Adolescents with harmful sexual behaviours in New Zealand: could assessment of personality-based classifications help guide therapeutic interventions?

Abstract

The purpose of the present study was to report on a pilot study using a personality-based classification of a community-based sample of 298 adolescent males with harmful sexual behaviours (AHSB) using the Millon Adolescent Clinical Inventory (MACI). Three personality-based profile types were identified using cluster analysis: Cluster 1. Detached, a detached and socially isolated type (n = 97); Cluster 2. Offending Supportive, an antisocial, offending-supportive type (n = 73); and Cluster 3. Typical Teen/Restricted, a type with few clinically elevated personality characteristics (n = 128). Further analyses indicated significant differences among the personality-based profiles on the MACI subscales that measured response styles, adolescents’ concerns and clinical problems. Few statistically significant relationships between the profile types and AHSB, offence and victim characteristics were found. The clinical implications of these findings are discussed, and the need for further research on the use of personality-based classification is recommended.

Keywords:

• adolescents
• harmful sexual behaviours
• Millon Adolescent Clinical Inventory
• personality
• pilot study
• typology

Introduction

Adolescents account for a considerable proportion of all harmful sexual behaviours (HSB; Ryan, Murrie, & Hunter, 2012). In the United States, approximately one in every five sexual offences is committed by persons under the age of 18 years (Federal Bureau of Investigation, 2012), and adolescents who present to Court adjudicated for HSB are responsible for over one third of sex offences committed against children each year in the United States (Sickmund & Puzzanchera, 2014).

Overall, however, it is important to note that adolescent serious violent offending, including sexual offending, in the United States has declined over the last two decades (Oudekerk & Morgan, 2016). Similarly, a more recent meta-analysis found that less than 5% of youth charged with a sexual offence were found to reoffend sexually within a mean follow-up period of 5 years, indicating that adolescents with harmful sexual behaviours (AHSB) are at low risk of sexual recidivism overall (Caldwell, 2010). Nonetheless, AHSB are a heterogeneous group (Everhart Newman et al., 2019; Lambie & Seymour, 2006), and a growing body of research has investigated differences among subgroups of AHSB to increase our understanding of the pathways into such offending and to guide the development of evidence-based interventions (Alexander et al., 2021; Richardson et al., 1997; Worling, 1995a, 1995b, 1995c).

Literature on AHSB has identified several offender, offence and victim characteristics that may be typical of this group (Seto & Lalumière, 2010). Attempts to classify subgroups on victim-based classification have provided mixed results, however, adding limited clarity to our understanding of this population (Hagan & Cho, 1996; Hsu & Starzynski, 1990; Worling, 2001). Fanniff and Kolko (2012) recommended that other methods, such as personality measures, may be more productive than continuing to focus on victim type, due to inconsistencies across studies based on victim age subtypes. Following this recommendation, researchers have explored the classification of AHSB on personality-based characteristics (Everhart Newman et al., 2019; Fox & DeLisi, 2018; Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001).

Despite differences in sample size, country of origin and the use of various personality measures and statistical analyses, most of these studies have identified 4–5 subgroups among AHSB. The most common personality classification across these studies was the Antisocial subgroup (Everhart Newman et al., 2019; Fox & DeLisi, 2018; Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001). As the name suggests, youth belonging to this subgroup displayed symptoms of conduct disorder and impulsivity and engaged in high levels of antisocial behaviour, including generalised offending. For these youth, their harmful sexual behaviour appears to be part of a more general antisocial and offending behaviour pattern. Another commonly identified classification across these studies is the Normal subgroup (Fox & DeLisi, 2018; Richardson et al., 2004; Smith et al., 1987). Relative to the other typologies, adolescents belonging to this group are marked by less psychopathology. For example, Smith et al. (1987) identified two groups who had ‘normal’ range profiles as measured by the Minnesota Multiphasic Personality Inventory (MMPI). Similarly, Fox and DeLisi’s (2018) Non-Disordered group was characterised by very low rates of psychopathology and offending behaviour. Males belonging to this subtype also had low levels of impulsivity and tended to feel empathy for victims (Fox & DeLisi, 2018). As such, there appears to be a subgroup of adolescents for whom harmful sexual behaviour was not directly related to marked psychological difficulties. Rather, such youth may have acted on sexually dominant and aggressive attitudes that are prominent in Western society (Brown & L’Engle, 2009).

