Juvenile Sex Offending Through a Developmental Life Course Criminology Perspective

DLC Research and Traditional Research on Juvenile Sexual Offending

DLC criminology contrasts with the more traditional view about juvenile sexual offending in at least two significant ways. First, to date, clinical research has been focused on the identification of traits associated with juvenile sexual offending, such as psychopathic traits, insecure attachment style, high sexual drive and sexual compulsivity, negative masculinity, and poor social competence to name a few (e.g., Hunter, Figueredo, Malamuth, & Becker, 2003; Knight & Sims-Knight, 2003; Marshall & Marshall, 2000; Seto & Lalumière, 2010). Although this approach provides a clinical profile of juveniles involved in sexual offenses, it does little to inform about the origins and the developmental course of juvenile sexual offending. Also, the trait-based approach cannot be reconciled with the low sexual recidivism rates observed across studies (e.g., McCann & Lussier, 2008). Current explanatory models of juvenile offending can be enhanced by incorporating juveniles’ past developmental antecedents such as past history of sexual and nonsexual delinquency as well as exposure to risk and protective factors at different developmental stages (e.g., pregnancy, birth, infancy; see Smallbone, 2006). These models can also be improved by incorporating the probable future development of delinquency as well as the expected exposure to risk and protective factors at different development stages in the future (e.g., late adolescence, emerging adulthood, adulthood). If the understanding of the development of juvenile delinquency has drastically improved in the past three decades with the advent of prospective longitudinal studies (e.g., DeLisi & Piquero, 2011; Piquero, Farrington, & Blumstein, 2003), the understanding of the development of juvenile sexual offending remains equivocal.

Second, several studies have been conducted to determine what is so special about JSO and whether they are a distinct group of offenders (e.g., Seto & Lalumière, 2010). The main strategy has been to compare a clinical sample of JSO with another sample of juvenile nonsexual offenders (JNSO) on a series of indicators. This comparative approach aims to identify between-individual differences that could be specifically associated with sexual offending during adolescence. The almost exclusive use of clinical or legal samples does not provide a representative view of JSO. It could be reasonably argued that the odds of entering a clinical or a legal sample is cumulative and related to the frequency and/or the seriousness of the offending behavior. In other words, juveniles having committed more or more serious sexual offenses are more likely to be found in those samples. Furthermore, this strategy contributes to the portrayal of JSO as a homogeneous group, and empirical evidence suggests that the “average” JSO is not representative enough to guide theoretical, research, and policy development. There is sufficient empirical evidence showing significant within-group variations among JSO along several dimensions and constructs (e.g., Butler & Seto, 2002; Knight & Prentky, 1993; McCuish, Lussier, & Corrado, 2014). Various classification models have been proposed to account for the heterogeneity among juveniles involved in sexual offenses, but such models are not informative about the developmental course of sexual offending and associated risk factors (e.g., Becker, 1998; Butler & Seto, 2002). In other words, there is a need for a research framework that can simultaneously take into account the between- and within-group heterogeneity characterizing juveniles involved in sexual offenses while allowing the identification of the different patterns of development associated with juvenile sexual offending.

This is what the DLC perspective is proposing, and, in many ways, it is consistent with a person-oriented approach (Bergman & Magnusson, 1997). Developmentalists introduced the person-oriented approach to favor a paradigm that shifts the focus from variables to individuals. The variable-oriented approach is based on aggregate data and average series across individuals, which can misrepresent individual patterns of development (von Eye & Bergman, 2003). Consequently, conclusions from variable-oriented studies might not apply to all or most individual cases. If the variable-oriented approach aims to identify and isolate the effect of a single individual-level factor, the person-oriented approach is concerned with the individual as an undivided whole (Bergman & Trost, 2006). For example, from a variable-oriented perspective, one could examine the role of pornography use on sexual offending, but from a person-oriented approach, this is problematic given that the relationship between pornography use and sexual offending could be reciprocal and the effect could change over time and across developmental stages (i.e., childhood, adolescence, adulthood). Therefore, to better account for the heterogeneity of individual development, the person-oriented approach focuses on the disaggregation of information and the identification of individual longitudinal patterns, with the understanding that some patterns are occurring more often than others (e.g., Lussier & Davies, 2011). In that regard, development can be conceptualized as a process characterized by states that can change over time (Bergman & Magnusson, 1997). Thus, repeated measurements become pivotal for the identification of continuity and change as individuals age. As such, the modeling of human development needs to account for the diversity of intercepts (i.e., initial level of the behavior) and slopes (i.e., trends of the behavior). To that end, nonlinear modeling becomes crucial to detect patterns of individual development. Configuration of variables, time-dependent variables and effect, longitudinal data with repeated measurements, nonlinear patterns, and changes over time best characterize the person-oriented approach. This perspective is in sharp contrast to the cross-sectional nature of studies typically conducted in the field of juvenile sexual offending. This is not to say that the person-oriented approach is superior to the variable-oriented one, but it provides a complementary perspective on human development (Bergman & Trost, 2006).

Juvenile Offending: A Dynamic Process View of Development

DLC criminologists have proposed theoretical models of offending that recognize human development, developmental stages, life events and transitions, and how such dynamic processes influence the development of offending over time (e.g., Farrington, 2005; Moffitt, 1993; Thornberry, 2005). Such models aim to explain both between- and within-individual stability and changes in offending over time (Loeber & LeBlanc, 1990). ¹ Unlike the measure of sexual recidivism, which only takes into consideration whether an individual will be rearrested (or convicted) for a sexual crime during some follow-up period, DLC researchers are concerned by the entire longitudinal sequence of offenses committed by an individual. This sequence of offending is best described by a series of boundary or time-related parameters and generic descriptors of offending (Table 1). On one hand, boundary or time-related parameters include the age of onset of offending, the age of offset or age at which the behavior stops, and the time in between (i.e., duration of offending). These parameters help contextualize the developmental stages (e.g., infancy, early and late childhood, early, middle and late adolescence, emerging adulthood, and so on) during which offending starts, unfolds, and terminates. On the other hand, generic descriptors refer to quantitative and qualitative aspects of offending over time as the offending unfolds. Quantitative aspects of offending refer, among other things, to the number of offenses committed (i.e., frequency) and the number of different types of offending someone has committed (i.e., variety). Qualitative aspects of offending refer to the nature of offending over time, such as escalation and seriousness. This is in sharp contrast with measures of sexual recidivism that do not specify the type of offense, the number of victims, the duration and frequency of offending, and so on (Lussier & Cale, 2013). In sum, if sexual recidivism informs of the probability that an adolescent may sexually reoffend during some specific time period, developmental parameters inform the field about the longitudinal quantitative and qualitative sequence of offending over time and the context in which the sequence unfolds.

