sexual and non-sexual offenses in adolescence) offenders. In this clinical sample, they observed that ASOs without a non-sexual criminal history had significantly fewer childhood and adolescent behavioral problems, more prosocial attitudes and beliefs, and a lower risk of future delinquency (Butler & Seto, 2002). In contrast, those with a history of non-sexual offenses most closely resembled ANSOs in terms of behavioral problems and prior offending variety.

These differences (sex-only and sex-plus) explain, in part, contrasting findings from studies that have broadly compared ASOs and ANSOs. For example, some studies indicate ASOs typically have less extensive histories than ANSOs (Seto & Lalumière, 2010). When Butler and Seto (2002) compared these two broad groups, they also found that ASOs had less extensive criminal histories and also were at lower risk for persistent delinquency compared with ANSOs. Similarly, using Dutch police data on arrests, Van Wijk, Mali, et al. (2007) found that ASOs were less likely than ANSOs to have criminal records prior to their index offense. Using these same data, Bullens, van Wijk, and Mali (2006) also observed that although ASOs had an earlier age of onset of criminal activity, the length of their involvement in crime was typically shorter compared with that of ANSOs.

Other findings have suggested that there are less extensive differences between these two broad groups (for a review see van Wijk et al., 2006). For example, using data from the Pittsburgh Youth Study, van Wijk and colleagues found minimal differences between these two groups in terms of general delinquent histories including theft, fraud, and serious delinquency (van Wijk et al., 2005). Similarly, the study by Bullens et al. (2006) also demonstrated that both ASOs and ANSOs were most likely to recidivate with a non-violent crime after their index offense (see also Nisbet, Wilson, & Smallbone, 2004). In effect, Butler and Seto’s (2002) early work firmly established that the utility of these broad comparisons (i.e., ASOs compared with ANSO) is limited at best without a more comprehensive understanding of the heterogeneity in the nature and extent of criminal involvement of this population.

The Unfolding of Criminal Activity Among ASOs

Early-onset offending is generally considered a harbinger of a life-course persistent (LCP) offending trajectory (Moffitt, 1993). It is considered to be a key marker of early psychosocial adversity that cascades and accumulates in subsequent developmental periods (e.g., beginning in childhood and through adolescence into adulthood). However, it is not clear that this is always the case for ASOs. Depending on the nature and extent of these adversities, different courses of offending or offending trajectories can emerge. In the context of ASOs, Vizard et al. (2007) examined whether the early onset of sexually abusive behavior (i.e., <11 years old) was associated with psychosocial deficits, sexual and non-sexual antisocial behavior, and conviction characteristics among a sample of ASOs referred for treatment. Their results indicated that early-onset persistent ASOs were characterized by significantly higher adversity along the lines of parental and family factors, attachment, psychosocial deficits, and trauma compared with late-onset ASOs. During childhood, the early-onset persistent group had generally higher rates of non-sexual antisocial behavior, but by adolescence, were more or less equivalent with the late-onset group.

Burton (2000) examined aspects of the sexual criminal activity of ASOs by classifying adolescents based on the onset and persistence of their sexual offending. Among a sample of ASOs referred for treatment in the United States, he identified three groups based on the sexual offending history: one for which the onset of sex offending occurred before age 12 (i.e., early offenders); those for whom sex offending was initiated in the teenage years (i.e., teen offenders); and finally those with sex offenses prior to and after the age of 12 (i.e., continuous offenders). Here, the “continuous” offenders were the most likely to be characterized by early sexual behavior problems and more severe risk factor profiles compared with the other two groups. This suggests that there is an underlying relationship between early sexual behavioral problems and early-onset, chronic, and versatile sex offending. Burton suggested these individuals potentially represent those at an elevated risk of continuing sex offending behavior into adulthood. In contrast, the early-onset (but not continuous) ASOs were also those with more serious non-sexual offending histories. Similar to the study by Vizard et al. (2007), this suggests that for some ASOs an early sex offense may initiate an offending trajectory that crosses over into other domains of non-sex offending, yet for others it is a relatively isolated event.

Carpentier et al. (2011) extended this line of investigation by combining elements of both sexual and non-sexual offending parameters, including the onset, variety, and desistence of criminal behavior. First, they distinguished early-onset sexual offending using age 12 as a cut-off. Next, they examined the variety of offending using the Butler and Seto (2002) dichotomy (i.e., sex-only vs. sex-plus), and finally, persistence/desistence was measured according to three groups: (a) those who had violent/sexual charges after their referral sex offense (stable-highs); (b) those with non-sexual non-violent charges after their referral offense (de-escalators); and (c) those with no charges after their referral offense (i.e., desisters). Their results mirrored those of prior studies (e.g., Bullens et al., 2006; Burton, 2000) indicating early-starters, sex-plus offenders, and stable-highs (i.e., violent/sexual recidivists) typically presented with the most severe risk factor profiles, had the earliest onset of sex offending, and were among the most aggressive and violent. In line with Burton’s (2000) study, the findings from Carpentier et al. (2011) suggest that the combination of aggressive and sexually deviant behaviors in childhood is associated with early-onset sex crimes in adolescence.

In a recent prospective longitudinal study, Lussier et al. (2012) examined sexual and non-sexual offending trajectories in a sample of Dutch male ASOs up to age 32 referred to specialized institutions for assessment. Using group-based modeling they identified five non-sexual offending trajectories (very low-rate, late-starters, adolescent-limited, late-bloomers, and high-rate persisters) and two sexual offending trajectories (high-rate slow desisters and adolescent limited) based on official convictions from adolescence to adulthood. Comparing these general and sexual offending trajectories uncovered interesting patterns. First, the sexual criminal involvement of the vast majority of ASOs in their study was limited to adolescence, ranging from the most criminally involved youth (high-rate persisters and late-bloomers) to the least criminally involved youth (very low rate and adolescent limited). In other words, early-onset, chronic serious and violent offending was neither a necessary nor a sufficient condition for the persistence of sexual offending into adulthood. In fact, the least antisocial ASOs identified in their study were those who were the most likely to be characterized by the sexual offending trajectory that continued into adulthood. In addition, an important subgroup (late-starters) had only sexual offenses in adolescence, but had non-sexual criminal activity in adulthood. In effect, these individuals would have been considered “sex-only” ASOs, yet it was in fact the case that their non-sexual offending followed rather than preceded their sexual offending. Taken together, these studies provide fairly clear evidence that there are multiple and diverse pathways leading to sexual aggression in adolescence and beyond.

