Victim Age-Based Subtypes of Juveniles Adjudicated for Sexual Offenses

Subtypes of Adult Offenders Based on Victim Age

One frequently studied method of subtyping adult sex offenders uses the age of their victim(s) to establish subgroups. Several models of the etiology of sexual offending against children highlight the role of sexual victimization (e.g., Hall & Hirschman, 1992; Seto, 2008; Ward & Siegert, 2002), with such victimization being hypothesized to lead to offending against children through the adoption of deviant sexual scripts (Ward & Siegert, 2002), modeling, conditioning, or other disruptions to normal sexual development (Seto, 2008). Other commonly identified etiological factors related to offending against children include social skills deficits (e.g., Hall & Hirschman, 1992; Marshall & Barbaree, 1990; Ward & Siegert, 2002), cognitive distortions (Hall & Hirschman, 1992; Ward & Siegert, 2002), and problems in self-regulation (Hall & Hirschman, 1992; Ward & Siegert, 2002). Some of the same factors have been hypothesized to be related to the development of all sex offending regardless of victim age (e.g., sexual victimization, Marshall & Marshall, 2000; self-regulation problems, Stinson, Sales, & Becker, 2008). Others suggest that adults who commit sexual offenses against postpubescent victims have negative views of women, substance abuse problems, condone violence, are hypermasculine (Robertiello & Terry, 2007) and are similar to violent nonsexual offenders (Harris, Mazerolle, & Knight, 2009).

Research on adult sex offenders has supported the existence of some expected differences between groups, as adults who select child victims are more likely to have been sexually abused (e.g., Jespersen, Lalumière, & Seto, 2009), have greater perceived deficits in social skills, and greater social anxiety (e.g., Segal & Marshall, 1985), whereas adults who offend against postpubescent victims are more generally antisocial (e.g., Olver & Wong, 2006; Porter, Fairweather, Drugge, Hervé, Birt, & Boer, 2000; Quinsey, Rice, & Harris, 1995) and have higher violent (nonsexual) recidivism rates (e.g., Prentky, Lee, Knight, & Cerce, 1997; Quinsey et al., 1995). Comparisons between subtypes of adult sex offenders based on victim age have thus consistently demonstrated meaningful differences between groups that are expected based on existing theory. These differences suggest potentially unique etiologies as well as treatment approaches for the subgroups. The success of this method of subtyping adult offenders has led to research examining the validity of this distinction among juveniles adjudicated for sexual offenses.

Subtypes of Juveniles Adjudicated for Sexual Offenses Based on Victim Age

Although this method of subtyping has intuitive appeal, the meaning of the selection of a child victim when the perpetrator is himself (or herself) still a minor is not entirely clear. It appears that theories regarding these subtypes of adult offenders have simply been extended downward to juveniles. Robertiello and Terry (2007) described the typically expected differences between these groups:

Those who abuse children . . . rely on opportunity, trickery, bribes, and threats, and often experience deficits in self-esteem and social competence. They lack social skills and show signs of depression. Juveniles who victimize peers and adults often commit sexual offenses in conjunction with other criminal behavior and exhibit a more generalized type of delinquency. These offenders are more likely to target strangers, use weapons, and cause injuries to their victims. (p. 515)

In addition, Seto and Lalumière (2010) hypothesized that “sexual victim age might . . . be an important moderator for some special explanations of adolescent sexual offending” (p. 532); therefore, theories regarding the etiology of sexually abusive behavior in juveniles might be specific to the etiology of offenses against children. Seto and Lalumière (2010) identified factors commonly theorized to contribute to the development of sexual offending in adolescents, which therefore may be associated specifically with the development of abuse of children: sexual victimization, poor childhood attachment, social incompetence, disruptions in sexual development, atypical sexual interests, psychopathology, and cognitive limitations. Support for these theories would thus be provided by demonstrating greater problems in these areas among juveniles who have victimized children.

Research regarding these expected differences has been largely inconsistent, however, in both methods and results. Studies differ in their operational definition of child victims (e.g., below age 12 and at least 3 years younger than offender: Carpenter, Peed, & Eastman, 1995; 4 or more years younger than offender: Richardson, Kelly, Bhate, & Graham, 1997); however, these definitional differences neither significantly alter the composition of subgroups nor the differences found between groups (Kemper & Kistner, 2010). In addition, existing studies differ in the assessment measures used (e.g., different measures of mental health functioning in Carpenter et al., 1995; Hunter, Figueredo, Malamuth, & Becker, 2003; Ronis & Borduin, 2007) and whether structured assessments are used at all (e.g., ‘t Hart-Kerkhoffs, Doreleijers, Jansen, van Wijk, & Bullens, 2009). Perhaps such methodological inconsistencies between studies have contributed to the inconsistent results produced to date.

Brief descriptive information and major findings of studies on victim-age-based subgroups are presented in Tables 1 and 2. As noted above, subtyping schemes can be validated by demonstrating theoretically expected differences between groups. The variables used to examine theoretically expected differences may be considered “psychologically meaningful” (Mann, Hanson, & Thornton, 2010, p. 194). The victim-age-based method of subtyping has not yet been accepted as a valid typology for JSOs because the few consistent differences identified to date are not related to etiology, treatment outcome, or recidivism. The review below highlights results related to theoretically expected differences (i.e., psychologically meaningful variables) which have produced largely inconsistent findings to date, results that have been consistent across studies, and results regarding risk and recidivism. The potential role of disruptions in sexual development, atypical sexual interests, and cognitive limitations will not be included in the review below as they could not be addressed in the current study.

OPEN IN VIEWER

Table 1.