OPEN IN VIEWER

Table 1.

Conceptual Parameters of Development.

Parameters	Definition
Boundary-type parameters
Age of onset	Age at first sexual crime
Age of offset	Age at last sexual crime
Duration	Time elapsed between age at first and last offense
Generic-type parameters
Frequency	Number of victims/number of sexual crime events
Persistence	Repetition of sexual offending over time
Versatility	Number of different sexual offenses committed

Boundaries and generic developmental parameters of offending can be chronologically organized to describe three developmental processes of offending (Figure 1), that is, activation, course or escalation, and desistance (LeBlanc & Fréchette, 1989). Onset marks the debut of offending, which is a precursor of persistence. For offending to become frequent and diversified, it needs first to persist. The activation phase is the most empirically examined stage of general offending and encompasses three developmental processes: (a) stabilization, that is, the impact of onset on the persistence of offending over time; (b) acceleration, that is, the impact of onset on the frequency of offending over time; and (c) diversification, that is, the impact of onset on the variety of offending over time. The escalation stage refers to the processes by which persistent offending is more likely to become chronic (i.e., chronicity) and move along a developmental sequence of increasingly serious offenses (i.e., progression). Finally, desistance is described by developmentalists as a slowing down of offending until complete termination (e.g., LeBlanc & Fréchette, 1989). Under such conceptualization, desistance refers to three processes: (a) culmination, that is, where offending stops progressing; (b) specialization, that is, a slowing down of the variety of offending or the gradual tendency to restrict offending to specific offenses; and (c) deceleration, that is, the gradual slowing down of frequency of offending. In other words, desistance is conceived as a process that may take some time and involves lapses and relapses until complete termination of offending. ²

OPEN IN VIEWER

Figure 1.

A process view of offending: An organizing framework.

The DLC approach provides concepts and processes that can shed some light on the origins and the developmental course of sexual offending over time. It does not suggest that all adolescents involved in sexual offenses have a sustained pattern of sexual offending comprised of multiple victims or multiple sexual crime events. In fact, it could be reasonably argued the DLC is not a suitable framework to study juvenile sexual offending given that only a small subgroup of adolescents, often overrepresented in clinical settings and in the juvenile justice system, shows such a pattern of sexual offending. The same argument can be raised regarding general delinquency and serious offending given that the most common patterns of juvenile delinquency are short-lived and involve trivial offenses (e.g., Loeber & LeBlanc, 1990). Knowing that, for most juveniles, sexual offending is limited to a single event is as important and informative for policy and theoretical development as knowing about the multiple sexual offenses committed by a relatively small subgroup of JSO. The DLC approach, first and foremost, provides a common language for researchers and practitioners from various disciplines to communicate and discuss more precisely about juvenile sexual offending, its nature, prevalence, and development. It also addresses a limitation of existing classification models of JSO that are not informative of longitudinal patterns of the behavior.

The DLC approach also provides concepts and processes that can be operationalized and measured to inform about patterns of sexual offending among juveniles. In fact, it is suggested that there is great heterogeneity in sexual offending across stages of offending (Figure 2). At one end of the continuum, most youth are hypothesized to be late-starters (or adolescent-onset), show little or minimal escalation or progression in sexual offending, and desist almost immediately from sexual offending. Is it also hypothesized that, at the other end of the continuum, there is a small group of adolescents who are characterized by an early onset (i.e., childhood-onset) who are more inclined to repeat and show evidence of escalation in terms of the nature of their sexually violent and/or abusive behaviors and whose desistance is more gradual over time. Such heterogeneity is not accounted for by existing explanatory models of juvenile sexual offending. In this article, a DLC framework is proposed to explore and describe the heterogeneity of sexual offending patterns among all youth involved in sexual offenses and not just those found in treatment programs or in the juvenile justice system. The next sections provide a review of the current state of knowledge on the developmental parameters of sexual offending.

OPEN IN VIEWER

Figure 2.

Hypothesized developmental stages of sexual offending and heterogeneity across stages.

The Prevalence of Juvenile Sexual Offending

Prevalence of offending refers to the proportion of individuals in a given group committing the behavior at least once during some specific time period. Although there is a plethora of prevalence studies of general juvenile delinquency (e.g., Piquero et al., 2003), these studies are not very informative of the prevalence of sexual offending given that self-report measures of delinquency do not include many items related to sexual offenses (e.g., rape, sexual assault). Consequently, these measures do not capture the whole spectrum of sexual offenses committed by juveniles and are likely to underestimate the actual prevalence of sexual offending. Surprising, given how significant concerns for juvenile sexual offenses have grown in recent years, is the lack of a self-report scale designed specifically to measure juvenile sexual offending. Therefore, most estimates of juvenile sexual offending come from official data (e.g., arrests). Police data show that the prevalence rate of sexual offending peaks during the teenage years, more precisely during the early to middle period of adolescence (e.g., Kong, Johnson, Beattie, & Cardillo, 2003). Official data also show that juvenile sexual offending is mostly a male phenomenon with more than 95% of the accused being males (Ryan, Miyoshi, Metzner, Krugman, & Fryer, 1996). ³ For example, the birth cohort studies conducted in England/Wales by the Home Office show that the overall official prevalence of juvenile sexual offending varied between about 0.3% and 0.5% (Marshall, 1997). Other birth cohort studies provide relatively similar estimates when controlling for the definition of sexual offense used (e.g., Zimring, Piquero, & Jennings, 2007). In the Cambridge longitudinal study of 411 boys followed from age 8, while 41% of them have been convicted at least once (for any crime), only 2.5% of the sample had a conviction for a sexual crime up to age 50 (Piquero, Farrington, Jennings, Diamond, & Craig, 2012). The study highlighted the presence of only four convicted JSO, for a prevalence of about 1%. More recently, using data from the 1984 Dutch birth cohort study, based on police data, Lussier and Blokland (2014) estimated that the prevalence of juvenile sexual offending was 0.4%. In sum, in spite of historical, sociocultural, and legal differences across studies, it is somewhat surprising to observe a certain regularity in the prevalence estimates of juvenile sexual offending using official data across jurisdictions and time periods. These numbers show that for every 1,000 individuals of a birth cohort, between 3 and 5 are arrested for a sexual offense during adolescence. This number illustrates how infrequently the phenomenon comes to the attention of the police and, relatedly, how difficult it is for researchers to isolate and identify possible explanations.