Heterogeneity in Criminal Involvement and Sex Offense Characteristics of ASOs

As there are multiple pathways to sexual aggression in adolescence and beyond, the nature of sexual offenses committed by ASOs varies dramatically as well. Furthermore, current evidence indicates there is at least some degree of concordance between the unfolding of criminal activity and sex offense characteristics of ASOs; some of the key differences between sex-only and sex-plus ASOs involve victim and offense characteristics. For example, sex-only offenders were less likely to offend against strangers in Butler and Seto’s (2002) study. Hunter, Figueredo, Malamuth, and Becker (2003) found that adolescents who targeted peer-aged female victims had more severe patterns of antisocial behavior compared with those who targeted children. Similarly, Bullens et al. (2006) noted that ASOs with peer-age victims also had more overall offenses than those who only targeted children and non-contact offenders (e.g., exhibitionists). They also indicated that the onset offense for a majority of adolescents who targeted children was sexual in nature (approximately 80%), whereas this figure was somewhat lower at approximately two thirds of those who targeted peers and adults (Bullens et al., 2006). Vizard et al. (2007) found that early-onset persistent ASOs were more likely to commit a wider array of sex offenses in adolescence (e.g., involving males, females, children, and adults) compared with late-onset ASOs who were more likely to selectively target either females or younger children. Similarly, in Burton’s (2000) study, the sexual offenses of continuous offenders were more likely to involve a wide range of offense behaviors, including non-contact, contact, and serious penetrative acts. Finally, in the prospective longitudinal study by Lussier et al. (2012), the least serious non-sexual offending trajectories (very low rate and late-starter groups) were the ones most likely to be characterized by sex offenses involving child victims. On the other hand, the proportion of offenders with peer age victims was distributed much more equally across the non-sexual offending trajectory groups they identified, and group offenders were typically found among the most serious antisocial trajectories (adolescent-limited, late-bloomers, and high-rate persisters).

Aims

Although there are evident links between the unfolding of criminal activity and the nature of sex offenses committed by ASOs, no studies have described the relationship between offending trajectories in adolescence and the unfolding of sexual and non-sexual criminal activity of ASOs. Theoretically, this is a crucial question because sex crimes are embedded in offending trajectories that can be characterized by multiple crime types and offending patterns. Furthermore, most studies investigating the criminal involvement of ASOs have been based on cross sectional samples making it difficult to adequately describe important changes in the unfolding of criminal activity.

Given that there is minimal continuity of sex offending into adulthood (e.g., Lussier et al., 2012; Zimring, Piquero, & Jennings, 2007), from a policy perspective there is also good reason to focus these analyses on the period of adolescence. Here the goal is to inform policy makers about the heterogeneity of ASOs, and whether and to what extent their individual offending patterns are related to the nature of the sex offenses they commit. Given these considerations, the aims of the current study were threefold. The first was to investigate general offending trajectories among a clinical sample of ASOs referred for treatment. Second, we sought to describe in depth, the unfolding of non-sexual and sexual criminal activity (i.e., onset, frequency, versatility/specialization for sex, violent non-sex, and non-violent offending) of ASOs on different offending trajectories, and finally, to what extent different offending trajectories in adolescence were informative of the characteristics of sex offenses committed by these youth.

Method

Sample

As part of a larger project on understanding and preventing youth sexual violence and abuse, the current study involved youth who were charged with at least one sexual offense between 2001 and 2009 and referred for assessment to a specialized treatment service in Australia. Upon referral to this treatment service, youth are given a preliminary assessment where their demographic and details of their referral (i.e., sexual) offense are recorded in clinical files. Not all youth referred for assessment proceed to treatment as this clinic prioritizes high-risk cases. This is determined on the basis of their risk profile, as well as capacity issues. At the time of data collection for the current study, researchers accessed the entire official criminal histories from the state Department of Justice and Attorney-General (DJAG) of all of the youth ever referred to the service since 2001 (n = 350). In this Australian state the minimum age of criminal responsibility is 10 years old. Data were retained for participants who were officially charged with a (hands-on) sexual offense between the ages of 10 and 17 and were 18 years or older at the time data collection took place (n = 217). This was done to ensure we had complete criminal history information between the ages of 10 and 17 for the sample. Therefore, this represents a retrospective longitudinal sample of youth referred to this specific treatment service. The average age of participants at the time of their referral to the treatment service was 15.6 years old (SD = 9.5, range = 11-18). Almost two thirds were Caucasian (61.3%) and just over half of the sample (56.1%) was enrolled in school at the time of referral. A much smaller proportion was employed at the time of their referral to treatment (19.9%), and just over one quarter (27.3%) lived in remote/rural locations across the state. Just over two thirds of the sample (35.5%) had served some period of time in custody, a minority were sexual recidivists (8.3%), and for nearly two thirds (62.2%), their first criminal charge in adolescence was a sexual one. Finally, 39.6% of the sample had only charges for a sexual offense(s) in their adolescence.

Procedures

Ethics for the current project were obtained as part of a larger Australian Research Council (ARC) grant through the host university. In line with these protocols, official criminal history data were obtained on the sample. This information included any and all specific charges and their respective dates. Next, researchers accessed state police service reports located in each clinical file that documented those charges that were associated with their referral to the treatment service, as well as details regarding the characteristics of the referral offense (i.e., nature of charges and victim characteristics).

Measures

Demographic characteristics

We examined basic demographic and descriptive characteristics of the sample. These included age at referral, Indigenous status (0 = no, 1 = yes), enrolled in school at the time of the offense (0 = no, 1 = yes), employed at the time of the offense (0 = no, 1 = yes), ever been incarcerated in adolescence (0 = no, 1 = yes), whether their first criminal charge in adolescence was for a sex offense (0 = no, 1 = yes), whether they had more than one conviction for a sex offense (i.e., “sexual recidivist”; 0 = no, 1 = yes), and whether they had only sex offenses in their adolescent criminal record (i.e., “sex-only”; 0 = no, 1 = yes). We also included a variable measuring whether or not they resided in a remote Australian community. This particular clinical service is unique in that clinicians also often travel to rural and remote locations in the state to deliver treatment. Importantly, many of these communities are characterized by high degrees of social disorganization and have serious problems with sexual abuse and assault (Cale, 2014; Smallbone & Rayment-McHugh, 2013). Therefore, geographical remoteness was measured using the Accessibility/Remoteness Index of Australia (ARIA; Department of Health and Aged Care, 2001). The ARIA uses road accessibility to services to calculate a standard classification of remoteness. Scores ranged from 0 to 12, with a score of 12 indicating highly remote regions. An ARIA score was assigned by researchers based on adolescents’ recorded last known address in clinical files. A score of 3.52 was used as a cut-off to dichotomize non-remote versus remote (coded as 0 and 1). This was based on the recommendations by the Department of Health and Aged Care (2001) that indicate a score of 3.52 and above indicates significantly restricted accessibility of goods, services, and opportunities for social interaction.