Previous Findings Regarding Differences Between Subgroups in Individual and Family Characteristics

Study	N	Sample characteristics	Individual characteristics	Family characteristics
Carpenter, Peed, and Eastman (1995)	36	17- to 19-year-olds ordered to receive sex offender treatment in a residential youth correctional facility. No MV group.	No differences in age, race/ethnicity, prior sex offenses, prior nonsex offenses, prior enrollment in treatment, or MCMI Histrionic, Narcissistic, or Antisocial scales. CV scored higher on the MCMI Schizoid, Avoidant, and Dependent scales.	NA
Ford and Linney (1995)	35	9- to 18-year-olds ordered into state-operated facilities. No MV group.	No differences in age, educational status, self-concept, and social skills. CV less likely to have prior offenses and reported greater exposure to pornographic magazines at young ages.	No differences in family structure, family size, or family criminal history. CV reported more intrafamily violence. A greater proportion of CV reported being sexually victimized.
Gunby and Woodhams (2010)	43	10- to 17-year-olds referred to East Midlands Youth Offending Teams. No MV group.	No differences in mental health problems. CV were more frequently the victims of bullying, had fewer age-appropriate friends, and had more negative self-esteem. PV more frequently came from deprived homes.	No differences in familial substance use. PV experienced more inconsistent supervision, had greater familial criminal involvement, and were more likely to have witnessed family violence. CV more frequently lived in “chaotic” households.
Hagan and Cho (1996)	100	12- to 19-year-olds who successfully completed mandatory treatment program in a correctional facility. No MV group.	NA	NA
Hagan, Gust-Brey, Cho, and Dow (2001)	100	Same sample as above	NA	NA
‘t Hart-Kerkhoffs, Doreleijers, Jansen, van Wijk, and Bullens (2009)	174	12- to 18-year-olds suspected of having committed a sex crime. No MV group.	No differences in externalizing problems, environmental conditions, substance abuse, school problems, or leisure-time functioning. CV demonstrated more problems in psychosocial development and internalizing problems.	CV more frequently reported problematic familial circumstances.
Hendriks and Bijleveld (2004)	116	11- to 18-year-olds prosecuted for a sexual offense for whom psychological testing had been requested. No MV group.	No differences in extraversion, sensation seeking, sexual abuse, or prior nonsex offenses. CV scored higher on neuroticism, were more likely to have psychopathology, were more often the victims of bullying, had lower self-image, had poorer peer relationships, were more likely to be of Dutch origin, were younger at the time of the offense, and were more likely to have a previous sex offense.	No differences in cruelty by parents, neglect by parents, or interparental violence.
Hummel, Thömke, Oldenbürger, and Specht (2000)	74	14- to 20-year-old defendants accused of sexual offenses. No MV group.	CV were more often victims of sexual abuse (significance testing not reported)	NA
Hunter, Hazelwood, and Slesinger (2000)	126	Records provided by police officers in training at FBI Academy. Age range not provided. No MV group.	No differences in age at time of offense, prior general arrests, prior sexual offense arrests. CV were predominantly White, PV were predominantly African American.	NA
Hunter, Figueredo, Malamuth, and Becker (2003)	182	12- to 18-year-olds treated in residential facilities (correctional and noncorrectional). No MV group.	No difference in prior arrests for sexual crimes. CV had greater deficits in psychosocial skills, were more likely to score in clinically significant range for depression and anxiety on the Youth Self-Report, and had a prior arrest for a nonsexual assault more frequently.	NA
Kaufman, Hilliker, et al. (1996) a	179	13- to 20-year-olds from two juvenile correctional facilities. Analyses used age of youngest victim.	A history of abuse was associated with selecting younger victims.	NA
Kemper and Kistner (2007)	296	12- to 19-year-olds in a juvenile residential training school (correctional facility).	No differences in FSIQ, age at admission, age at first sex offense, number of commitments to DJJ facilities, or age at first arrest. CV were more likely to be White, those who offended against peers were more likely to be African American. PV had more nonsexual charges.	NA
Kempton and Forehand (1992)	15	11- to 18-year-olds in a correctional facility. No MV group.	No differences between groups on the Anxiety, Inattentive, Aggressive, or Social/Withdrawal subscales of the Teacher Report Form.	NA
Långström and Grann (2000)	46	15- to 20-year-olds referred for an evaluation of insanity. No MV group.	NA	NA
Parks and Bard (2006)	156	12- to 17-year-olds in a juvenile correctional facility.	No differences between groups on age at index offense, age at admission or discharge from the institution, FSIQ, VIQ, or PIQ. PV were more often African American, with more Whites in the CV and MV groups.	NA
Rasmussen (1999) b	170	7- to 18-year-olds adjudicated for a sexual offense.	NA	NA
Richardson, Kelly, Bhate, and Graham (1997)	100	11- to 18-year-olds referred for psychological and/or psychiatric assessment. Half of MV group victimized a younger sibling and an unrelated child.	No differences in diagnosis of psychosis, depression, or conduct disorder. MV had earlier age of onset than CV and PV. A greater proportion of MV had been sexually abused than CV and PV. PV were more likely to have engaged in theft and assault and to have antisocial peers than other groups. MV and PV were more likely than CV to have school behavior problems.	No differences between groups in rates of experiencing or witnessing family violence. MV and CV were more often considered at risk of neglect or abuse by social services and were more often categorized as a “problem family” by social services.
Ronis and Borduin (2007)	46	10- to 17-year-olds referred to the Missouri Delinquency Project. No MV group.	No differences in age, number of arrests, SES, race, sexual or physical abuse, scores on the Revised Behavior Problem Checklist (anxiety, attention, conduct disorder, and aggression), scores on the Global Severity Index of the Brief Symptom Inventory, scores on the Missouri Peer Relations Inventory (emotional bonding and aggression) or academic performance.	No differences in proportion living with two parents, scores on the Family Adaptability and Cohesion Evaluation Scales–II, or ratings of negative affect, or facilitative exchange in an observational task.
Smith and Monastersky (1986)	112	10- to 16-year-olds referred to a community evaluation and treatment program. No MV group.	NA	NA
van Wijk et al. (2007)	578	12- to 20-year-olds who were evaluated by the forensic mental health service in the Netherlands. No MV group.	No differences in age. Developmental disorder and cluster C personality disorders were more common among CV. PV were similar to violent nonsex offenders.	NA
van Wijk, van Horn, et al. (2005)	112	Same as above.	No differences in age, self-esteem, or being a victim of bullying. CV were more often White, whereas PV were more often ethnic minority.	NA
			Post hoc tests not reported but differences reported between groups on neuroticism (CV higher), IQ (CV higher), school-dropout rate (CV lower), and bad contact with peers (PV higher).
Worling (1995)	90	12- to 19-year-olds in outpatient treatment. No MV group.	No differences in age or SES. No differences between juveniles with female child and female peer/adult victims in sexual abuse rates, which were significantly lower than the rates of sexual abuse among juveniles with any male child victim.	NA

Note: CV = Juveniles with child victims; PV = Juveniles with peer/adult victims; MV = Juveniles with mixed victim types. Some samples included nonsexual offenders; in these instances, only the number of sexual offenders are included in the N. A meta-analysis by Graves, Openshaw, Ascione, and Ericksen (1996) is excluded because the “sexual assault offender” group is described as including victims of varying age, including victims younger than the offender.

a.