Several other typologies have been identified and classified including: Unusual/Isolated, Overcontrolled/Reserved and Confident/Aggressive (Worling, 2001), Submissive, Dysthymic/Inhibited and Dysthymic/Negativistic (Richardson et al., 2004), Inadequate, Conforming and Passive-Aggressive (Oxnam & Vess, 2008) and Anxious, Depressed/Anxious and Dysthymic/Disorganised (Everhart Newman et al., 2019). While precise comparisons of these subgroups are difficult due to differences in sample, personality measures and variables of interest included in each study, findings consistently point to a cluster characterised by personality characteristics related to social isolation, worry and fear, depressive symptoms, passive-aggressive behaviours, internalising behaviours and insecurity. For these youth, their offending may be born out of a need to connect and feel better about themselves.

Table 1 provides an overview of personality-based studies of AHSB.

Table 1. Personality-based studies of adolescents who have sexually offended.

Study	Personality measure	Number of subgroups found	Typologies
Smith et al. (1987)	Minnesota Multiphasic Personality Inventory (MMPI)	4	I. Normal range profile
			II. Most disturbed profile
			III. Normal range profile
			IV. Abnormal range profile^a
Worling (2001)	California Psychological Inventory (CPI)	4	Antisocial/Impulsive
			Unusual/Isolated
			Overcontrolled/Reserved
			Confident/Aggressive
Richardson et al. (2004)	Millon Adolescent Clinical Inventory (MACI)	5	Normal
			Antisocial
			Submissive
			Dysthymic/Inhibited
			Dysthymic/Negativistic
Oxnam and Vess (2008)	Millon Adolescent Clinical Inventory (MACI)	4	Inadequate
			Antisocial
			Conforming
			Passive-Aggressive
Fox and DeLisi (2018)	Positive Achievement for Change Tool (PACT)	4 (males)	Non-disordered males
			Impulsive unempathetic
			Early-onset chronic
			Male Victim offender
		2 (females)	Non-disordered females
			Female victim offenders
Everhart Newman et al. (2019)	Millon Adolescent Clinical Inventory (MACI)	4	Anxious, Depressed/Anxious, Dysthymic/Disorganised, Antisocial

^aThe first normal range profile was characterised by shyness, anxiety, social isolation and positive self-image impression management; the second normal range profile was characterised by frankness, realistic self-image presentation, outgoing personality and appropriate affect.

Despite these findings, studies have not shown similar consistency between personality-based profiles and background offender, offence and victim characteristics (Higgins, 2008; Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001). The relationship between personality subgroups and such characteristics thus remains unclear. It is possible that certain personality typologies may be associated with particular offence profiles and victim characteristics. Such young people may therefore have different therapeutic needs. Personality-based research indicates that AHSB are a heterogeneous group. Given that such youth are likely to have different treatment needs, further understanding of personality-based typologies and their relationship with offence and victim characteristics may assist the development of more responsive and nuanced therapeutic interventions.

Considering the only personality-based typology study based on New Zealand (NZ) data used a relatively small sample size (Oxnam & Vess, 2008), the present pilot study aimed to investigate the classification of AHSB on personality characteristics using a significantly larger NZ sample and to compare these with those previously identified in research. The Millon Adolescent Clinical Inventory (MACI) was chosen for the pilot project as this had been most utilised in previous studies on adolescents with HSB, therefore making direct comparison most applicable. The second aim was to explore the relationship between AHSB, offence and victim characteristics with the identified personality-based profiles.

Method

Participants

A file review was conducted of 298 AHSB who had been referred to SAFE, a community-based harmful sexual behaviour treatment programme located in Auckland, NZ. The types of harmful sexual behaviours committed by adolescents referred to SAFE range from non-contact to penetrative offences. SAFE provides individual, group and family-focused counselling for young people, primarily using a client-centred, trauma-informed, strengths-based therapy, within a systemic framework incorporating broader client systemic support systems.

Inclusion criteria for the present study were that participants were male, were between the ages of 13 and 19 years and had valid MACI scores (Millon, 1993). Participants’ ages ranged from 13 to 18 years with a mean age of 14.74 years (SD = 1.27). The ethnicities of participants were reported as Pākehā/New Zealand European (56.4%), Māori (31.2%) and Pacific Island (7%). The remaining participants (5.4%) were classed as belonging to other ethnicities and classified as ‘Other’.

Young people are referred to SAFE by the child welfare services (Oranga Tamariki) of NZ and undergo treatment as part of their Court diversion. Formal investigation of the young person’s offending typically takes place via Police Youth Aid investigation or an Oranga Tamariki Social Worker assessment.