Offense trajectories in adolescence

To examine offense trajectories in youth, we used eight repeated measures of finalized charges for sexual, violent non-sexual, and non-violent offenses at each age between 10 and 17 years old. Non-violent charges included any charges for property, weapon possession, or drug related offenses. Administrative charges (e.g., breach of probation) were excluded from the calculation of trajectories. We favored the use of finalized charges over convictions for two main reasons. First, it is difficult to ascertain the extent to which convictions reflect the activity of youth courts’ discretion in sentencing compared with the actual criminal activity of youth. Second, while charges on the other hand reflect to a certain extent the activity of police, diversion policies and the use of alternative measures are prominent in the state and also in Australia more broadly. Therefore, examining charges instead of convictions, at least for some youth, may capture a more detailed portrayal of their criminal activity in adolescence (see also DeLisi, Neppl, Lohman, Vaughn, & Shook, 2013). To establish the offending trajectories in youth of the current sample, the total number of charges for an individual was examined each year in adolescence beginning at 10 years old, until the time they were 17.

Criminal activity parameters and referral sex offense characteristics

For descriptive purposes, we examined the age at first charges, frequency of charges, and variety/specialization (i.e., for any offenses, non-violent offenses, violent non-sexual offenses, and sexual offenses) in adolescence. Variety of offending was calculated on a scale from 1 to 6 by examining the presence of (a) charges for a sexual offense, (b) charges for a violent offense, (c) charges for a property offense, (d) charges for a drug offense, (e) charges for a weapons offense, and (f) charges for public order or administrative offenses (e.g., motor vehicle charges, breach of conditions, etc.). Therefore, youth charged only for a sexual offense were scored with a 1, indicating there was no variety in their adolescent offense profiles, whereas youth who had the presence of all six types of charges in their adolescent offense history received a 6, indicating the highest level of variety according to the scale. Specialization for each offense category (i.e., non-violent, violent non-sexual and sexual offenses) was calculated as a ratio of the number of charges in the respective offense category to the total number of charges in an adolescent’s criminal history and then multiplied by 100 to reflect a percentage.

The average age at first charges for any offense was 14.6 (SD = 1.8) years old, the average number of charges in adolescence for the sample was 19.6 (SD = 31.4) charges, and on average, the level of variety was a score of 2.5 (SD = 1.5) out of 6. Just over half of the sample had ever been charged for a non-violent crime (55.3%), the average age these charges first occurred was 15.4 (SD = 1.0) years old, the average number of charges was 14.6 (SD = 30.2), and overall, non-violent charges represented approximately two thirds (M = 36.0, SD = 37.8) of the overall proportion of offenses for youth in the sample. Just over two thirds (38.7%) of the sample had ever been charged for a violent non-sexual crime, the average age at first charges for a non-sexual violent offense was 14.8 (SD = 1.7) years old, the average number of charges for non-sexual violence in adolescence was 1.2 (SD = 2.4), and on average, these charges represented a small proportion of the overall offending of youth in the sample (M = 6.2, SD = 57.7). Finally, the average age at first charges for a sexual offense was 15.5 (SD = 1.4) years old, the average number of sexual charges in the sample was 3.6 (SD = 3.6), and on average, sexual charges represented just half of the proportion of overall charges in the sample (M = 57.7, SD = 40.6).

Next we examined the nature of charges that accompanied the youth’s referral to treatment. More specifically, this included whether youth had, on referral, charges for only sexual offenses (65.9%), sexual and non-violent charges (24.0%), sexual and violent non-sexual charges (21.2%), and all three types of charges on referral (11.1%). In terms of the victim characteristics examined, for approximately one third of the sample the victim was a family member (33.2%), and approximately one quarter (25.8%) had a male victim. For over two thirds of the sample (68.7%), the referral offense involved a child victim below the age of 12, just over one-fifth (22.1%) involved a peer-age victim between the ages of 13 and 17, and finally, 16.1% involved an adult victim.

Analytic Strategy

To analyze offense trajectories in adolescence, we used semi-parametric group-based trajectory modeling (Nagin & Land, 1993). Group-based trajectory modeling is a specialized application of finite mixture modeling (Nagin, 1999, 2005) designed to identify clusters of individuals following similar progressions of behavior/outcome over age and time (Jones & Nagin, 2007). In this case, such mixture models are useful for modeling unobserved heterogeneity in a population because they do not assume there is an average age–crime curve for each individual. Here, offense trajectories between ages 10 and 17 years old were modeled as charges per youth per age year, using a zero-inflated Poisson model because the charge (i.e., count) data analyzed were characterized by intermittency (i.e., periods of non-offending; Nagin, 2005). In addition, given that a small proportion of youth had substantially high charge counts, we top-coded at 1% of the peak values in the data to reduce the influence of these very large counts (Weakliem & Wright, 2009). The shapes of trajectories identified each correspond to a polynomial function of time (i.e., intercept, linear, quadratic, and cubic; Nagin, 2005). To identify the appropriate number of trajectories, first we examined the Bayesian Information Criterion (BIC) which rewards parsimony in a trajectory model while imposing penalties to model fit for the addition of trajectory groups. Second, we examined the extent to which the content and shape of the trajectories was substantively meaningful in the current clinical context. Finally, considering that just over one third of the sample was incarcerated at some point in adolescence, it was important to also control for exposure time. Exposure per year was calculated as follows:

{Exposure}_{j i} = \frac{1 - (No . of months detained \times . 60)}{24},

for every person at time period j and age i (see van der Geest, Blokland, & Bijleveld, 2009). Given that specific sentence lengths were not available in the records obtained, we modified the original equation from days to months, and calculated 60% of the official recorded incarceration sentence because youth sentenced to prison in the current state serve a minimum of anywhere between 50% to 75% of the incarceration sentence.