(K. L. Kaufman, et al., 1996) did not create two categories (child vs. peer/adult). Victims divided into the following age ranges: 0-3, 4-6, 7-9, 10-12, 13-17.

b.

Rasmussen (1999) did not compare based on victim age categories but used age of victim as a predictor of recidivism.

OPEN IN VIEWER

Table 2.

Previous Findings Regarding Differences Between Subgroups in Offense Characteristics, Victim Characteristics, and Risk to Reoffend

Study	Offense characteristics	Victim characteristics	Risk to reoffend and recidivism rates
Gunby and Woodhams (2010)	PV were more likely to use violence to control victims.	CV more often selected victims known to the offender.	NA
Hagan and Cho (1996)	NA	NA	No differences in sexual or general recidivism rates over 2-year follow-up. No differences in the severity of subsequent offenses.
Hagan, Gust-Brey, Cho, and Dow (2001)	NA	NA	No differences in sexual recidivism over 8-year follow-up.
‘t Hart-Kerkhoffs, Doreleijers, Jansen, van Wijk, and Bullens (2009)	CV less likely to use excessive violence. A greater proportion of CV reported feeling sexual excitement during offense.	CV more likely to offend against a male and less likely to offend against a stranger than PV.	No differences in sexual recidivism. PV who committed offense in a group had higher general and violent recidivism rate than PV who committed offense alone and CV.
Hendriks and Bijleveld (2004)	CV more likely to use verbal violence or threats and less likely to use physical violence than PV.	CV less likely to offend against a stranger and more likely to have a male victim than PV.	NA
Hunter, Hazelwood, and Slesinger (2000)	No differences in rates of transporting the victim, murdering the victim, or use of a weapon. PV more likely to commit the offense with co-offenders, more likely to commit the offense in the context of another crime, and more likely to use force than CV.	No difference in the rates of having multiple victims. PV had more female victims and more stranger victims than CV. Child victims more likely to use no or passive resistance strategies, peer/adult victims more likely to use verbal or physical resistance strategies.	NA
Hunter, Figueredo, Malamuth, and Becker (2003)	No differences in rates of actual or attempted intercourse or presence of co-offenders. PV used greater force, used a weapon more often, and were under the influence of alcohol or drugs at the time of the offense more often. CV more likely to commit the offense at victim’s home.	No difference in likelihood of having multiple victims. CV more likely to have offended against a relative.	NA
K. L. Kaufman, et al. (1996) a	Juveniles who offended against younger victims were more likely to give gifts and/or use love and attention to gain victim’s trust, less likely to use alcohol or drugs, more likely to threaten the relationship and to use withdrawal of benefits.	NA	NA
Kemper and Kistner (2007)	MV had more adjudications and victims (even when compared with CV and PV with more than 1 victim). CV less likely to engage in penile penetration. CV and MV more likely to have engaged in oral sex. MV engaged in a greater variety of sexual acts than CV, who in turn engaged in a greater variety than PV.	CV and MV more likely than PV to have victims of both genders, less likely to have only female victims, and more likely to victimize only family members.	No differences in nonsexual or sexual recidivism rates over an average follow-up of 5.22 years. MV more likely than both other groups to be treatment noncompleters, although there were no differences in rates of removal from group treatment.
Långström and Grann (2000)	NA	NA	No differences in general or sexual recidivism over an average 5-year follow-up.
Parks and Bard (2006)	NA	NA	No differences on PCL:YV Interpersonal and Antisocial factors, sexual recidivism, nonsexual recidivism (maximum time at risk of 11 years).MV less likely to successfully complete treatment than CV or PV. MV scored higher than CV on PCL:YV Total score, Affective and Behavioral factors, as well as JSOAP-II Impulsive-Antisocial Behavior scale, Intervention scale, and Total score. On the J-SOAP-II Sexual Drive/Preoccupation Scale, MV scored higher than CV, who scored higher than PV.
Rasmussen (1999) b	NA	NA	Age of youngest victim was not associated with sexual recidivism over a 5-year follow-up.
Richardson, Kelly, Bhate, and Graham (1997)	No differences in types of sexually abusive behaviors, means of overcoming victim resistance. MV had longer duration of offending than CV	PV more likely to offend against strangers. MV and CV more likely to offend against acquaintances. CV more likely to have a male victim.	NA
	or PV. MV and CV engaged in a greater range of sexual acts than PV. PV more likely to commit offense in public.
Smith and Monastersky (1986)	NA	NA	No significant difference in general or sexual recidivism.

Note: CV = Juveniles with child victims; PV = Juveniles with peer/adult victims; MV = Juveniles with mixed victim types. The following studies were excluded from this table as they did not include information relevant to either offense characteristics, victim characteristics, or risk to reoffend/recidivism: Carpenter, Peed, and Eastman (1995), Ford and Linney (1995), Hummel, Thömke, Oldenbürger, and Specht (2000), Kempton and Forehand (1992), Ronis and Borduin (2007), van Wijk and colleagues (2005, 2007), and Worling (1995).

a.

Kaufman et al. (1996) did not create two categories (child vs. peer/adult). Victims divided into the following age ranges: 0-3, 4-6, 7-9, 10-12, 13-17.

b.

Rasmussen (1999) did not compare based on victim age categories but used age of victim as a predictor of recidivism.

Research Regarding Psychologically Meaningful Variables

Juveniles who offend against children are expected to have higher rates of sexual victimization than those who offend against peers or adults. A number of studies demonstrate such a finding, although they differ regarding whether the effect was noted for all juveniles with child victims (Ford & Linney, 1995; ‘t Hart-Kerkhoffs et al., 2009; Hummel, Thömke, Oldenbürger, & Specht, 2000), only those with male child victims (Worling, 1995), or those that have offended against both child and peer/adult victims (Richardson et al., 1997), and some studies did not find a significant difference (‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Ronis & Borduin, 2007). A recent meta-analysis did find that juveniles with child victims were significantly more likely to have a history of sexual (but not physical) abuse than juveniles with peer/adult victims (Seto & Lalumière, 2010). Further evidence of differences between victim-age based subgroups is needed, given the mixed results to date, to support theories that propose a major etiological role for child sexual victimization in the development of offending against children.