Researchers would argue that juvenile sexual offending is largely underestimated by official data and that other means for estimating the prevalence is necessary. Given that the field lacks a valid and reliable measure of juvenile sexual offending, it is not surprising that few empirical studies have estimated the prevalence of juvenile sexual offending using self-report measures. Recently, Kjellgren, Priebe, Svedin, and Langström (2010) estimated the self-reported prevalence of sexually coercive behaviors in a Swedish sample of high school students aged between 17 and 20 years. For this study, sexually coercive behaviors were defined as “talked someone into, used pressure or forced somebody to masturbate them, to have sexual intercourse, oral sex or anal sex” (p. 1162). In total, 5% of the sample reported any lifetime coercive sexual behaviors. Kjellgren et al. (2010) also reported that not only several key criminogenic risk factors of antisocial behavior were related to sexually coercive behaviors by males (e.g., aggression, risk-taking, substance use, depression, poorer perceived parental care) but also risk factors that appear more specific to sexual offending such as the endorsement of rape myths, frequent use of pornography, early age of onset of sexual intercourse, and having more sexual partners. This estimated prevalence found by Kjellgren et al. (2010) is about 12 times higher than those reported in studies using official data (e.g., Marshall, 1997). Comparatively speaking, data with adults suggest that the self-report of sexual coercion is up to about 50 times higher than what is reported in official data (Lussier & Cale, 2013). These comparative estimates are hardly reliable given that they are based on data stemming from different samples and using different definitions of sexual violence and abuse. Prevalence estimates using both official and self-report data on the same samples are needed to better estimate the prevalence of sexual violence and abuse across developmental stages. To achieve this, first and foremost, the development of measures of juvenile sexual offending should be a high priority among researchers.

The Age of Onset of Juvenile Sexual Offending

The age of onset refers to the age at first sexual offense and it is particularly important because it marks the origins of the behavior (Lussier & Healey, 2010). The limited research reporting on the age of onset has been conducted with clinical samples of juveniles and these studies tend to report an average age of 14 years (e.g., Carpentier, Leclerc, & Proulx, 2011; Jacobs, Kennedy, & Meyer, 1997; Ryan et al., 1996). Similar findings have been reported in a birth cohort study (Lussier & Blokland, 2014) as well as in an at-risk community sample (Lussier, Blokland, Mathesius, Pardini, & Loeber, 2014), reinforcing the idea that middle adolescence is a critical period for the onset of sexual offending. This corresponds to a period, at least for U.S. schools, where children going through puberty typically transition from eighth (middle school) to ninth grade (high school). This age is also associated with the onset of reckless and rebellious behaviors such as vandalism at school, driving without a driver’s license, using hard drugs, and prostitution (see LeBlanc & Loeber, 1998). Also notable, the average age of onset of juvenile sexual offending is younger than the average age of first sexual intercourse (Zimmer-Gembeck & Helfand, 2008).

In line with this observation, prospective longitudinal research has shown that an early onset of sexual intercourse is a developmental risk factor of juvenile sexual offending (e.g., Lussier et al., 2014). There is clearly a need to examine the onset of sexual offending in the context of what constitutes normal sexual development during early to middle adolescence. Furthermore, longitudinal studies of general offending have shown that self-reported onset occurs significantly earlier than official onset age (e.g., Loeber & LeBlanc, 1990. Research conducted with JSO also suggests such a trend, but the length of the discrepancies between the two onset measures remains unknown. In addition, sexual offending encompasses a broad range of manifestations, and empirical studies suggest that the age of onset may vary across these behaviors. In that regard, the Zolondek, Abel, Northey, and Jordan (2001) findings show that atypical sexual behaviors may precede in some cases the onset of sexual offending. Unfortunately, such a hypothesis has not been empirically examined. Also, there is some but limited evidence suggesting that juveniles who offend against significantly younger victims tend to be significantly younger than those offending against victims their own age (Groth, 1977; Hendriks & Bijleveld, 2004). These findings may speak, among other things, to the physical and psychological maturity necessary to create a context conducive to a sexual offense against adolescents as opposed to children. Clearly, existing studies highlight the need for studies combining official and self-report data that describe the onset of sexual offending. Studies using both methodologies with adult sexual offenders show, on average, a 7-year gap between the actual and official ages of onset of convicted adult male sexual offenders (Lussier & Mathesius, 2012), and it is unclear how this actual–official onset gap characterizes the development of sexual offending in JSO. This is important given that adolescents may be active for quite some time before their sexual offending is assessed and intervention strategies are proposed. This is especially relevant for the identification of the early-onset group of JSO.

For developmentalists, the identification of early onset, persistent offenders is pivotal given that this group is at risk of a long-term pattern of chronic, violent, and versatile offending (e.g., LeBlanc & Loeber, 1998; Moffitt, 1993). Furthermore, researchers have highlighted the fact that onset ages of juvenile delinquency are not distributed in a bimodal way as the early–late onset categorization suggests, but follows a relatively normal distribution (e.g., Thornberry, 2003), raising issues about whether an early-onset group is a meaningful concept. It is unclear, however, whether such considerations apply specifically to sexual offending. Given that the typical official onset of juvenile sexual offending occurs during middle adolescence, researchers have been concerned with a group whose sexual offending is activated in an earlier developmental stage, that is, childhood.

There has been limited research estimating the prevalence of early, childhood-onset sexual offending. Findings based on clinical samples of JSO suggest that a small group starts offending during childhood, with estimates ranging between 5% and 26% across studies (Awad & Saunders, 1991; Carpentier et al., 2011; Ryan et al., 1996). A retrospective file review of 280 cases referred to a community assessment and treatment service by Vizard and colleagues (Vizard, Hickey, French, & McCrory, 2007; Vizard, Hickey, & McCrory, 2007) examined and compared early- and late-onset offenders. Several developmental differences were found between the two groups, suggesting that the early-onset group was more likely to come from disorganized familial environments, to have been neglected and victimized, and more likely to be disruptive, impulsive, hyperactive, and to show a difficult temperament. The early-onset group, as opposed to the late-onset, was more likely to have abused male victims and less likely to use verbal coercion or to penetrate their victim. Of importance, Vizard and colleagues defined individuals with an early onset of sexually abusive behaviors as those who (a) had committed a sexually abusive act prior age 11 and (b) who also continued to sexually abuse during adolescence. In other words, their early-onset group excluded those who did not persist into adolescence. On closer examination of Vizard et al.’s numbers, it can be estimated that 34% of early-onset children did not persist into adolescence and were excluded from the analyses. The exclusion of this early-onset desister or “recovery” group might have contributed to the clinical portrayal of the early-onset group as being more dysfunctional than the late-onset group. This exclusion also raises issues about who this early-onset desister group is, what their clinical profile is, and whether this group is significantly different from the early-onset persisters.