Results

Offense Trajectories Between 10 and 17 Years Old

In the current analysis, the BIC value to select the appropriate number of groups steadily increased with the addition of up to six groups suggesting that statistically, more than six groups characterize the data (Table 1). However, in such instances, parsimonious models should be selected without concealing distinctive (i.e., theoretically and empirically important) developmental features of the data (Nagin, 2005). Therefore, we selected a four group model primarily based on the substantive usefulness and validity as it related to the clinical context of the sample and the research questions in the current study, namely, distinctions between the trajectories based on criminal activity parameters and the referral sexual offense characteristics (Nagin & Odgers, 2010). Figure 1 displays the offense trajectories of the four groups identified: (a) rare offenders (RA; 53.0%), (b) late-bloomers (LB; 25.3%), (c) low-rate chronics (LR; 10.1%), (d) and high-rate chronics (HR; 11.5%). Offending ranged, on average, from 0 to approximately 14.5 (calculated based on top-coded data) charges depending on the trajectory. The RA trajectory consisting of approximately half of the sample remained low and relatively stable in adolescence with less than two charges on average per age period. The LB trajectory also had relatively few charges in early adolescence but by the end of adolescence surpassed the other three trajectory groups. At the same time, the low-rate trajectory was characterized by a moderate peak in charges (on average above five) in mid-adolescence and declining thereafter, whereas the high-rate trajectory was characterized by a substantial and high volume of offending peaking slightly later in adolescence and declining rapidly approaching young adulthood.¹ Table 1 also shows the average posterior probabilities and ranges for assignment to each of these trajectory groups, which was computed as a model fit diagnostic. The average posterior probabilities for group assignment were all high above .89 and above indicating a small probability that individuals were assigned into an incorrect group.

Table 1.

BIC for Model Selection and Average Assignment Probabilities.

Number of groups	BIC
2	−2,843.80
3	−2,696.35
4	−2,605.38
5	−2,555.00
6	−2,524.77
Average assignment probabilities		Adolescent offense trajectories
Assignment group	% (n)	Rare offenders	Late-bloomers	Low-rate chronic	High-rate chronic	Range
Rare offenders	53.0% (n = 115)	.91	.06	.02	.01	.51-1.00
Late-bloomer	25.3% (n = 55)	.08	.90	.02	.00	.51-1.00
Low-rate chronic	10.1% (n = 22)	.02	.07	.89	.03	.50-1.00
High-rate chronic	11.5% (n = 25)	.00	.01	.04	.95	.47-1.00

Note. BIC = Bayesian information criterion.

Figure 1.

General offending trajectories of adolescent sex offenders (n = 217).

Demographic Characteristics and Offense Trajectories

The four offense trajectories differed on a number of demographic and descriptive characteristics (Table 2). There was a significant overall difference in age at time of referral across the four groups (RA > LB; p < .01). Youth in the RA trajectory were also most likely to have been employed at the time of referral. The LR and HR trajectories had the highest proportions of Indigenous youth (54.5% and 56.0% respectively) and also, the highest proportion of youth who had served some period of incarceration (63.6% and 72.0% respectively). The LB and LR trajectories had the highest proportion of youth that resided in remote/rural locations of the country (38.5% and 47.6% respectively). For a majority of youth in the RA trajectory (80.0%), their onset offense in adolescence was a sexual one, nearly two thirds (60.0%) had only sex offenses in adolescence (i.e., sex-only). The LB and the LR trajectories were also the most likely to have sexual recidivists, followed by the HR group, although because of the overall low base rate of sexual recidivism and low expected cell counts, these comparisons should be interpreted with caution.

Table 2.

Offending Trajectories in Adolescence and Demographic Characteristics.

	Total sample	Group 1 (RA)	Group 2 (LB)	Group 3 (LR)	Group 4 (HR)	F(df); χ²(df)	Partial η²; Cramer’s V
	(N = 217)	(n = 115)	(n = 55)	(n = 22)	(n = 25)	F(df); χ²(df)	Partial η²; Cramer’s V
Age at the time of referral, M (SD)^a	15.6 (1.4)	16.0 (1.1)	15.1 (1.6)	15.2 (1.7)	15.4 (1.3)	F(3, 54.6) = 5.5**	.08
Indigenous (%)	38.7	29.6	43.6	54.5	56.0	χ²(3) = 10.0*	.22
Remote Australian community (%)	27.3	20.0	38.5	47.6	27.3	χ²(3) = 10.3*	.22
Enrolled in school at time of referral (%)	48.8	61.8	50.0	50.0	47.6	χ²(3) = 2.9 n.s.	.13
Employed at time of referral (%)	17.5	27.7	9.6	12.5	13.6	χ²(3) = 8.4*	.21
Ever incarcerated during adolescence (%)	35.6	14.9	50.9	63.6	72.0	χ²(3) = 48.9***	.48
First criminal charge was sexual (%)	62.2	80.0	50.9	27.3	36.0	χ²(3) = 37.2***	.41
Sexual recidivist (%)^b	8.3	0.9	18.2	18.2	12.0	χ²(3) = 18.7***	.29
Sex-only offenders (%)	39.6	60.0	21.8	9.1	12.0	χ²(3) = 43.7***	.45

Note. RA = rare offender trajectory; LB = late-bloomer trajectory; LR = low-rate chronic trajectory; HR = high-rate chronic trajectory.

Equality of variance not assumed. Robust test of equality of means (Welch).

Low expected cell counts.

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

Criminal Activity Parameters of Offense Trajectories

Given that the same data were used to compute offending trajectories and criminal activity parameters, the following analyses are presented strictly for descriptive purposes. Youth in the RA trajectory were significantly older at their age of onset of general offending (M = 15.6, SD = 1.1 years old) than the three other trajectories (RA > LB, p < .001; RA > LR, p < .001; RA > HR, p < .001; Table 3). Given that official charges were used to calculate trajectories, each trajectory significantly differed from one another in terms of the frequency of any charges. Similarly, in terms of the variety of offending, youth in the RA trajectory showed a significantly lower crime-mix compared with the other three trajectories (RA < LB, p < .001; RA < LR, p < .001; RA < LR, p < .001).

Table 3.

Description of Criminal Activity in Adolescence that Characterizes the Four Offending Trajectories.