Juveniles who offend against children are expected to have higher rates of mental health problems as well. Currently, evidence regarding differences in problems with anxiety is mixed (Hunter et al., 2003; Kempton & Forehand, 1992; Ronis & Borduin, 2007), although there is some consistent evidence that those with child victims are more likely to have internalizing problems (‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; van Wijk, van Horn, Bullens, Bijleveld, & Doreleijers, 2005). Research to date has found not found significant differences in externalizing symptoms between groups (‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Kempton & Forehand, 1992; Richardson et al., 1997; Ronis & Borduin, 2007). Consistent findings that JSOs with child victims are more likely to have internalizing problems would demonstrate a potentially important difference in the etiology of offending against children that has implications for treatment planning.

Juveniles who offend against children are expected to have lower self-esteem and greater deficits in social skills than juveniles who offend against peers/adults. Although there is some support for greater problems in peer relationships among those who select child victims (Gunby & Woodhams, 2010; Hendriks & Bijleveld, 2004; Hunter et al., 2003; van Wijk, van Horn, et al., 2005), other research has not confirmed this hypothesis (Ford & Linney, 1995; Ronis & Borduin, 2007). The literature regarding differences in self-esteem is also inconsistent (Ford & Linney, 1995; Gunby & Woodhams, 2010). The inconsistent results cast doubt on the theory that poor social skills or peer relationships play a role in the etiology of offending against children, and further research is needed to clarify the mixed results to date.

Juveniles adjudicated for sexual crimes against peer/adult victims have been hypothesized to be more generally delinquent and antisocial than those who commit crimes against children. Some research demonstrates the expected difference in prior arrests (Ford & Linney, 1995; Kemper & Kistner, 2007; Richardson et al., 1997); however, other studies have not found a significant difference (Carpenter et al., 1995; Hendriks & Bijleveld, 2004; Hunter, Hazelwood, & Slesinger, 2000) or have found the reverse (Hunter et al., 2003). A recent meta-analysis found that those with only peer/adult victims were more likely to have indicators of delinquency risk or a history of arrests than those with any child victims (Seto & Lalumière, 2010). It is somewhat difficult to interpret these findings given the lack of differences identified when externalizing symptoms are studied; it is possible that a difference in arrest history or delinquency risk factors is more likely to be detected in correctional samples, although such differences were not found in all studies with such samples. Further research is needed to clarify whether JSOs with peer/adult victims are more generally antisocial.

If juveniles who offend against peers/adults are hypothesized to be similar to general delinquents, then the characteristics commonly seen in families of delinquents may be seen in this subgroup more frequently. Family risk factors for delinquency include poor parental monitoring, punitive or inconsistent discipline, emotionally distant parents, child physical abuse, low family income (Murray & Farrington, 2010), and witnessing family violence (Zinzow et al., 2009). There is some initial evidence that those who select peer/adult victims are more likely to have been exposed to inconsistent discipline and socioeconomic deprivation (Gunby & Woodhams, 2010). Findings regarding differences in exposure to family violence (Ford & Linney, 1995; Gunby & Woodhams, 2010; Richardson et al., 1997) and exposure to family criminality (Ford & Linney, 1995; Gunby & Woodhams, 2010) are inconsistent. Seto and Lalumière’s (2010) meta-analysis found no difference between these subtypes on family problems. If a consistent body of literature shows differences between subgroups in family characteristics, this would provide further evidence for the validity of the distinction between these groups. Additional research is needed given the inconsistent results to date.

Differences Consistently Found Between Victim-Age-Based Subgroups

Much of the research to date on victim-age-based subgroups focuses on subgroup differences that are not clearly tied to any theoretical expectations, yet these comparisons have produced the most consistent findings across studies. There is a variety of research suggesting that juveniles with peer/adult victims are more likely to use force or violence (Gunby & Woodhams, 2010; ‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Hunter et al., 2000, 2003; cf. Richardson et al., 1997). There is also substantial evidence that juveniles with peer/adult victims are more likely than those with child victims to offend against strangers (Gunby & Woodhams, 2010; ‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Hunter et al., 2000, 2003; Richardson et al., 1997). Juveniles who have any child victims (including mixed offenders) are more likely to offend against family members (Hunter et al., 2003; Kemper & Kistner, 2007) or acquaintances (Richardson et al., 1997). In addition, juveniles who have any child victims are consistently found to be more likely than those with only peer/adult victims to have offended against at least one male (‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Hunter et al., 2000; Kemper & Kistner, 2007; Richardson et al., 1997). Although the importance of such differences is currently unclear, demonstrating similar differences in new samples suggests some comparability to prior research.

Risk to Reoffend/Recidivism

Theories regarding victim-age-based subgroups could reasonably lead to a hypothesis that those who offend against peers/adults will have higher general recidivism rates due to higher overall antisociality; however, the research to date has not found any significant differences in general or nonsexual rearrest rates (Hagan & Cho, 1996; Kemper & Kistner, 2007; Långström & Grann, 2000; Parks & Bard, 2006; Smith & Monastersky, 1986). In addition, there is evidence that juveniles with child victims and those with mixed victims score higher than juveniles with peer/adult victims on a risk assessment instrument designed for JSOs (Parks & Bard, 2006), yet research has consistently failed to demonstrate significant differences in sexual recidivism (Hagan & Cho, 1996; Hagan, Gust-Brey, Cho, & Dow, 2001; Kemper & Kistner, 2007; Långström & Grann, 2000; Parks & Bard, 2006; Rasmussen, 1999; Smith & Monastersky, 1986). Selecting male child victims is believed to be a risk factor for sexual recidivism in JSOs (Prentky & Righthand, 2003), perhaps because research has demonstrated that sexual interest in children (Hanson & Morton-Bourgon, 2005) and selection of male child victims (Hanson & Bussière, 1998) are associated with higher recidivism rates. In addition, the selection of child victims may be perceived as indicative of increased risk for sexual recidivism because child victims are more easily intimidated or coerced and less capable of making effective reports. Therefore, a potential difference in recidivism between victim-age-based subgroups is an important area for continued study, especially given the heightened emphasis in recent years on identifying high-risk youth (e.g., Elkovitch, Viljoen, Scalora, & Ullman, 2008; Prentky, Li, Righthand, Schuler, Cavanaugh, & Lee, 2010).