Selected personality measure

The MACI is a self-report inventory with 160 items measuring personality styles, current concerns and clinical problems of adolescents aged from 13 to 19 years (Millon & Davis, 1993). The MACI is used widely for measuring personality typologies in adolescent clinical populations (Salekin et al., 2002) and is recommended for use in research (McCann, 1999; Millon, 1993).

The MACI is made up of four Modifying Indices, 12 Personality Patterns scales, eight Expressed Concerns scales and seven Clinical Syndromes scales. The Modifying Indices are made up of one reliability scale, which assesses random responding, and three validity scales (Disclosure, Desirability and Debasement), which are designed to measure an individual’s willingness to be open and self-disclosing, as well as positive and negative impression management. The 12 Personality Patterns scales are: Introversive, Inhibited, Doleful, Submissive, Dramatising, Egotistical, Unruly, Forceful, Conforming, Oppositional, Self-Demeaning and Borderline Tendency. Patterns of elevations (configurations) on these subscales provide a style of personality that incorporates cognitive styles, coping mechanisms, behavioural tendencies and biophysical vulnerabilities (McCann, 1999).

The reliability and validity of the MACI have been supported by studies indicating sound psychometric properties (e.g. Blumentritt & Wilson Vanvoorhis, 2004; Penney et al., 2008). Areas that have provided support for the MACI’s ability to differentiate personality styles and related dysfunction have included psychopathy, conduct disorder, depression, substance abuse and childhood abuse (Higgins, 2008; McCann, 1999; Penney et al., 2008).

The Millon inventories are scaled based on a cut-off system identified by prevalence rates in clinical samples (Millon, 1993; Tringone et al., 2007). Items are weighted according to how much the items tap into the core construct. Item weights are summed together to create a total raw score for each subscale, which is then converted into base rate (BR) scores for each scale in line with normative groups (Tringone et al., 2007). BR scores range from 0 to 115. A BR score of 85 or more indicates that the core traits of that scale play a prominent role in the adolescent’s functioning (McCann, 1999; Millon, 1993; Tringone et al., 2007). A BR of 75 to 84 suggests that the characteristics of the scale are likely to be present as features in the adolescent’s functioning. A BR score of 60 to 74 indicates that some of the characteristics of that scale likely play a role in the adolescent’s functioning (McCann, 1999; Millon, 1993; Tringone et al., 2007).

Offender, offence and victim variables

Table 2 describes AHSB, offence and victim-related variables obtained from the files of participants.

Table 2. Definition of offender, offence and victim variables.