	Total sample (N = 217)	Group 1 (RA; n = 115)	Group 2 (LB; n = 55)	Group 3 (LR; n = 22)	Group 4 (HR; n = 25)	F(df); χ²(df)	Cramer’s V; Partial η²
Sample members (%)	—	53.0%	25.3%	10.1%	11.5%	—	—
General offending
Age at first charge^a	14.6 (1.8)	15.6 (1.1)	13.9 (1.7)	12.9 (1.7)	12.9 (1.8)	F(3, 53.8) = 40.7***	.39
Frequency^b	19.6 (31.4)	4.4 (3.7)	20.3 (17.0)	36.4 (27.4)	72.9 (56.3)	F(3, 213) = 105.0***	.60
Variety^b	2.5 (1.5)	1.7 (1.0)	3.1 (1.5)	3.4 (1.3)	3.8 (1.4)	F(3, 213) = 32.6***	.32
Non-violent crimes
Participation (%)	55.3	33.0	72.7	90.9	88.0	χ²(3) = 51.9***	.49
Age at first charge^a	13.9 (1.8)	15.4 (1.0)	13.5 (1.4)	12.9 (1.8)	13.2 (1.8)	F(3, 48.9) = 25.4***	.34
Frequency^a,c	14.6 (30.2)	1.5 (3.3)	13.7 (16.8)	30.3 (26.4)	63.1 (58.6)	F(3, 64.7) = 36.2***	.30
Specialization^a	36.0 (37.8)	17.1 (27.6)	36.3 (4.9)	71.2 (29.3)	72.0 (34.1)	F(3, 57.9) = 37.2***	.34
Violent crimes
Participation (%)	38.7	20.0	58.2	50.0	72.0	χ² (3) = 38.6***	.42
Age at first charge^a	14.8 (1.7)	15.8 (1.0)	15.2 (1.3)	13.4 (1.8)	13.8 (1.8)	F(3, 33.2) = 9.6***	.30
Frequency^a,c	1.2 (2.4)	0.3 (0.9)	1.7 (2.0)	2.6 (4.3)	3.6 (3.5)	F(3, 53.7) = 18.8***	.23
Specialization^a,b	6.2 (57.7)	4.9 (11.6)	9.0 (12.6)	6.4 (9.6)	5.9 (7.6)	F(3, 60.6) = 6.5***	.08
Sexual crimes
Age at first charge^a	15.5 (1.4)	15.9 (1.1)	14.9 (1.4)	14.9 (1.4)	15.2 (1.4)	F(3, 54.8) = 10.0***	.13
Frequency^a,b	3.6 (3.6)	2.6 (1.7)	4.9 (3.8)	3.4 (2.8)	6.2 (6.9)	F(3, 55.4) = 9.1***	.12
Specialization^a,b	57.7 (40.6)	77.8 (31.6)	46.1 (38.0)	22.1 (31.3)	22.1 (35.4)	F(3, 59.3) = 34.1***	.32

Note. Original means and standard deviations are reported. Analysis of Variance (ANOVA) was applied to examine the main effects between adolescent offense trajectory groups and criminal career parameters. The Scheffe test was used for post hoc comparisons because of the inequality in group sizes and the fact that it is a conservative procedure that allows for the examination of all possible group differences. In cases where the equality of variance assumption was not met, Tamhane’s T2 test was used in post hoc analyses. RA = Rare offender trajectory; LB = Late-bloomer trajectory; LR = Low-rate chronic trajectory; HR = High-rate chronic trajectory.

Equality of variance not assumed. Robust test of equality of means.

A log transformation was performed on the variable. Original means are displayed.

An inverse transformation was performed on the variable. Original means are displayed.

†

p < .10. *p < .05. **p < .01. ***p < .001.

A majority of youth in the LB, LR, and HR trajectories had charges for non-violent offenses compared with only approximately one third of those in the RA trajectory. Youth in the RA trajectory were also significantly older at the time of their first non-violent charges (RA > LB, p < .001; RA > LR, p < .001; RA > LR, p < .001). Similarly, those in the RA trajectory also had a lower frequency of non-violent charges (RA > LB, p < .001; RA > LR, p < .001; RA > LR, p < .001). At the same time, the lowest proportion of non-violent charges was evident for the RA trajectory (RA > LB, p < .001; RA > LR, p < .001; RA > LR, p < .001) and the LB trajectory (LB > LR, p < .05; LB > HR, p < .05).

A differential pattern emerged in terms of parameters measuring violent charges between the trajectory groups. First, youth in the RA trajectory did not differ from those in the LB trajectory at the age of first charges for violence, nor did those in the LR trajectory differ from youth in the HR trajectory (RA > LR, p < .001; RA > HR, p < .01; LB > LR, p < .01; LB > HR, p < .05). Youth in the RA trajectory had the least overall number of violent charges followed by those in the LB trajectory (RA < LB, p < .001; RA < LR, p < .01; RA < HR, p < .001; LB < HR, p < .05) and also had the lowest proportion of violent charges in their adolescent criminal history (RA < LB, p < .001; RA < HC, p < .05).

A differential pattern was also evident in terms of sexual offending parameters between the trajectory groups. The age at the first sexual offense was similar between youth in the RA and HR trajectories (RA > LB, p < .001; RA > LR, p < .05). Youth in the RA trajectory also had the least number of charges for sexual offenses (RA < LB, p < .001) and at the same time, had the highest proportion of sexual crimes in their criminal history (RA > LB, p < .001; RA > LR, p < .001; RA > HR, p < .001). Youth in the LB trajectory also had a higher proportion of charges for a sex crime compared to both chronic offending trajectories (LB > LR, p < .05; LB > HR, p < .05).

Referral Sex Offense Characteristics of Offending Trajectories

Table 4 shows that offense trajectories in adolescence were also differentially related to specific characteristics of the referral sexual offense. Youth in the RA trajectory were by far the most likely to have been referred based on strictly sexual charges (81.7%). Those in the LB and HR group were equally likely to have non-violent charges accompany sexual ones in their referral offense (41.8% and 48.0% respectively); however, youth in the LB trajectory were those most likely to also have referral charges for violence (41.8%). Differential patterns also emerged across the trajectories in terms of the age of the victim. Youth in the RA and LR trajectories were those most likely to have offended against a child (75.7% and 81.8% respectively), those in the HR trajectory were most likely to offend against peer-age victims (48.0%) and adult victims (36.0%). Finally, those youth in the LB trajectory were also among the most likely to offend against adult victims (27.3%).

Table 4.

Referral Sexual Offense Characteristics of the Four Offending Trajectories.