Although these findings are promising, they should be interpreted with caution. These clinical studies are based on retrospective data of the age of onset and, consequently, may overestimate the continuity between atypical childhood sexual behaviors and juvenile sexual offending. In fact, although the identification of early-onset offenders may seem straightforward, it raises the issue of identifying criteria defining who the early onset offenders are and whether they can be prospectively identified. It also raises conceptual issues as to whether the term “early-starter” is appropriate in the absence of detailed information about normative and nonnormative sexual behaviors at the earliest developmental stages ⁴ (for a discussion, see Chaffin et al., 2008). Whereas retrospective studies suggest that atypical childhood sexual behaviors may be early signs of persistent juvenile sexual offending, prospective longitudinal studies may show that most children with atypical childhood sexual behaviors do not become JSO. For example, a prospective longitudinal study by Carpentier, Silovsky, and Chaffin (2006) showed that only 2% of a group of children with sexual behavior problems who took part in a cognitive-behavioral treatment had a sexual abuse perpetration report over a 10-year follow-up period. The current state of knowledge remains equivocal as to what are normative and nonnormative childhood sexual behaviors and what may appear to some as atypical sexual behaviors (e.g., a 4-year-old wanting to touch other children’s genitals) may in fact be part of a relatively normative pattern of sexual development. To uncover the presence of a developmental pathway (or pathways) of normative and nonnormative sexual development, prospective longitudinal data on sexual behaviors at the earliest developmental stages are needed.

Persistence and Continuity in Sexual Offending

Persistence of offending encompasses a behavioral (i.e., repetition of the behavior) and a temporal dimension (i.e., duration of offending). Prior clinical studies using retrospective data measured persistence using mainly behavioral indicators of prior sexual offenses and tend to report a relatively high level of persistence. For example, the study by Groth (1977) with a small sample of JSO assessed at a forensic mental health facility showed that more than 60% of them had a prior sexual offense. Other reports provided results consistent with Groth’s portrayal of JSO as persistent sexual offenders (e.g., Awad & Saunders, 1991; Awad, Saunders, & Levene, 1984; Becker, Kaplan, Cunningham-Rathner, & Kavoussi, 1986; Fehrenbach, Smith, Monastersky, & Deisher, 1986). These findings led some to conclude that “sex offenders have established a repetitive pattern of deviant behavior prior to an arrest” (Boyd, Hagan, & Cho, 2000, p. 139). The image of the persistent offender that emerged from clinical studies in the 1980s appears to have been a function of the selective nature of the samples used to conduct the studies and the differential criminal justice response to juvenile sexual offending prevailing at the time. ⁵

Indeed, the portrayal of JSO as persisters does not fit the general picture provided by other empirical studies (e.g., Doshay, 1943; Way & Urbaniak, 2008; Worling & Långström, 2006). For example, in the Ryan et al. (1996) study of more than 1,000 JSO, only 7.5% had previously been charged with a sexual offense and less than half of them had been found guilty. Prospective longitudinal studies also portray persistence of juvenile sexual offending as something rather unusual. A meta-analysis has shown that sexual recidivism rates typically reported in longitudinal studies tend to vary between 5% and 10% (McCann & Lussier, 2008). These findings do not appear to be the result of a relatively short follow-up period following their release (about 3-5 years) as longitudinal research with JSO has shown that it is uncommon for them to sexually reoffend after 5 or 6 years following their release (Langström, 2002). Few empirical studies report relatively higher base rate of sexual recidivism, but those are probably attributable to the nature of their sample (e.g., Hagan, Gust-Brey, Cho, & Dow, 2001; Langström & Grann, 2000). Reliance on self-report data, as opposed to the often used official data on offending, provides similar conclusions that persistence of sexual offending is an unusual phenomenon. In that regard, Bremer (1992) reported a 6% reconviction rate in a sample of serious JSO, but the recidivism rate rose to 11% when based on self-reports. Therefore, while the use of official data underestimates the true recidivism rates, it cannot explain why the vast majority are not rearrested for a sexual crime.

A related concept of persistence is the continuity of sexual offending in adulthood. Research has shown how important it is to separate retrospective and prospective continuity. The importance of distinguishing between retrospective and prospective continuity was highlighted by Robins (1978). Robins observed that while most antisocial children do not go on to become antisocial adults, adult antisocial behaviors virtually requires prior antisocial behaviors. In other words, she argued that highly antisocial behaviors rarely or never arose de novo in adulthood. These conclusions, therefore, suggest that while most JSO do not go on to become adult sexual offenders, most adult sexual offenders were previously JSO. In line with empirical studies examining persistence, the current state of empirical knowledge shows that most JSO do not become adult sexual offenders. In that regard, an examination of the Philadelphia birth cohort longitudinal data led Zimring, Jennings, Piquero, and Hays (2009) to conclude that “the most striking feature of the Philadelphia data was the lack of overlap between juvenile sexual offending and adult sex offending” (p. 65). They observed that for every 10 JSO, only 1 had a sexual offense record in adulthood. The group of JSO accounted for roughly 8% of the total adult sexual offenses committed by the entire members of the birth cohort. The lack of continuity found in the Zimring et al. (2009) studies (see also Zimring et al., 2007) were confirmed in another birth cohort study. Indeed, Lussier and Blokland (2014) reported in the 1984 Dutch birth cohort study that, while the prevalence of adult sexual offending in non-JSO was 0.5% (n = 360), it rose to 3% for one-time JSO (n = 8) and 12% for JSO who sexually reoffended during adolescence (n = 9). Percentages, however, can be somewhat misleading in the absence of absolute numbers on the continuity of sex offending. Indeed, the study shows that 360 of the 377 adult sex offenders found in the study had no prior record for a sex crime during adolescence while 17 did. In other words, only 4.5% of adult sex offenders were previously JSO. Therefore, the overlap between juvenile and adult sexual offending, as far as official data show, is minimal.

Albeit some continuity between the two, juvenile and adult sexual offending are relatively two distinct phenomena. The empirical knowledge explaining why they are relatively independent of one another is only tentative and the DLC approach could help shed some light on this. From a DLC perspective, it could be argued that the motives for sexual offending are developmental stage–specific; that is, factors that are specific to adolescence may explain juvenile sexual offending while factors that are more specific to adulthood may explain adult sexual offending. It could also be argued that even within the period of adolescence (i.e., early, middle, and late adolescence), different developmental factors (e.g., puberty, onset of sexual contacts, substance use, peer pressure, and delinquency) may put individuals at risk of committing a sexual offense.