Variable	Definition
Age	In years from MACI scoring form at assessment
Ethnicity	Pākehā/New Zealand European: if this was the exclusive recorded ethnicity.
	Māori: if ethnicity was recorded as exclusively Māori or Māori and any other ethnicity.
	Pacific Island (PI): if ethnicity was recorded as exclusively PI or PI and other ethnicity (other than Māori).
	Other: if recorded ethnicity was not Māori, Pākehā/New Zealand European or Pacific Island.
Physical victimisation	Experience of physical assaults by at least one primary caregiver.
Sexual victimisation	Experience of any unwanted sexual contact, or sexual contact with person who was at least four years older than the adolescent at time of incident. These criteria have been used to measure sexual victimisation in earlier research (e.g. Worling, 2001).
Exposure to intimate partner violence (IPV)	Exposure to incidents of physical assaults between older family members/primary caregivers.
High family stress	Experience of a high level of on-going stress within the family prior to his sexual offending, such as self-reported or significant other reported family stressful environment.
Victimisation by bullying	Experience of verbal, emotional or physical intimidation by either his peers or siblings.
Perpetration of bullying	Perpetration of verbal, emotional or physical non-sexual acts of manipulation, teasing or threats.
Use of pornography	Viewed pornographic material via the Internet, videos or magazines.
Living at home	Residing with any family-related member at the time of assessment.
Parental stability	Biological parents were living together in marriage or de-facto relationship at the time of assessment.
Stable relationship with mother	Consistent, supportive and non-abusive relationship with his biological mother.
Stable relationship with father	Consistent, supportive and non-abusive relationship with his biological father.
Consensual sexual experience with peers	Consensual sexual intimacy (touching, kissing or intercourse) with a peer of no more than three years difference in age.
General offending	Engaged in at least one type of offending behaviour other than sexual offending.
Criminal versatility	Dishonesty (stealing/burglary/theft/shoplifting), fire-setting, violence, animal cruelty, property damage (vandalism/tagging) or drug. Categories for criminal versatility were: no other offending, one other type, two to three other types and four or more other types of offending.
Early offending onset	Non-sexual or sexual offending behaviours that began at age 12 or before.
Still at school	Still attending school at assessment.
Behavioural problems at school	Severe or frequent problems with attendance, co-operation and conflict in relationships or offending behaviours (e.g. sexual, dishonesty, violence, drug or property damage).
Academic/learning problems at school	Severe or frequent problems with academic achievement or learning difficulties.
Level of denial	Assessed by clinicians at assessment: a Likert scale of 1 (Adolescent FULLY acknowledged ALL aspects of ALL sexual offences) to 6 (Adolescent denied sexual offending).
Instability of environment	Multiple moves of living residence or schools.
Family history (FH) of sexual offending	Any other family-related member who had committed sexual offences.
Family history (FH) of non-sexual offending	Any other family member who had committed non-sexual offences (e.g. dishonesty, drug, property damage, violence, animal cruelty, driving).
Family history (FH) of physical victimisation	Violence within the family history that did not include perpetration of violence within the current family environment.
Family history (FH) of substance abuse	Alcohol or other drug abuse in the family history.
Family history (FH) of physical/psychological illness	Severe or chronic psychological or physical illness (e.g. depression, psychosis, cancer).
Family history (FH) of economic hardship	Chronic unemployment or chronic financial poverty.
Involvement in hobby or sports	Engagement in any prosocial recreational activity (e.g. team or individual sports, art, writing, reading, computer games).
Types of friends	Classified into four groups (mostly offending supportive, mostly prosocial, mixed, and no or limited friends).
Substance abuse	Past or current (at assessment) alcohol or other drug abuse.
History of self-harm/suicide	Prior attempt to self-harm or suicide, OR ideation of either.
Anal penetration	Actual or attempted anal penetration of at least one victim.
Vaginal penetration	Actual or attempted vaginal penetration of at least one victim.
Exclusively hands-off	Sexual behaviours that did not include any physical contact with any victim (such as exhibitionism or downloading pornographic material from the Internet).
Exclusively genital touching or oral	Sexual touching and/or oral contact with genitals of at least one victim, and there was no anal or vaginal penetration.
Ever bestiality	Any sexual activity with at least one nonhuman animal.
Use of force/threats/weapons	Any evidence from file of the use of force, weapons or threats of either during the sexual offence.
Victim age	Sexually offended against at least two child victims. A victim was considered a child if the offender was at least four years older than the victim and the victim was less than 12 years of age at the time of the offence.
Victim gender	Sexually offended against at least two male victims.
Victim relationship	Sexually offended against at least two stranger victims. A stranger victim was defined as any victim who was not known to the offender previous to the offence (i.e. not intrafamilial full, half, step, adopted, whangai, or foster siblings, cousins, nieces, or nephews, OR known through family or school class mate).

Note: MACI = Millon Adolescent Clinical Inventory.

Procedure

The administration of the MACI was piloted as part of the assessment procedure at SAFE to test its possible utility in assisting clinicians in guiding both assessment and therapeutic interventions. This paper reports on the findings of this pilot study. Tests were administered and scored at the time of assessment by clinicians with strict adherence to administration and scoring guidelines as detailed in the manual (Millon, 1993). The AHSB, offence and victim characteristics were obtained from participants’ hard copy files.

Data analyses

Cluster analysis (CA) is used in research to explore the organisation of data into meaningful structures or groups, which are also referred to as clusters, groups, taxonomies or profiles when exploring personality characteristics (Hair & Black, 2000; Hartigan, 1975; Sharma, 1996). The aim is to ascertain a set of clusters whereby the cases (individuals) within each cluster are similar to each other (homogeneous) but different to other clusters on the characteristics that are measured (Everitt & Dunn, 1991; Hair & Black, 2000; Sharma, 1996).

The following cluster techniques were used in this study: Ward’s method (Ward, 1963) and squared Euclidean distances, average linkage (between-groups) using cosine, average linkage (within groups) using squared Euclidean distance, and k-means using squared Euclidean distances. Although other researchers have included the Expressed Concerns and Clinical Syndromes scales in their cluster analyses (e.g. Higgins, 2008), only BR scores of the 12 Personality Patterns scales were included in this research as this approach is more consistent with Millon’s theoretical perspective of personality (Millon, 1993; Millon & Davis, 1993).