Characteristics of the referral sexual offense	Total sample (N = 217)	Group 1 (RA)	Group 2 (LB)	Group 3 (LR)	Group 4 (HR)	χ²(df); Cramer’s V
Characteristics of the referral sexual offense	Total sample (N = 217)	(n = 115)	(n = 55)	(n = 22)	(HC; n = 25)	χ²(df); Cramer’s V
Only sexual charges	65.9%	81.7%	43.6%	59.1%	48.0%	χ²(3) = 29.0***; 0.37
+ Non-violent charges	24.0%	11.3%	41.8%	18.2%	48.0%	χ²(3) = 28.1***; 0.36
+ Violent charges	21.2%	8.7%	41.8%	27.3%	28.0%	χ²(3) = 25.9***; 0.35
+Non-violent + Violent^a	11.1%	1.7%	27.3%	1.5%	24.0%	χ²(3) = 30.1***; 0.37
Familial victim(s)	33.2%	36.5%	30.9%	27.3%	28.0%	χ²(3) = 1.4 n.s.
Male victim(s)	25.8%	26.1%	34.5%	18.2%	12.0%	χ²(3) = 5.4 n.s.
Child (0-12) victim(s)	68.7%	75.7%	61.8%	81.8%	40.0%	χ²(3) = 15.1**; 0.23
Peer-age (13-17) victim	22.1%	19.1%	20.0%	13.6%	48.0%	χ²(3) = 11.4*; 0.23
Adult victim (18+)	16.1%	7.8%	27.3%	9.1%	36.0%	χ²(3) = 19.0***; 0.30

Note. RA = rare offender trajectory; LB = late-bloomer trajectory; LR = low-rate chronic trajectory; HR = high-rate chronic trajectory.

Low expected cell counts.

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

Discussion

The results of the current study support the utility of distinguishing broadly between criminally versatile and non-versatile adolescent sexual offenders (i.e., sex-only, sex-plus; Butler & Seto, 2002) given that over half of the sample was classified as “rare” offenders and the vast proportion of their limited offending over the course of adolescence was of a sexual nature. Yet, sexual recidivism among this sample of ASOs overall was extremely low, but varied substantially according to offending trajectories. As demonstrated in the recent longitudinal study of Lussier et al. (2012), in the current context, early-onset, chronic serious, and violent offending was not a necessary nor sufficient condition for the persistence of sexual offending (i.e., sexual recidivism), in this case in adolescence. The results also indicate that there is benefit to gaining a deeper understanding of the unfolding of both sex and non-sex offending criminal activity patterns of criminally versatile ASOs. These ASOs do not represent a homogeneous group of adolescents that all partake in “cafeteria style” offending. Furthermore, offending trajectories also provide information about the nature of sex offenses committed. This potentially sheds light on differential motivations for sex offending in adolescence.

Heterogeneity in Sex and Non-Sex Offending of Criminally Versatile ASOs

A small proportion of the sample (high-rate chronic pattern) showed an early onset of offending in adolescence, characterized by a progression to more serious and frequent crimes following the typical age–crime curve over the course of adolescence. These ASOs were criminally versatile and their sex offending appeared to have occurred after a sequence of escalating offense types in adolescence, beginning with an early adolescent onset of non-violent and violent offenses followed by sex offenses. This group had the lowest proportion of youth who were enrolled in school and this group was the most likely to have experienced a period of incarceration at some point in their adolescence. These findings are broadly consistent with previous research (i.e., Carpentier et al., 2011; Lussier et al., 2012), and demonstrate that for this portion of criminally versatile ASOs, sex offending in adolescence occurred after a process of escalation in severity of their offending (Elliott, 1994).

In addition to being the most criminally active group, they were also among those most likely to target peer and adult victims, compared with the other trajectory groups. Importantly, however, ASOs in the high-rate chronic group were more or less equally likely to target child, peer-age, and/or adult victims. In addition, they had the lowest rate of sexual recidivism (with the exception of ASOs in the rare offender trajectory). Taken together, these findings suggest that for these criminally versatile ASOs, their sex offending may not be primarily sexually motivated, but rather represents an extension of antisocial tendencies (i.e., violating the rights of others) escalating into their sexual lives. Importantly, however, this pattern only described approximately one quarter of the criminally versatile ASOs in the sample.

Adolescents characterized by the low-rate chronic offending trajectory showed a similar escalation to sex offending, but they were the most likely of the criminally versatile groups to have had a child victim. In effect, these were ASOs characterized by low frequency non-sexual offending who demonstrated a pattern of specifically targeting child victims. There are at least three possible explanations for this finding. The first is the possibility that the motivations for the sex offenses of ASOs on this trajectory are sexual in nature. In a study of a clinical sample of ASOs who targeted child victims, Dennison and Leclerc (2011) found that repeat ASOs were more likely to be characterized by histories of sexual abuse and inappropriate sexual behavior. In the current study, low-rate chronic ASOs were the most likely to target children and among the most likely to be sexual recidivists, along with ASOs on the late-bloomer trajectory. It is possible that early psychosocial deficits of these ASOs have had a specific impact on their sexual development.

Second, Indigenous ASOs were over-represented in this trajectory group, as were ASOs from rural and remote regions. Importantly, police charging practices may differ in remote regions of Australia compared with large urban centers, and in the current study, trajectories are based on finalized charges. At the same time, however, these findings also suggest there are broader situational and environmental circumstances that may, in part, also explain this pattern. Therefore, the third possible explanation is that in many remote Indigenous communities in Australia, differential opportunity structures and access to children (i.e., due to residential overcrowding), community breakdown, and lack of general supervision may explain, at least in part, why these ASOs were more likely to target children (Cale, 2014; Smallbone & Rayment-McHugh, 2013). In addition, these related social structural factors may also explain in part why youth in this trajectory group were among those with the highest rate of sexual recidivism. In effect, these findings also provide evidence about the importance of understanding persisting transitory risk factors for adolescent sex offending (Lussier et al., 2012).

Late-Bloomers and Ensnarement

The context in which sex offenses were committed also sheds some light on the interesting pattern that emerged regarding the late-blooming offending trajectory. First, low-level (i.e., property) offending characterized the initiation of criminal activity early in adolescence, and by mid-adolescence violent and sex crimes followed contemporaneously; these youth were among the most likely to have both violent and non-violent charges accompany their referral sex offense. Of the criminally versatile groups, they demonstrated the highest specialization in sex offending; partly a function of the fact they had minimal criminal histories until later in adolescence. In effect, their criminal histories did not suggest that sex offending was an escalation of a pattern of serious and violent offending, yet by the end of adolescence these youth had surpassed the frequency of offending of all three of the other offending trajectories. If we consider the longitudinal study of Lussier et al. (2012), this pattern is somewhat consistent with the late-bloomer trajectory uncovered in their study. One hypothesis may be that alcohol and drug abuse/intoxication characterize part of the late-bloomer profile, although it was not possible to test this in the current study. In addition, they were among the most likely to be sexual recidivists; potentially indicating these ASOs may be those more likely to exhibit continuity of sex offending into adulthood.

Although we did not follow these adolescents into adulthood, this may potentially represent the prototypical adolescent-limited antisocial youth who experienced ensnarement (Moffitt, 1993). Moffitt conceptualized the notion of ensnarement due to events in adolescence such as criminal convictions and drug addiction. This may be relevant in the current context when considering the potential ensnarement effect a sex offense would have on a prototypical “adolescent-limited” offender, especially given half of these adolescents were also incarcerated at some point in adolescence.