Study Rationale and Aims

Although there are a number of theoretically expected differences between victim-age-based subgroups of JSOs, consistent findings (or consistent null results) have been elusive to date. The most consistent results are for differences that are not clearly connected to etiological theory, treatment outcome, or recidivism, and thus the validity of making distinctions between groups based on victim age is uncertain. Additional research is needed to provide more evidence, where results have been inconsistent to date, and to shift the focus to theoretically relevant variables.

In addition, further research is required to address the limitations of the existing research. Many studies have been limited to juveniles in residential or correctional settings (e.g., Ford & Linney, 1995; Hagan & Cho, 1996; Hagan et al., 2001; Hunter et al., 2003; K. L. Kaufman, et al., 1996; Kemper & Kistner, 2007; Parks & Bard, 2006). Other studies have included youth not yet adjudicated for sexual offenses (‘t Hart-Kerkhoffs et al., 2009; Långström & Grann, 2000). Concerns regarding the generalizability of these types of samples suggest that additional research regarding juveniles participating in outpatient treatment is needed. In addition, a number of studies do not explicitly articulate the method used to classify victims as children or peers (e.g., Graves et al., 1996; Hagan & Cho, 1996; Hagan et al., 2001; Kempton & Forehand, 1992; van Wijk, van Horn, et al., 2005). Furthermore, many studies have relied exclusively on file reviews rather than interviews or assessment measures completed by the child or parent (e.g., Gunby & Woodhams, 2010; Hunter et al., 2000; Richardson et al., 1997). The inclusion of parent-report and the use of standardized assessment measures has been a rarity to date, and most studies include relatively few of the variables described above (for an exception, see Ronis & Borduin, 2007). Finally, the majority of studies have excluded offenders with both child and peer/adult victims.

The current study seeks to build on this existing literature by comparing JSOs from an outpatient treatment sample who were classified into one of three groups based on the age(s) of their victim(s): child victims (at least 4 years younger than offender), peer/adult victims, or both types of victims (referred to as mixed). The study draws on a comprehensive database that includes a wide range of variables gathered using a variety of methods and informants (i.e., self-report, parent-report, semistructured clinical interview, and legal documentation). The study focuses first on theoretically relevant variables, as differences on these variables inform the validity of this method of subtyping juvenile offenders. Second, we will examine some of the victim characteristics commonly studied in prior research, to demonstrate the comparability of our sample with previous studies. Finally, we will examine risk and recidivism, given the importance of identifying high-risk youth.

As compared to those with peer/adult victims, juveniles with child victims were hypothesized to: 1) have higher rates of sexual victimization but not physical abuse; 2) be more likely to be diagnosed with anxiety and affective disorders, but not other mental health diagnoses; 3) have more severe symptoms of depression, anxiety, posttraumatic stress, and general internalizing problems but will not differ in symptoms of externalizing problems or anger; 4) have more peer problems and lower social competence.

Given the theory- and research-based expectations that juveniles who offend against peer/adult victims are more generally antisocial, we hypothesize that that this group will: 5) be more likely to have a history of prior arrests; 6) score higher on a measure of antisocial tendencies; 7) demonstrate more problematic scores on scales assessing parenting practices and activities; 8) have higher rates of witnessing domestic violence.

Based on prior consistent findings in the literature, we hypothesize that juveniles with child victims will: 9) be more likely to have male victims and to have offended against a biological relative and 10) score higher on the Juvenile Sexual Offender Assessment Protocol–II (J-SOAP-II; and 11) not differ in recidivism rates.

Few studies have included offenders who have both child and peer/adult victims and a theoretical basis for making predictions regarding this group is lacking. Therefore, the current study will examine differences between this group and the other two groups of offenders in an exploratory manner, comparing groups on only those domains that have been studied in prior research with mixed offenders: mental health functioning, antisocial tendencies, victim characteristics, and recidivism.

Method

Results

Juveniles With Child Victims Compared With Juveniles With Peer/Adult Victims

Abuse history

There were no significant differences between groups on self- or parent-reported physical, sexual, or emotional abuse based on the ACSBI abuse items or based on clinicians’ ratings of the presence of each type of maltreatment on the K-SADS-PL (see Table 3). There is a nonsignificant trend for the parents of juveniles with child victims to report that their child was a victim of sexual abuse on the relevant ACSBI item more often than the parents of juveniles with peer/adult victims.

OPEN IN VIEWER

Table 3.

Comparisons Between Juveniles With Child Victims and Those With Peer/Adult Victims