After an optimum number of clusters (personality-based profile groups) were established, a series of analyses of variance and chi-square tests were conducted to investigate relationships with profile groups on the remaining MACI scales (Modifying Indices, Expressed Concerns and Clinical Syndromes) as well as offender, offence and victim characteristics. Post hoc comparisons with a Bonferroni correction were conducted for significant analysis of variance tests. For the analyses of variance tests, a new adjusted alpha was used to assess statistical significance.

Results

Cluster analysis

Using Ward’s method and squared Euclidean distances, the tree-diagram and the coefficient column of the Agglomeration Schedule table indicated that either three or five clusters provided the most within-group similarity and between-group variability. A more detailed inspection of the means across clusters indicated that the five-cluster solution had two groups (Clusters 4 and 5) with no mean scores over the 75 cut-off BR points. This was considered of limited use due to the presence of two groups without mean score elevations. Thus, due to the superiority of results generated from a three-cluster solution using Ward’s method and squared Euclidean distances, coupled with its well-documented use in other relevant research (e.g. Oxnam & Vess, 2008), this was deemed the best solution (see Table 3). The three personality profiles were named Detached (n = 97), Offending Supportive (n = 73) and Typical Teen/Restricted (n = 128). This group was considered more similar to young people who do not engage in HSB.

Table 3. Mean BR scores on each MACI Personality Patterns Scale for three clusters using Ward’s method (squared Euclidean distances).

MACI scale	Cluster
	(1) Detached (n = 97)	(2) Offending Supportive (n = 73)	(3) Typical Teen/ Restricted (n = 128)
Introversive*+	73.76 (13.82)^a	58.81 (11.80)^b	48.34 (15.69)^c
Inhibited*+	75.14 (12.80)^a	49.93 (16.40)^b	46.12 (17.43)^b
Doleful*+	68.65(14.02)^a	61.55 (18.64)^b	26.76 (13.02)
Submissive*	54.40 (16.23)^b	52.03 (10.58)^b	62.67 (11.83)^a
Dramatising*+	35.24 (13.62)^c	54.70 (9.99)^b	61.91 (12.77)^a
Egotistic*+	30.71 (12.51)^c	47.23 (10.15)^b	56.53 (11.42)^a
Unruly*+	61.53 (17.07)^b	75.48 (12.68)^a	57.41 (17.70)^b
Forceful*+	47.70 (22.63)^b	58.74 (20.51)^a	32.97 (17.33)^c
Conforming*+	38.52 (12.64)^c	46.62 (10.60)^b	59.45 (15.22)^a
Oppositional*+	70.23 (11.95)^a	72.53 (8.53)^a	46.42 (16.95)^b
Self-Demeaning*+	68.74 (13.74)^a	52.88 (17.50)^b	26.38 (10.74)^c
Borderline Tendency*+	60.91 (16.25)^a	57.08 (17.01)^a	27.25 (11.05)^b

Note: Standard deviations in parentheses. BR = base rate; MACI = Millon Adolescent Clinical Inventory. The symbol + denotes a large effect size for the scale.

Values with the same lower case letter (a, b and c) were not statistically different from each other.

*p < .004.

Personality Patterns Scales

A one-way between-groups multivariate analysis of variance (MANOVA) was conducted on the Personality Patterns scales. Statistically significant differences were found across personality profile groups, F(24, 570) = 29.08, p < .05, Pillai’s Trace = 1.101, with a large effect size (η²_p = .55). Of the 12 Personality Patterns scales, 11 showed large effect sizes (η²_p ranging from .16 to .65), and one (Submissive) showed a moderate to large effect (η²_p = .11). Size of effect was taken from Cohen’s (1988) guidelines (.01 = small effect, .06 = moderate effect, .14 = large effect).

The Detached (i.e. Cluster 1) group scored significantly higher than the other two personality profile groups on the Introversive, Inhibited and Doleful and Self-Demeaning Personality Patterns scales (ps < .004). The Offending Supportive group (i.e. Cluster 2) scored significantly higher on the Unruly scale than the other two groups, and both the Detached and Offending Supportive groups scored significantly higher on the Oppositional and Borderline Tendency scales than the Typical Teen/Restricted group (i.e. Cluster 3; ps < .004). The Typical Teen/Restricted group had significantly higher scores than the other two groups on the Submissive and the Dramatising scales; however, no mean scores for the Typical Teen/Restricted group reached the BR score of 75 or above (see Table 3).