The Frequency of Sexual Offending

Criminologists have long been concerned with the identification of chronic juvenile offenders (e.g., Loeber & Farrington, 1998). Such concerns arose from the work of Wolfgang et al. (1972) who identified in the Philadelphia birth cohort study that a small group of adolescents (about 6%) were responsible for more than half of all crimes committed by all members of the birth cohort and an even larger proportion of more serious offenses, including rape. These numbers have been replicated in many countries, across gender, using both self-report and official data (Piquero et al., 2003). The identification of chronic offenders raises two key issues. First, chronic offending refers to the frequency of offending and requires an operational definition to identify chronic offenders (e.g., five or more arrests during adolescence), a definition without any theoretical rationale and regarded by many as arbitrary. Second, this definition is not very useful from a prevention standpoint given that one needs to chronically offend to be identified as a chronic offender. It is unclear, however, how these two considerations translate to chronic sexual offending. Unlike some nonsexual offenses, the frequency of sexual crimes refers to two dimensions of offending (Lussier & Cale, 2013).

Traditionally, the frequency of sexual offending has referred to the number of different people an offender has sexually victimized. It can also refer, however, to the number of sexual crime events, that is, the total number of different times or occasions an individual has sexually abused his or her victim(s). This approach provides a more precise picture of the extent of the sexual offending, but is more difficult to estimate, especially for cases involving multiple and repeated victimization over time. Research with adults conducted by Lussier, Bouchard, and Beauregard (2011) has shown that it is more advisable to distinguish persistent offenders who pursue a victim-oriented strategy (i.e., multiple victims victimized on a very limited number of occasions) from those pursuing an event-based strategy (i.e., limited number of victims victimized multiples times). It is unclear, however, if such dichotomy also applies to JSO.

There has been limited research on the frequency of sexual offending in JSO. In a sample of more than 300 convicted JSO, Carpentier et al. (2011) found that the self-reported average number of victims was about 2, ranging from 1 to 16. These offenders primarily victimized children exclusively (close to 60%), followed by peers or adults exclusively (about 25%). Jacobs et al. (1997) reported similar numbers for a sample of 78 JSO in a residential specialized treatment program. It is unclear, however, if they referred to the number of events or the number of different victims or both. The inspection by Becker et al. (1986) of self-report data from a sample of 67 juveniles referred to a clinic revealed that the frequency of offending when based on the number of victims is about 1, but rose to about 2 for adolescents having sexually abused a young boy. For those having offended against similar age female victims, the average number of victims was about 1.5. Data regarding the average number of events were quite similar with the exception of those having offended against intrafamilial victims as well as males of a similar age for whom the number of events per victim was much higher. Furthermore, Miranda and Corcoran (2000) compared the frequency and duration of the abuse between a small sample of JSO and a small sample of adult sexual offenders. The findings showed that, on average, juveniles had been involved in about 2 child sexual abuse incidents over a period of about 9 months and had offended less frequently and over a shorter period of time compared with adult sexual offenders. Taken together, these findings suggest that most JSO offend against 1 or 2 victims. The number of victims remains relatively low unless sexual offending is against boys in an intrafamilial context.

Other research on JSO suggests that the frequency of sexual offending may be higher than what is typically reported in empirical research (e.g., Ryan et al., 1996; Zolondek et al., 2001). This variation is probably due to sample and sampling differences as well as the presence of a subgroup of offenders disproportionately more criminally active than others. This group of juveniles is more likely to be overrepresented in clinical settings or in juvenile detention centers. For example, a study by Wieckowski, Hartsoe, Mayer, and Shortz (1998) revealed that a small sample of adjudicated JSO had on average a little more than 3 hands-on victims and were involved on average in more than 70 hands-on events (median = 12). In contrast, the average number of hands-off victims reported was well more than 30 (median = 14) while the mean number of events was more than 100 (median = 51). This is suggestive that frequency is higher for hands-off than hands-on sexual offenses. In other words, it is more likely that hands-off offenses involve multiple simultaneous victims as opposed to hands-on offenses. Wieckowski et al.’s (1998) findings suggest that some forms of sexual offending may be more conducive to repetition because it may involve concealing behaviors that may go unnoticed or might be perceived as minor or not serious enough to be reported to the authorities. The findings also show a significant gap between the mean and the median of offending frequency, suggesting an asymmetric distribution of offending frequency. Such asymmetric distribution is likely to occur when a small group of individuals present a frequency of offending that is much higher than most other offenders included in the sample. ⁶ To date, no studies have examined the prevalence of chronic sexual offending and whether chronic JSO present a clinical profile distinct from other nonchronic JSO. The frequency of sexual offending examined in past research does not take into consideration the time at risk or how long offenders were active, or as criminologists call it, the lambda (i.e., offending rate; see Deslauriers-Varin & Beauregard, 2014).

Crime Versatility and Specialization

Versatility in offending has often been described and defined by opposition to specialization. Developmentalists, however, define versatility and specialization as two distinct phases of persistent offending (e.g., Loeber & LeBlanc, 1990). Indeed, persistent offending is said to become quite diverse and versatile and, with experiences and contingencies, to become more patterned and specialized over time. Such processes have been shown with adult sexual offenders (e.g., Miethe, Olson, & Mitchell, 2006) but have not been examined with JSO. The specialization/versatility issue first refers to whether JSO tend to limit their offending to sexual crimes (i.e., specialization) or whether their offending, if persistent, is not limited to sexual crimes (i.e., versatility). Studies conducted in clinical settings as well as those with samples of adjudicated juveniles have shown that between 40% and 60% of JSO have a history of nonsexual offenses (e.g., Awad et al., 1984; Lussier & Blokland, 2014; Ryan et al., 1996). It is then safe to say that among adjudicated JSO, a substantial proportion of recidivists have been through the youth justice system before for nonsexual crimes. For example, in Ryan et al.’s (1996) examination of the criminal history profile of more than 1,000 JSO, close to 28% were known to have three or more nonsexual offenses. The most prevalent offenses were shoplifting, theft, assault, runaway, and vandalism, all of which are common among juvenile offenders. Consequently, to recognize the presence of a subgroup of JSO involved in nonsexual offenses, researchers have suggested distinguishing JSO into sex-only and sex-plus groups.

Butler and Seto (2002) distinguished JSO who had been charged only with sexual crimes (referred to as sex-only; n = 22) from JSO who had been charged with sexual crimes and other crimes (referred to as sex-plus; n = 10). The two groups of JSO were compared with a group of nonsexual juvenile offenders. Butler and Seto (2002) found few significant differences between these two groups of juvenile offenders. They did find, however, that the sex-plus group was more similar to nonsexual juvenile group than the sex-only group. The sex-only group had fewer conduct problems, more prosocial attitudes and beliefs, and a lower expected risk of future delinquency than the sex-plus group. The study was based on retrospective data and did not provide a longitudinal view of their sexual offending. The nondevelopmental approach taken by Butler and Seto (2002) raises several questions about the unfolding of JSO behaviors over time. For example, is it possible that the sex-only JSO are late-onset nonsexual offenders. Similarly, is it possible that the sex-plus group includes individuals who committed nonsexual crimes that are sexually motivated (e.g., stealing underwear, break and entry). Also, it appears that the approach taken by Butler and Seto (2002) might be too restrictive to account for the various antisocial patterns among JSO (Lussier, van den Berg, Bijleveld, & Hendriks, 2012; McCuish et al., 2014).