	Child victim group	Peer/adult victim group
	M (SD) or % (n)	M (SD) or % (n)	df	Statistic	p
Abuse history
Clinician’s indication that child maltreatment occurred (K-SADS)
Physical abuse	22.94% (25)	17.78% (8)	1	χ2 = 0.50	.478
Sexual abuse	17.59% (19)	8.89% (4)	1	χ2 = 1.88	.170
Child’s report on ACSBI Items 47, 48, 49
Physical abuse	21.36% (22)	15.91% (7)	1	χ2 = 0.58	.447
Sexual abuse	15.53% (16)	6.82% (3)	1	χ2 = 2.08	.149
Emotional abuse	33.33% (34)	20.45% (9)	1	χ2 = 2.45	.117
Parent’s report on ACSBI Items 47, 48, 49
Physical abuse	21.51% (20)	19.05% (8)	1	χ2 = 0.11	.744
Sexual abuse	29.03% (27)	9.76% (4)	1	χ2 = 5.95	.015
Emotional abuse	55.91% (52)	47.62% (20)	1	χ2 = 0.80	.371
Mental health
Diagnoses based on K-SADS-PL
Disruptive behavior disorder	38.53% (42)	44.68% (21)	1	χ2 = 0.52	.473
Anxiety disorder	10.09% (11)	0% (0)	1	χ2 = 5.10	.035°
Affective disorder	3.67% (4)	2.13% (1)	1	χ2 = 0.25	1.00°
Substance use disorder	0% (0)	6.38% (3)	1	χ2 = 7.09	.026°
Trauma Symptom Checklist for Children
Anxiety	47.06 (8.43)	46.45 (6.61)	143	t = 0.42	.673
Depression	48.06 (8.55)	45.16 (5.82)	143	t = 2.05	.042
Anger	45.19 (7.59)	44.98 (6.91)	143	t = 0.16	.875
Posttraumatic stress	48.60 (8.92)	46.45 (7.55)	143	t = 1.40	.165
Dissociation	47.50 (7.90)	45.39 (6.42)	143	t = 1.56	.121
Sex concerns a	48.79 (11.14)	46.68 (8.70)	103.60	t = 1.23	.222
Underreporting	53.86 (11.78)	55.80 (12.08)	143	t = −0.90	.368
Child Behavior Checklist/Strengths and Difficulties Questionnaire
Internalizing	0.05 (0.96)	−0.10 (1.09)	149	t = 0.83	.409
Externalizing	−0.09 (1.03)	0.19 (0.92)	149	t = −1.63	.106
Social skills
CBCL social competence b	40.13 (9.05)	42.37 (7.25)	89.63	t = −1.44	.154
CBCL/SDQ		−0.09 (1.04)	149	t = 0.80	.426
Social/peer problems	0.05 (0.94)
Criminal history/antisocial tendencies
Any prior nonsexual arrests	6.36% (7)	14.89% (7)	1	χ2 = 2.95	.123°
Antisocial Process Screening Device
Callous/unemotional	3.91 (2.20)	4.18 (2.24)	136	t = −0.67	.505
Narcissism	3.79 (3.13)	3.77 (2.43)	136	t = 0.04	.967
Impulsivity	4.36 (2.19)	4.95 (1.84)	136	t = −1.55	.123
Family functioning/parenting practices
Alabama Parenting Questionnaire
Involvement	37.88 (5.30)	36.74 (5.78)	116	t = 1.04	.302
Positive parenting	24.71 (3.61)	24.33 (3.27)	116	t = 0.536	.593
Poor monitoring	16.74 (5.14)	20.77 (5.02)	115	t = −3.88	<.001
Inconsistent discipline	13.95 (3.98)	14.09 (3.95)	116	t = −0.17	.867
Corporal punishment	4.42 (2.15)	4.76 (2.22)	114	t = −0.78	.438
Clinician’s indication that child maltreatment occurred (K-SADS)
Witnessed domestic violence	46.73% (50)	26.67% (12)	1	χ2 = 5.28	.022
Victim characteristics
Any male victim	35.09% (40)	8.00% (4)	1	χ2 = 12.99	<.001
Victim was biological relative	37.50% (42)	6.00% (3)	2	χ2 = 17.10	<.001
Risk to reofffend/recidivism
J-SOAP-II Total score	21.50 (6.90)	18.22 (6.80)	28	t = 1.20	.240
Any rearrest	7.94% (5)	30% (9)	1	χ2 = 7.74	.011°
Sexual rearrest	0% (0)	3.33% (1)	1	χ2 = 2.12	.323°

Note: K-SADS-PL = Schedule for Affective Disorders and Schizophrenia for School Aged Children–Present and Lifetime Version, ACSBI = Adolescent Clinical Sexual Behavior Inventory, APSD = Antisocial Process Screening Device, TSCC = Trauma Symptom Checklist for Children, CBCL = Child Behavior Checklist, SDQ = Strengths and Difficulties Questionnaire, APQ = Alabama Parenting Questionnaire, J-SOAP-II = Juvenile Sex Offender Assessment Protocol–II. °Fisher’s Exact Test used because at least one cell has an expected count less than 5. Maximum scores on the APSD scales are 14 for Narcissism, 10 for Impulsivity, and 12 for Callous-Unemotional. TSCC scores are T-scores (M = 50). CBCL/SDQ Internalizing, Externalizing, and Social/Peer Problems are z scores. CBCL Social Competence score is a T-score (M = 50). Maximum scores on the APQ scales are 50 for Involvement, 30 for Positive Parenting, 50 for Poor Monitoring, 30 for Inconsistent Discipline, and 15 for Corporal Punishment.

a.

Levene’s test F = 4.15, p = .044.

b.

Levene’s test F = 4.16, p = .044.

Mental health

There were no significant differences between juveniles with child victims and juveniles with peer/adult victims on any mental health variables (see Table 3). Juveniles with child victims were marginally more likely to have an anxiety disorder diagnosis, and juveniles with peer/adult victims were marginally more likely to have a substance use disorder. In addition, juveniles with child victims scored marginally higher on the Depression scale of the TSCC.

Social skills

There were no significant differences between groups on the Social Competence scale of the CBCL or on the combined CBCL/SDQ variable indicating social/peer problems (see Table 3).

Criminal history/antisocial tendencies

There were no significant differences between groups on any indicators of criminal history or antisocial tendencies (see Table 3).

Family functioning/parenting practices

There was a significant difference between groups on poor monitoring on the APQ. The parents of juveniles with peer/adult victims reported greater problems in monitoring than the parents of juveniles with child victims. There were no other significant differences between groups on family functioning or parenting practices. Clinicians indicated that juveniles with child victims had witnessed domestic violence at a marginally higher rate than juveniles with peer/adult victims.

Victim characteristics

There were significant differences between groups on both victim characteristics studied: any male victims and any biological relative victims (see Table 3). Juveniles with child victims were more likely to have a male victim and more likely to have a biologically related victim.

Risk to reoffend/recidivism

There was not a significant difference between groups on J-SOAP-II total score (see Table 3). There was not a significant difference in rearrest for sexual offenses, but there was a difference in general recidivism rates, with juveniles with peer/adult victims having a higher rearrest rate than juveniles with child victims.

Comparisons Including All Three Subgroups

Mental health

There were no significant differences between groups in rates of mental health diagnoses when juveniles with mixed victims were included in analyses (see Table 4). There were marginal differences in rates of anxiety disorder and substance use disorder. In both categories of diagnosis, juveniles with mixed victims appeared to be more similar to juveniles with child victims than those with peer/adult victims.

OPEN IN VIEWER

Table 4.