Naive Experimentation Versus the Young Male Syndrome

In the current study, the onset of adolescent criminal activity was more likely to be characterized by a sex offense for those adolescents in the rare offending trajectory. Consistent with previous accounts (e.g., Butler & Seto, 2002), this tends to support the notion that these were adolescents with fewer behavioral problems and more prosocial attitudes because they were the most likely to be employed at the time of their referral sex offense, and this group was also characterized by the highest proportion of youth enrolled in school. Furthermore, they were the most likely to have only sex offenses that typically occurred by mid-adolescence and thus a relatively shorter and homogeneous adolescent criminal activity pattern compared with the other groups (Van Wijk, Mali et al., 2007 ). In other words, these rare/late-onset offenders closely resemble the prototypic “sex-only” type; a majority had child victims, and their referral offenses were not typically characterized by any additional violent or non-violent charges.

Importantly, however, approximately one quarter also had peer-age or adult victims. On the one hand, “rare” offenders with peer-age victims may reflect the “young male syndrome” (e.g., Lalumière, Harris, Quinsey, & Rice, 2005; Seto & Barbaree, 1997; Wilson & Daly, 1985). Here, sex offending may represent temporary/contextual difficulties associated with finding a consensual sexual partner, particularly in adolescence, that lead to the use of coercion (e.g., “date-rape”). On the other hand, for “rare” offenders whose sexual offenses involved children, the offenses may represent motivations such as curiosity resulting in inappropriate sexual contacts (e.g., O’Brien & Bera, 1986). It is possible that early sexual behavior problems in childhood may have set the stage for sex offending of these adolescents (e.g., Burton, Nesmith, & Badten, 1997). Given that these adolescents had minimal criminal involvement, a focus on sexual development may provide some additional insight into their sex offending. One possibility is that in the absence of other major psychosocial deficits, early childhood sexual victimization has a unique impact on sexual development (e.g., Kendall-Tackett, Williams, & Finkelhor, 1993; Putnam, 2003) and possibly the emergence of adolescent sex offending against children in the context described here (Hunter, Figueredo, & Malamuth, 2010). However, it is also crucial to consider whether and to what extent other psychosocial deficits such as impulsivity, ego-centricity, under-developed abstract reasoning about the future consequences of behavior, and, to some extent, situational circumstances, such as the availability of a vulnerable potential victim, play a role in this context. For example, no group differences were evident in terms of whether there was a familial victim or whether the victim was male.

Limitations

While the results from the current study provide some insight into the offending trajectories of ASOs, and how these are related to the unfolding of criminal activity in general and sex offenses specifically, there are several methodological limitations that need to be taken into consideration. First, the study was based on a sample of youth who were referred to a clinical service in Australia for committing a sexual offense over approximately a 9-year period and it was not possible to control for potential cohort effects in the longitudinal analyses. Given the sample is unique in this sense any generalizations should also be made with caution. It also was not possible to explore dynamics of group offenses in the current study because the base rate of this phenomenon in the current sample was too low. Similarly, official criminal data often do not capture a wide range of undetected/unreported offending, and this limitation may be even more salient in the context of sex offending. This is undoubtedly an important dynamic of adolescent sex offenses to explore, especially considering links between antisocial involvement and group offenses among ASOs (e.g., Bijleveld & Hendriks, 2003). Finally, the data from the current study are based on finalized charges, and therefore also reflect, to a certain extent, the activities of police and more broadly the juvenile justice system of this particular state in Australia.

Conclusion

The current study provided detailed information about the unfolding of criminal activity across heterogeneous offending trajectories of ASOs, and also provided evidence that these trajectories are differentially associated with the characteristics of the sexual offenses they commit. While such a framework should not necessarily be interpreted as a panacea for the issue of classification, the current study demonstrates its utility as a conceptual framework by clinicians working with this population. It provides a framework to explore how individual differences and risk factors are related to individual offending patterns, and can provide insight into the onset and nature of sexual offenses committed by ASOs. The utility from a clinical perspective is that understanding how ASOs sexual and non-sexual criminal activity unfold over the course of adolescence, and how they are related to each other, can assist in developing and tailoring innovative and individualized treatment strategies. Methodologically, this is also a particularly innovative approach to examine the impact of treatment interventions on patterns of within-individual change beyond broad measures of recidivism (Nagin & Odgers, 2010), something future studies should consider.

Footnotes

Acknowledgements

We would like to acknowledge the support and assistance of the Queensland Police Service, Department of Justice and Attorney-General and Griffith Youth Forensic Service. The views expressed herein are solely those of the authors, and do not necessarily reflect the views or policies of these organisations. We would also like to express our thanks to the three annonymous reviews for their constructive feedback, as well as the editors, James Cantor and Anthony Beech.

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 research was supported under the Australian Research Council Discovery Projects Funding Scheme (Project: DP110102126).

Notes

References

Barbaree

H. E.

Hudson

S. M.

Seto

M. C.

(1993). Sexual assault in society: The role of the juvenile offender. In Barbaree

H. E.

Marshall

W. L.

Hudson

S. M.

(Eds.), The juvenile sex offender (pp. 1-24). New York, NY: Guilford.

Becker

J. V.

(1998). What we know about the characteristics and treatment of adolescents who have committed sexual offenses. Child Maltreatment, 3, 317-329.

Becker

J. V.

Kaplan

M. S.

Cunningham-Rathner

Kavoussi

(1986). Characteristics of adolescent incest sexual perpetrators: Preliminary findings. Journal of Family Violence, 1, 85-97.

Bijleveld

C. C. J. H.

Hendriks

(2003). Differential personality and background characteristics of juvenile group-and solo-sex offenders. Psychology, Crime & Law, 9, 237-245.

Blumstein

Cohen

Roth

Visher

(1986). Criminal careers and “career criminals.” Washington, DC: National Academy Press.

Bullens

van Wijk

Mali

(2006). Similarities and differences between the criminal careers of Dutch juvenile sex offenders and non-sex offenders. Journal of Sexual Aggression, 12, 155-164.

Burton

D. L.

(2000). Were adolescent sex offenders children with sexual behavior problems? Sexual Abuse: A Journal of Research and Treatment, 12, 37-48.

Burton

D. L.

Nesmith

Badten

(1997). Clinician’s views of sexually aggressive children: A theoretical explanation. Child Abuse & Neglect, 21, 157-170.

Butler

S. M.

Seto

M. C.

(2002). Distinguishing two types of juvenile sex offenders. Journal of the American Academy of Child and Adolescent Psychiatry, 41, 83-90.