The specialization/versatility issue also refers to whether persistent sexual offenders tend to repeat the same offense across sexual crimes (specialization) or whether their sexual offending is versatile and includes a wide array of sexual offenses. This issue is important given that it may help to delineate the nature of treatment and intervention offered. If sexual offending is repetitive and specific in nature, it may require a more circumscribed and specialized intervention aligned to the factors associated with the type of offense committed. Measuring versatility in sexual offending is challenging because sexual crimes are multidimensional and implies that sexual offenses can vary along multiple dimensions, such as victim’s age and gender, victim–offender relationship, the type and level of coercion used, and the type and level of sexual intrusiveness (Lussier, Leclerc, Healey, & Proulx, 2008). Studies conducted with JSO rarely include more than one or two dimensions of sexual offenses therefore limiting the examination of versatility in sexual offending. The general trend is that juveniles tend to repeat the same type of sexual offense in terms of the offender–victim age difference and the victim’s gender (Awad & Saunders, 1991; Awad et al., 1984; Fehrenbach et al., 1986). Limited research also suggest that those who initiate their sexually abusive behaviors in childhood and persist into adolescence are more likely to commit a different sexual offense than those who initiate their abusive behaviors in adolescence. For example, Vizard, Hickey, French and McCrory (2007) reported that 55% of childhood-onset persisters and 33% of adolescent-onset offenders had committed sexually abusive behaviors against both male and female victims. Furthermore, 31% of childhood-onset persisters had sexually abused both a child and an adult compared with 19% of the adolescent-onset group. Both differences were statistically significant, which suggest that versatility in sexual offending follows a diversification process.

Sexual Offending Trajectories

The preceding sections aimed to describe the current state of knowledge regarding specific developmental parameters of juvenile sexual offending. It is hypothesized that these parameters can be organized into theoretically informative sexual offending trajectories. The dual taxonomy of juvenile offenders proposed by Moffitt (1993) is a prominent example of theoretical model describing and explaining distinct general juvenile offending patterns. The dual taxonomy has been the subject of numerous empirical examinations generally supportive of the model (e.g., Piquero & Moffitt, 2005) but has also been subject to reformulation (e.g., Moffitt, Caspi, Harrington, & Milne, 2002), commentaries, and critiques (e.g., Sampson & Laub, 2005; Skardhamar, 2009). Moffitt’s (1993) taxonomy distinguished a group of life course persisters (LCP) and a group of adolescent-limited (AL) offenders. The taxonomy includes not only several predictions about developmental antecedents, individual differences, and familial environment, but also the course of delinquency and crime. More specifically, Moffitt (1993) argued that LCP represents about 5% to 10% of a male birth cohort. On one hand, they show a childhood onset of conduct disorder, an earlier onset of delinquency, a more pronounced pattern of offending versatility and aggravation, a tendency to specialize in violent offenses (e.g., assault, intimate partner abuse), and a much greater likelihood of persisting into adulthood, than the AL group. Members of the AL group, on the other hand, are described by an adolescence-onset of offending with a tendency to specialize in property and statutory offenses. These individuals typically desist from offending before reaching adulthood. Moffitt (1993) argued that during adolescence, the LCP and AL groups might be virtually indistinguishable in terms of their offending, thus creating a challenge for clinicians and practitioners involved in the case management and treatment of juveniles. It is likely that a similar situation may characterize JSO, which has contributed to the proliferation of risk assessment tools to guide clinicians distinguishing the sexual recidivists from the nonrecidivists.

Longitudinal studies have shown that the dual taxonomy might be too restrictive and should be expanded to account for other offending patterns, such as common offending and the low chronic offending (LeBlanc & Loeber, 1998; Piquero, 2008). Of importance, Moffitt’s LCP group was hypothesized to be most likely to escalate to rape and sexual assault as such behaviors would reflect a complex process of heterotypic continuity of antisocial behavior (Lussier, Leclerc, Cale, & Proulx, 2007). The LCP group is reminiscent of the antisocial JSO group identified by Becker (1998) as well as Butler and Seto (2002). The LCP syndrome, however, is unlikely to characterize all JSO as research suggests that no more than 40% of JSO meet the criteria of a conduct disorder (e.g., France & Hudson, 1993; Langström & Grann, 2000; Seto & Lalumière, 2010). Furthermore, the proportion of child-onset and adolescent-onset conduct disordered JSO remains unknown. The AL group was initially described as adolescent-onset offenders who were involved in crimes that matched their desire for an adult status (e.g., driving without a license, property theft), rather than violent offenses. This prediction is therefore counterintuitive to the idea of AL offenders committing sexual crimes. Some researchers have raised some objections (Lalumière, Harris, Quinsey, & Rice, 2005; Seto & Barbaree, 1997), arguing that members of the AL group may escalate their antisocial behaviors to sexual coercion by, among other things, mimicking the behaviors of LCP offenders who may be involved in sexual assaults. Based on this hypothesis, it would be expected that a relatively high proportion of AL offenders would be initially involved in gang rape or pressured into sexual offenses by their peers

Moffitt’s (1993) taxonomy of antisocial behaviors represents a starting point for the elaboration of a developmental classification model of juvenile sexual offending. This taxonomy is useful given that (a) sexual offenses committed by juveniles are first and foremost crimes and not necessarily manifestations of sexual deviance (Smallbone, 2006); (b) there is a significant presence of behavioral problems and conduct disorder, antisocial attitudes and cognitions, and general delinquency among adolescents involved in sexual crimes (e.g., Butler & Seto, 2002; France & Hudson, 1993; Seto & Lalumière, 2010; van Wijk et al., 2006); (c) JSO who recidivate are more likely to be rearrested for a nonsexual crime (Caldwell, 2002; Hendriks & Bijleveld, 2008; McCann & Lussier, 2008); and (d) the criminal behavior preceding and following a sexual crime is mainly nonsexual in nature (van Wijk, Mali, Bullens, & Vermeiren, 2007). Accordingly, sexual crimes committed by JSO are more often than not committed in isolation as part of a more general nonsexual offending pattern. The extent to which the taxonomy can account for both nonsexual and sexual offending patterns remained unclear until recently. In that regard, Hendriks, van den Berg, and Bijleveld (2014); Lussier et al. (2012); and more recently McCuish, Lussier, and Corrado (2015) established that the dual taxonomy does not account for general offending trajectories of JSO. Furthermore, a recent study (Lussier et al., 2012) reported modest synchrony between sexual offending and nonsexual offending trajectories with respect to Moffitt’s (1993) dual taxonomy of antisocial behaviors (i.e., AL and LCP). In other words, AL sexual offending is not necessarily associated with AL nonsexual offending, and a pattern of high-rate sexual offending is not necessarily associated with a pattern of life course persistence of nonsexual offending. This finding is crucial in that it shows that Moffitt’s taxonomy does not account for patterns of juvenile sexual offending and it emphasizes the need for a specific developmental model of juvenile sexual offending. Put differently, given the current state of knowledge, a specific developmental classification model of juvenile sex offending is more advisable and complementary to a developmental classification model of general delinquency.