Select Comparisons of Juveniles With Child Victims, Peer Victims, and Both Types of Victims

	Child victim group	Peer/adult victim group	Mixed age victim group
	M (SD) or % (n)	M (SD) or % (n)	M (SD) or % (n)	df	Statistic	p
Mental health diagnoses based on K-SADS-PL
Disruptive behavior disorder	38.53% (42)	44.68% (21)	50.00% (6)	2	LR = 0.93	.627
Anxiety disorder	10.09% (11)	0% (0)	8.33% (1)	2	LR = 8.27	.016°
Affective disorder	3.67% (4)	2.13% (1)	0% (0)	2	LR = 1.02	.599°
Substance use disorder	0% (0)	6.38% (3)	0% (0)	2	LR = 7.79	.020°
Criminal history/antisocial tendencies
Any prior nonsexual arrests	6.36% (7)	14.89% (7)	8.33% (1)	2	LR = 2.73	.255
Antisocial process screening device
Callous/ unemotional	3.91 (2.20)	4.18 (2.24)	3.58 (2.61)	(2, 147)	F = 0.40	.670
Narcissism	3.79 (3.13)	3.77 (2.43)	4.75 (3.11)	(2, 147)	F = 0.60	.552
Impulsivity	4.36 (2.19)	4.95 (1.84)	4.67 (1.97)	(2, 147)	F = 1.23	.295
Victim characteristics
Any male victim	35.09% (40)	8.00% (4)	8.33% (1)	2	LR = 13.04	<.001
Victim was biological relative	37.5% (42)	6.00% (3)	8.33% (1)	2	LR = 23.23	<.001
Recidivism
Any rearrest	7.94% (5)	30.00% (9)	11.11% (1)	2	LR = 7.33	.026°
Sexual rearrest	0% (0)	3.33% (1)	11.11% (1)	2	LR = 4.64	.098°

Note: LR = Likelihood Ratio statistic, used when ≥1 cell had an expected count less than 5. K-SADS-PL = Schedule for Affective Disorders and Schizophrenia for School Aged Children–Present and Lifetime Version, APSD = Antisocial Process Screening Device. Maximum scores on the APSD scales are 14 for Narcissism, 10 for Impulsivity, and 12 for Callous-Unemotional.

Criminal history/antisocial tendencies

There were no significant differences between groups on rates of previous arrests or scores on the APSD (see Table 4).

Victim characteristics

Differences in rates of offending against males and against biological relatives remained significant when mixed offenders were included in the analyses (see Table 4). Juveniles with mixed victims appeared more similar to those with peer/adult victims in victim selection.

Recidivism

There was a marginal difference between groups in general rearrest rates, with juveniles with peer/adult victims having the highest rate of general rearrest. There was not a significant difference in sexual rearrest rates.

Discussion

In a sample of juveniles adjudicated for sexual offenses mandated to an outpatient treatment program, a comparison of subgroups based on the ages of their victim(s) revealed more similarities than differences. The few differences identified, however, were consistent with hypotheses drawn from theory and prior research. The results relevant to each hypothesis will be discussed in the context of the existing literature. The study’s limitations and clinical implications will also be addressed.

There were no significant group differences in rates of sexual victimization, which was contrary to Hypothesis 1 but not unprecedented (‘t Hart-Kerkhoffs et al., 2009; Hendriks & Bijleveld, 2004; Ronis & Borduin, 2007). There are several possible explanations for this. First, Worling (1995) noted that juveniles who offend against male children are more likely than other juveniles to have a history of sexual victimization. The current sample included relatively few youth with male victims, perhaps limiting the ability to detect such a difference. Second, it is possible that sexual abuse serves as a risk factor for engaging in sexually abusive behaviors in general rather than specifically for committing offenses against children (e.g., Johnson & Knight, 2000; Seto et al., 2010; Sigurdsson, Gudjonsson, Asgeirsdottir, & Sigfusdottir, 2010). Consistent with this hypothesis, the rate of experiencing child sexual abuse was higher across groups than that found in general population males (7.5% lifetime prevalence; Finkelhor, Turner, Ormrod, & Hamby, 2009) regardless of informant (12% based on child report, 14% based on clinician ratings that abuse occurred, 24% based on parent report). The marginal effect regarding sexual victimization based on parent report may be the result of parents reporting that their child may have been abused because it is a potential explanation for their child’s behavior rather than because they are aware of a specific incident of sexual abuse. Indeed, of the parents who indicated possible sexual abuse, only 34% indicated their child was “definitely” abused, with the majority indicating that such abuse might have happened. Overall, the current results do not support theories that postulate a major role for childhood sexual victimization in the development of sexual offending against children (e.g., Hall & Hirschman, 1992; Seto & Lalumière, 2010; Ward & Siegert, 2002), although findings from previous research suggest this variable may be important (Seto & Lalumière, 2010).

Contrary to Hypotheses 2 & 3, there were no significant differences between groups in mental health problems, including internalizing symptoms and disorders. There was a marginal difference in rates of anxiety disorder diagnosis, and it is notable that no juveniles with peer/adult victims met criteria for an anxiety disorder. Unexpectedly, there was also a marginal difference in rates of substance use disorder, and again, although not significant, it is notable that no juveniles with child victims met criteria for a substance use disorder. Prior research on mental health differences between juveniles with child victims and those with peer/adult victims has been mixed, although there is more support for differences in internalizing than externalizing problems. The lack of significant differences in externalizing symptoms and disorders in the current study is consistent with Hypothesis 3 as well as prior research on externalizing symptoms (Kavoussi, Kaplan, & Becker, 1988), in that overall rates of disruptive behavior disorder diagnosis were high (with general population prevalence rates of ODD and CD being estimated at 3.2% and 3.3%: Canino, Polanczyk, Bauermeister, Rohde, & Frick, 2010; 3.3% prevalence rates for ADHD: Polanczyk et al., 2010), but there were no differences between groups on externalizing symptoms or diagnoses. The implications of potential differences in mental health problems between groups for theories positing a role for psychopathology in the development of sexual offending against children is unclear as all research that has studied this question, including the present study, relied on assessments made after the offense occurred. Nonetheless, there is a pattern across studies suggesting greater anxiety and internalizing problems in juveniles with child victims, which suggests that internalizing symptoms may be more likely to be a focus of treatment in this subgroup.

There were no significant differences between groups on peer problems and social competence in the current study, contrary to Hypothesis 4, and prior research on this topic is mixed. Social skills training is commonly included in treatment programs for adolescents (used by 94% of community and 99% of residential programs; McGrath, Cumming, Burchard, Zeoli, & Ellerby, 2010). Such treatment may be justified by a possible link to recidivism in juvenile populations (Kenny, Keogh, & Seidler, 2001; although not in adult sex offenders, Mann et al., 2010), but given the inconsistency of research on this topic to date, there is little basis for assuming that juveniles with child victims have a greater need for such treatment than those with peer/adult victims.