10.

Cale

(2014). Innovations in the prevention and treatment of youth sexual violence and abuse in Australia. The Canadian Criminal Justice Association: Justice Report, 29, 6-9.

11.

Carpentier

Leclerc

Proulx

(2011). Juvenile sexual offenders: Correlates of onset, variety and aggravation of criminal behavior. Criminal Justice and Behavior, 38, 854-873.

12.

Department of Health and Aged Care. (2001) Measuring remoteness: Accessibility/Remoteness Index of Australia (ARIA). Revised Edition. Occasional Papers: Series No. 14.

13.

DeLisi

Neppl

T. K.

Lohman

B. J.

Vaughn

M. G.

Shook

J. J.

(2013). Early starters: Which type of criminal onset matters most for delinquent careers? Journal of Criminal Justice, 41, 12-17.

14.

Dennison

Leclerc

(2011). Developmental factors in adolescent child sex offenders: A comparison of nonrepeat and repeat sexual offenders. Criminal Justice and Behavior, 38, 1089-1102.

15.

Elliott

D. S.

(1994). Serious violent offenders: Onset, developmental course and termination. Criminology, 32, 1-22.

16.

Hunter

J. A.

Figueredo

A. J.

Malamuth

N. M.

(2010). Developmental pathways into social and sexual deviance. Journal of Family Violence, 25, 141-148.

17.

Hunter

J. A.

Figueredo

A. J.

Malamuth

N. M.

Becker

J. V.

(2003). Juvenile sex offenders: Toward the development of a typology. Sexual Abuse: A Journal of Research and Treatment, 15, 27-48.

18.

Jones

B. L.

Nagin

D. S.

(2007). Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociological Methods Research, 35, 542-571.

19.

Kendall-Tackett

K. A.

Williams

L. M.

Finkelhor

(1993). Impact of sexual abuse on children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113, 164-180.

20.

Lalumière

M. L.

Harris

G. T.

Quinsey

V. L.

Rice

M. E.

(2005). The causes of rape: Understanding individual differences in male propensity for sexual aggression. Washington, DC: American Psychological Association.

21.

Loeber

Le Blanc

(1990). Toward a developmental criminology. In Morris

Tonry

(Eds.), Crime and justice (pp. 375-473). Chicago, IL: University of Chicago Press.

22.

Lussier

van den Berg

Bijleveld

Hendriks

(2012). A developmental taxonomy of juvenile sex offenders for theory, research, and prevention: The adolescent-limited and the high-rate slow desister. Criminal Justice and Behavior, 39, 1559-1581.

23.

McCuish

E. C.

Corrado

Lussier

Hart

S. D.

(2014). Psychopathic traits and offending trajectories from early adolescence to adulthood. Journal of Criminal Justice, 42, 66-76.

24.

Miner

M. H.

(2012). Factors associated with recidivism in juveniles: An analysis of serious juvenile sex offenders. Journal of Research in Crime and Delinquency, 39, 421-436.

25.

Moffitt

T. E.

(1993). “Life-course-persistent” and “adolescent-limited” antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701.

26.

Nagin

D. S.

(1999). Analyzing developmental trajectories: A semiparametric, group-based approach. Psychological Methods, 4, 139-157.

27.

Nagin

D. S.

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

28.

Nagin

D. S.

Land

K. C.

(1993). Age, criminal careers, and population heterogeneity: Specification and estimation of a nonparametric, mixed Poisson model. Criminology, 31, 327-362.

29.

Nagin

D. S.

Odgers

C. L.

(2010). Group-based trajectory modeling in clinical research. Annual Review of Clinical Psychology, 6, 109-138.

30.

Nisbet

I. A.

Wilson

P. H.

Smallbone

S. W.

(2004). A prospective longitudinal study of sexual recidivism among adolescent sex offenders. Sexual Abuse: A Journal of Research and Treatment, 16, 223-234.

31.

O’Brien

M. J.

Bera

(1986). The PHASE typology of adolescent sex offenders. Salem, OR: Department of Human Resources, Children’s Services Division.

32.

Putnam

F. H.

(2003). Ten-year research update review: Child sexual abuse. Journal of the American Academy of Child and Adolescent Psychiatry, 42, 269-278.

33.

Seto

M. C.

Barbaree

H. E.

(1997). Sexual aggression as antisocial behavior: A developmental model. In Stoff

D. M.

Breiling

Maser

J. D.

(Eds.), Handbook of antisocial behavior (pp. 524-533). New York, NY: John Wiley.

34.

Seto

M. C.

Lalumière

M. L.

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

35.

Smallbone

Rayment-McHugh

(2013). Preventing youth sexual violence and abuse: Problems and solutions in the Australian context. Australian Psychologist, 48, 3-13.

36.

van der Geest

Blokland

Bijleveld

(2009). Delinquent development in a sample of high-risk youth: Shape, content, and predictors of delinquent trajectories from age 12 to 32. Journal of Research in Crime and Delinquency, 46, 111-143.

37.

van Wijk

Loeber

Vermeiren

Pardini

Bullens

Doreleijers

(2005). Violent juvenile sex offenders compared with violent juvenile nonsex offenders: Explorative findings from the Pittsburgh Youth Study. Sexual Abuse: A Journal of Research and Treatment, 17, 333-352.

38.

Van Wijk

Mali

Bullens

Vermeiren

(2007). Criminal profiles of violent juvenile sex and violent juvenile non-sex offenders. Journal of Interpersonal Violence, 22, 1340-1355.

39.

van Wijk

Vermeiren

Loeber

Hart-Kerkhoffs

Doreleijers

Bullens

(2006). Juvenile sex offenders compared to non-sex offenders: A review of the literature 1995–2005. Trauma, Violence, & Abuse, 7, 227-243.

40.

Vizard

Hickey

McCrory

(2007). Developmental trajectories associated with juvenile sexually abusive behaviour and emerging severe personality disorder in childhood: 3 - year study. British Journal of Psychiatry, 190, s27-s32.

41.

Weakliem

D. L.

Wright

B. R. E.

(2009). Robustness of group-based models for longitudinal count data. Sociological Methods and Research, 38, 147-170.

42.

Wilson

Daly

(1985). Competitiveness, risk taking, and violence: The young male syndrome. Ethology and Sociobiology, 6, 59-73.

43.

Zimring

F. E.

Piquero

A. R.

Jennings

W. R.

(2007). Sexual delinquency in Racine: Does early sex offending predict later sex offending in youth and young adulthood. Criminology and Public Policy, 6, 507-534.