The scientific literature highlights the presence of much discontinuity, albeit some continuity, in sexual offending over time. Becker (1998) suggested that JSO may include an abstainer group (i.e., nonrecidivist), an antisocial group (i.e., sexual offense as part of a general tendency to engage in delinquency), and a sexual group (i.e., at risk of persistence in sexual offense). The model is clinically intuitive but has not been empirically tested. It does suggest, however, that the sexual group will never desist from sexual offending, which is not supported by the empirical literature. It also suggests that the antisocial group is not at risk of sexually reoffending, which is counterintuitive with the fact that adolescents whose sexual offending persists in adulthood are more involved in nonsexual offending than those who do not persist (Lussier & Blokland, 2014; Zimring et al., 2007). Furthermore, Becker’s (1998) focus on recidivism informs neither about other developmental aspects of offending nor about offending trajectories. Empirical research conducted with adult sexual offenders has shown that general offending trajectories and sexual trajectories are different in number, shape, and rate (Francis, Harris, Wallace, Knight, & Soothill, 2014; Lussier & Davies, 2011; Lussier, Tzoumakis, Cale, & Amirault, 2010). Emerging findings highlight similar trends with JSO (Lussier et al., 2012), thus reinforcing the importance of distinguishing sexual and nonsexual offending. The proposed developmental classification system focuses on sexual offending.

The current state of theoretical and empirical knowledge suggests the presence of two distinct trajectories of juvenile sexual offending: (a) the AL sexual offenders and (b) the high-rate/slow-desisters (Table 2). It is hypothesized that these two trajectories can be distinguished on a series of developmental indicators. The AL JSO are hypothesized to represent the vast majority of juveniles who are involved in sexual offenses. Their prevalence, however, might not be as important in clinical settings or in the juvenile justice system. This group may share some similarities with the young male syndrome described by Lalumière et al. (2005). This group is unlikely to show sexual behavior problems during childhood and is hypothesized to be characterized by a relatively normal sexual development up to puberty. Their offending tends to start between the period of early to midadolescence. It is also argued that the growth of their sexual offending is very limited given that these young offenders may offend only once although some of them may repeat their behavior. Persistence, therefore, is possible if the associated risk factors are present and the protective factors are limited. In the context where there is persistence of sexual offending over time, it is argued that the sexual offending behavior will tend to be of the same nature.

OPEN IN VIEWER

Table 2.

A Developmental Taxonomy of Juvenile Sexual Offenders.

Developmental features of sexual offending	Developmental taxonomy
	Adolescence-limited	High-rate/slow-desisters
Official prevalence	About 90%	About 10%
Onset	Adolescence	Childhood
Frequency	Limited	Repetitive
Persistence	Limited, if any, to nonsexual offending	Yes, and involves continuity in adulthood
Recidivism	May be present, but limited to adolescence	Present in adolescence, probabilities declining in adulthood
Versatility	Limited	Present
Specialization	Limited	Limited in adolescence, some evidence in adulthood
Desistance	Rapid, in adolescence	Slow and gradual, in adulthood

It is believed that, for this group, the risk factors for sexual offending are transitory and more specific to the period of adolescence (e.g., puberty, peer influence, binge drinking, delinquency involvement, sexual arousal, opportunity). It is also hypothesized that sexual offending may take various shapes (e.g., child abuse, peer abuse) because situational, contextual, and social factors will be pivotal in creating opportunities for illegal sexual behaviors. It is therefore argued that their offending will neither be reflective of overwhelming deviant sexual thoughts, fantasies, or urges, nor of a deviant sexual preference in the making. However, these individuals may show a pattern of nonsexual juvenile delinquency. It is hypothesized that desistance from sexual offending will be rapid, if not immediate, for most of them and will occur in either late adolescence or in emerging adulthood. If there is persistence of offending beyond that period, it is expected that offending will be nonsexual in nature. This group is most likely to be found in community-based samples and therefore reflects trends and observations found in community-based studies. Currently, given the absence of a developmental model to guide clinical assessment and the similarities in terms of offending during adolescence, these adolescents may be misclassified as high-rate/slow-desisters.

The second group is referred to as the high-rate/slow-desisters and they represent a small subgroup of adolescents having committed a sexual offense. The high-rate/slow-desisters are most likely to be found in clinical samples and therefore reflect trends and observations found in clinical studies. This group is unlikely to be found in self-reported, population-based community samples given their overall low prevalence. This pattern of juvenile sexual offending, however, is more prevalent in criminal justice settings, especially those handling more serious cases and juvenile sexual recidivists (i.e., detention, inpatient treatment programs). It is hypothesized that their onset of sexual offending occurs in childhood. This group is likely to show atypical sexual behaviors during childhood, which may precede or co-occur with their sexual offending. This is not suggesting that all children showing atypical sexual behaviors go on to become JSO, but rather that the atypical sexual behaviors of this group in particular persist beyond childhood. In other words, this group is at risk of sexual offending during adolescence, especially if the exposure to risk factors of sexual offending persists and continues to overcome the protective factors.

The growth of their sexual offending will be gradual and constant without any intervention. This group is more likely to persist in their sexual offending beyond the period of adolescence. It is argued that these juveniles will eventually desist from sexual offending, but the desistance process is significantly longer due to the long-lasting effect of the multiple risk factors to which they have been exposed to early. They are more likely to be characterized with developmental risk factors related to sexual offending (e.g., childhood sexual victimization, exposure to sexually deviant models). The persistence of their sexual offending is reflective of the presence and the role of more stable risk factors and individual differences conducive to the commission of sexual offenses. These juveniles are at risk of manifesting one or multiple paraphilia and deviant sexual preferences later in adulthood. The high-rate/slow-desister group is also the one most likely to show evidence of diversification of sexual offending, which is most likely to occur during adolescence and young adulthood, and progressive evidence of increasing specialization in sexual offenses over time until termination of sexual offending.