Based on theory and research on both juveniles (e.g., Seto & Lalumière, 2010) and adults (e.g., Harris et al., 2009; Olver & Wong, 2006), we hypothesized that juveniles with peer/adult victims would be more generally antisocial than those with child victims (Hypotheses 5 and 6). The current study did not detect any differences between groups in past arrests or in antisocial tendencies; however, very few participants had a history of arrests. It is possible that juveniles with more arrests were considered too dangerous for outpatient treatment, resulting in a highly selected, low-risk sample. If this is the case, the low scores on the APSD in the current sample would also be expected, and thus the lack of significant differences should be interpreted very cautiously. If further research supports the hypothesis that juveniles with peer/adult victims are more generally antisocial (e.g., Seto & Lalumière, 2010), the application of empirically supported interventions designed for the treatment of general delinquency, such as multisystemic therapy (MST) and functional family therapy (e.g., Henggeler & Sheidow, 2003), may be especially effective in the treatment of juveniles with peer/adult victims. Such approaches may prove useful with all JSOs regardless of victim type, given research that suggests all JSOs have similarities to general delinquents (e.g., Ronis & Borduin, 2007; van Wijk, Loeber, et al., 2005) and given that MST has been shown to be efficacious in treating JSOs (e.g., Letourneau et al., 2009). Alternatively, outpatient samples like those in the current study may have relatively little in common with general delinquent samples, given the rarity of prior arrests and low scores on a measure of antisocial tendencies. Less intensive interventions may therefore be warranted, and empirically supported treatments for low-risk delinquents may prove useful.

The current study found only one significant difference on a standardized measure of parenting practices consistent with Hypothesis 7: the parents of juveniles with peer/adult victims reported more problems in monitoring their children. Poor monitoring has been associated with general delinquent activity in a variety of research (e.g., Hoeve et al., 2009); the current findings suggest it might play a similar role in sex offenses committed against peers/adults. Recent longitudinal research has found that active efforts to monitor adolescents’ behavior has no effect on future delinquency, suggesting that previous results may be more indicative of the effect of youth disclosure rather than active monitoring activities (Kerr, Stattin, & Burk, 2010). Nonetheless, changes in parenting practices have been shown to partially mediate the effect of MST on reoffense in general delinquent offenders (Huey, Henggeler, Brondino, & Pickrel, 2000) as well as problematic sexual behaviors in a sample of JSOs (Henggeler et al., 2009). Thus, despite current controversies in the field regarding parental monitoring efforts, treatment focused on parenting practices is likely to improve outcomes for JSOs. This may be particularly important for juveniles with peer/adult victims; however, treatment that includes the family has been shown to be effective for mixed samples of JSOs (e.g., Borduin, Schaeffer, & Heiblum, 2009; Letourneau et al., 2009; Worling & Curwen, 2000), therefore family functioning is an appropriate treatment target regardless of victim type.

Contrary to Hypothesis 8, there was not a significant difference between groups in rates of witnessing domestic violence. Furthermore, the marginal effect suggested that those with child victims, rather than those with peer/adult victims, had somewhat higher rates of this experience. Potential explanations for an association between witnessing domestic violence and offending against children are not clear, however, and further research is needed to clarify the role of this experience. In addition, it is notable that there were high rates of witnessing domestic violence and emotional abuse (parent reported) across groups. The rates of these forms of maltreatment in the current sample (47% and 56%, respectively) appear markedly higher than in the general population, as 20.3% of a large, nationally representative sample reported witnessing a family assault and 10.9% reported experiencing emotional abuse (Finkelhor, Turner, Ormrod, & Hamby, 2009). Witnessing domestic violence and experiencing emotional abuse both have been shown to have serious long-term consequences for children (e.g., Chan & Yeung, 2009; Kitzmann, Gaylord, Holt, & Kenny, 2003; Spinhoven et al., 2010; van Harmelen et al., 2010). Witnessing domestic violence is a risk factor for general delinquent behavior (e.g., Zinzow et al., 2009) and has been found to be a risk factor for sexual coercion in a community sample (Sigurdsson et al., 2010). Experiencing emotional abuse has also been shown to be associated with sexual aggression perpetration (among females) and victimization (among males; Zurbriggen, Gobin, & Freyd, 2010). Future research on JSOs should include these forms of child maltreatment, given their potentially important etiological role and the relative dearth of research on each in the JSO literature. In addition, in providing treatment for the child’s own traumatic experiences, clinicians should explore potential traumatization subsequent to witnessing family violence and experiencing emotional abuse, just as for sexual and physical abuse.

Consistent with prior research and with Hypothesis 9, juveniles with child victims were more likely to offend against biological relatives and males than juveniles with peer/adult victims. Neither of these differences support any particular theoretical explanation of the etiology of sexually abusive behaviors nor are there clear indications that such victim characteristics are associated with unique risks or needs that should be addressed in treatment. These victim characteristics may reflect situational factors (e.g., ease of access to young family members) rather than indicating anything particular about the offender in question. Additional research is needed to clarify the significance of these differences in victim characteristics and to determine whether they merit different treatment approaches.

There were no differences in J-SOAP-II scores in the current study, in contrast to prior research and Hypothesis 10. Very few participants had been scored on the J-SOAP-II, and therefore this may be due to lack of power rather than a lack of true differences between groups. In contrast to Hypothesis 11, juveniles with peer/adult victims had higher general rearrest rates than those with child victims. This difference is consistent with the hypothesis that juveniles who select peer/adult victims are more generally antisocial than those who select child victims, but no other studies to date have found higher general recidivism rates among juveniles with peer/adult victims (see Table 2). Thus, there is not sufficient basis to assume this group presents greater risk. The lack of significant differences in sexual recidivism is consistent with prior research directly comparing subtypes; however, a meta-analysis found that offending against any victims significantly different in age (i.e., any victim other than a peer) was associated with sexual recidivism (McCann & Lussier, 2008). As most studies, including the current one, did not separate juveniles with adult victims from those with peer victims, existing comparisons may underestimate the relevance of victim age for recidivism risk. Based on existing research, there is little reason to recommend that one group receive more intensive treatment or be subject to more restrictive social control policies than another.

Comparisons including offenders with both types of victims were included for exploratory purposes, focusing on characteristics that have been included in previous research. The results indicate that those with mixed victims appear similar to offenders with child victims in some respects (e.g., mental health diagnoses) and appear more similar to offenders with peer/adult victims in other respects (e.g., victim characteristics). We were unable to examine some characteristics that appear specific to those with mixed victims based on prior research (e.g., those with mixed victims have been found to have an earlier age of onset and longer duration of offending than both other groups; Richardson et al., 1997). More research is needed, with larger samples of mixed offenders, before conclusions can be drawn regarding this group.