The dual taxonomy of JSO has received preliminary empirical support from the field of sexual violence and abuse. Using both retrospective and prospective longitudinal data of a group of close to 500 adjudicated JSO, Lussier et al. (2012) used semiparametric group-based modeling to examine their sexual offending and nonsexual offending trajectories from age 12 to about age 32. Those authors identified that a two sexual offending trajectory model best fitted the longitudinal data, and the identified trajectories were in line with the dual taxonomy of JSO, that is, a group that mirrors the AL pattern (89.6%) and a group that resembles the high-rate/slow-desister pattern (10.4%). Key observations of the study were made about the two sexual offending trajectories. First, both groups had a similar proportion of sexual recidivists during adolescence but, albeit with a few exceptions, only the high-rate group included adult sexual recidivists. Second, the types of offender (i.e., child abusers, peer abusers, group offenders) were proportionally distributed across sexual offending trajectories, suggesting the dual taxonomy applied to different sexual offender types. Clearly, the Lussier et al. (2012) study requires replication, but it does point out that, during adolescence, both groups may look similar in terms of their sexual offending. Without additional research helping to identify developmental factors discriminating the two groups, classification errors are likely to occur.

Conclusion

The DLC approach presented in this article is one possible avenue that can guide research and facilitate this paradigm shift and to inform policy makers about the origins and development of sexual offending. The current state of knowledge on the origins and developmental course of sexual offending is limited to crude developmental parameters based on a limited number of studies conducted mainly with small samples of adjudicated juveniles that are not representative of the general population involved in sexual offenses. In spite of these limitations, the current state of knowledge does highlight the presence of much heterogeneity in terms of onset, frequency, persistence, continuity, and specialization in sexual offenses. Based on available evidence, it is argued that JSO are best represented by the presence of at least two distinct juvenile sexual offending trajectories: the AL and the high-rate/slow-desisters. These patterns are not considered by current policies, treatment models, or clinical assessment protocols, which may lead to the portrayal of all JSO as potential adult sexual offenders. Although the taxonomy presented in this article may be clinically intuitive and practically appealing, it should be seen as tentative for the time being, given the scarcity of DLC research on sexual offending. In fact, based on the available evidence, it is neither possible to prospectively identify juveniles at risk of following a pattern of high-rate slow desistance, nor to identify JSO presenting this offending pattern. The article also highlights several conceptual and methodological issues that limit the possibility of drawing firm conclusions about the origins and developmental course of juvenile sexual offending. To date, research conducted with JSO has been focused on between-individual differences either through trait-based or comparative studies to determine what is unique about this group. Future research now needs to be more focused on within-individual stability and changes or how one person evolves across developmental stages and life transitions and how the stability and changes affect sexual offending.

To address these issues, there is an urgent need for prospective longitudinal cohort studies examining the onset and the developmental course of normative and nonnormative sexual behaviors in representative samples drawn from the general population. Such longitudinal studies should start prior adolescence and the onset of juvenile sexual offending to examine prospectively the onset and course of juvenile sexual offending over time and associated risk and protective factors (e.g., Lussier et al., 2014). Given the low prevalence of official juvenile sex offending in general population studies, prospective longitudinal research should also be conducted with at-risk, vulnerable population presenting developmental risk factors of juvenile sex offending. Such populations may include, for example, samples of children presenting nonnormative sexual behaviors and those presenting early behavioral problems (e.g., Carpentier et al., 2006). These studies would inform about not only the continuity and the discontinuity of maladaptive behaviors over time but also the context in which continuity and discontinuity is most likely to occur. Finally, while the field of research has been heavily focused on juvenile sexual offending by males, prospective longitudinal studies should include the study of females’ normative and nonnormative sexual development. A pivotal aspect of DLC research will be to determine the extent and nature of gender differences to be taken into account for the development of explanatory and prevention models of juvenile sex offending (e.g., Lanctôt & Le Blanc, 2002). Using longitudinal cohort data, repeated measurements of normative and nonnormative sexual behaviors across developmental stages are needed to better understand the dynamic development of sexual behavior over time. Research should assist and guide practitioners as to what sexual behaviors, qualitatively (e.g., shape, context, etc.) and quantitatively (e.g., frequency), are developmental stage–normative and what behaviors are not. A challenge for researchers will be to develop a valid and reliable instrument that can measure sexual development (normative and nonnormative) from the earliest developmental stages and across these stages (e.g., Friedrich, Lysne, Sim, & Shamos, 2004). Given the complex association between general delinquency and sexual offending, the cooccurrence of sexual and nonsexual offending, it is recommended that the longitudinal study of normative and nonnormative sexual behaviors include longitudinal measures of antisocial behaviors. To do this, developmentalists have proposed various ways of looking simultaneously at between-individual and within-individual differences over time to inspect for the presence of offending trajectories and associated risk and protective factors. Examples of methods used to analyze longitudinal and repetitive measurements include latent-growth curve modeling (Duncan & Duncan, 2004), semiparametric group-based modeling (Nagin, 2005), as well as latent-transition class (Collins & Lanza, 2010).

Research in the area of general delinquency has shown that serious, chronic, and violent delinquency is a multistage phenomenon that develops over time with different and cumulating risk factors across various developmental stages (e.g., Thornberry, 2005). Research has also shown that serious, chronic, and violent delinquency can be, to some degree, prospectively screened when simultaneously taking into account risk and protective factors across multiple domains (e.g., biological, criminological, psychological, psychiatric, social) and multiple levels (e.g., individual, family, peer, school, neighborhood; for example, Loeber, Stouthamer-Loeber, Farrington, & White, 2008). Emerging longitudinal research suggest that sexual offending trajectories may follow relatively distinct courses than nonsexual offending trajectories (Lussier et al., 2012), but the similarity of some risk and protective factors may not preclude the intersection of these trajectories at some time point during childhood, adolescence, and/or adulthood. The implementation of prospective longitudinal cohort studies with repeated measurements of normative and nonnormative sexual behaviors across developmental stages could not only provide key policy information on between- and within-individual changes in sexual behaviors over time but also inform about pathways leading to sexually violent and abusive behaviors. In this article, it is hypothesized that at least two main developmental patterns best represent the sexual offending patterns among juveniles. This information is pivotal for the elaboration and implementation of multistage prevention and intervention programs. Currently, the scarcity of research on the origins and developmental course of juvenile sexual offending precludes the development of screening instruments that may support and guide the clinical assessment and treatment of children who may or may not be at risk of committing sexual offenses as well as those at risk of a long-term sexual offending trajectory.