Modifying Indices, Expressed Concerns and Clinical Syndromes scales

A subsequent one-way between-groups MANOVA was conducted to investigate personality-based differences across the three personality profile groups on Modifying Indices (response styles), Expressed Concerns and Clinical Syndromes scales. There was a statistically significant difference between the personality profile groups, and further post hoc tests showed that differences across groups reached statistical significance using a Bonferroni adjusted alpha level of .002 for all 18 scales (see Table 4). Of these 18 scales, 15 showed large effect sizes (η²_p ranging from .16 to .65), two scales (Sexual Discomfort and Anxious Feelings) a moderate to large effect (η²_p = .09 and .10, respectively), and one scale (Desirability index) a small to moderate effect (η²_p = .05).

Table 4. Mean scores of the MACI modifying indices, expressed concerns, and clinical syndromes scales for the three identified profile groups.

MACI Scale	Cluster
	(1) Detached	(2) Offending Supportive	(3) Typical Teen/Restricted
Disclosure*	71.79 (13.81)^a	62.05 (9.29)^b	33.39 (14.83)^c
Desirability*	52.80 (18.54)^b	59.93 (17.04)^b	61.98 (16.47)^a
Debasement*	73.76 (13.73)^a	57.52 (11.90)^b	39.83 (6.89)^c
Identity Diffusion*	64.69 (19.30)^a	55.79 (16.08)^b	33.74 (12.83)^c
Self-Devaluation*	76.88 (18.82)^a	51.67 (16.32)^b	26.20 (13.70)^c
Body Disapproval*	49.54 (27.84)^a	29.45 (21.43)^b	19.76 (14.60)^b
Sexual Discomfort*	58.94 (17.39)^a^b	53.27 (15.32)^b	66.10 (16.87)^a
Peer Insecurity*	73.65 (21.70)^a	41.47 (21.51)^b	42.90 (20.42)^b
Social Insensitivity*	49.30 (14.58)^b	66.62 (14.10)^a	59.91 (12.88)^a
Family Discord*	65.94 (17.80)^b	76.18 (12.91)^a	57.23 (18.64)^c
Childhood Abuse*	59.12 (26.66)^a	47.75 (25.86)^a	29.37 (22.38)^b
Eating Dysfunctions*	37.24 (27.59)^a	20.44 (15.64)^b	12.85 (9.68)^b
Substance Abuse Proneness*	52.71 (23.93)^a	60.66 (23.40)^a	33.80 (22.70)^b
Delinquent Predisposition*	56.51 (16.27)^c	74.37 (14.20)^a	63.96 (14.93)^b
Impulsive Propensity*	66.65 (18.71)^a	73.37 (14.91)^a	48.02 (20.56)^b
Anxious Feelings*	61.15 (19.92)^a	48.04 (13.98)^b	58.55 (13.10)^a
Depressive Affect*	83.51 (14.09)^a	65.10 (19.04)^b	38.59 (18.83)^c
Suicidal Tendency*	49.86 (25.13)^a	34.16 (21.76)^b	14.35 (9.31)^c

Note: Standard deviations in parentheses. MACI = Millon Adolescent Clinical Inventory.

Values with the same lower case letter (a, b and c) were not statistically different from each other.

*p < .002.

The Detached group was more self-revealing, self-devaluing and self-debasing, had a higher level of current concern regarding their identity and peer relationships, and self-reported more depressive symptoms than did the other two groups. Although the actual scale elevations for a measure of suicidal ideation were not above the cut-off line for any of the groups, it is worth noting that the Detached group had a score significantly higher than those of the other two groups. The Detached and Offending Supportive groups were both more worried about family issues and had more substance abuse problems and impulsivity than the Typical Teen/Restricted group. The Offending Supportive group showed a tendency for antisocial behaviours more than the other two groups and was the least anxious of the three groups. The Typical Teen/Restricted group had no scale elevations that were elevated at, or above, the recommended cut-off for clinical concern.

AHSB characteristics

Of the AHSB characteristics that were investigated using a series of chi-square tests, only non-deviant sexual experience, χ²(2, N = 298) = 6.28, p < .05, academic/learning difficulties, χ²(2, N = 298) = 8.81, p < .05, and suicidal and self-harm attempts/ideation, χ²(2, N = 298) = 38.75, p < .001, were statistically different across the personality profile groups. The Offending Supportive group had a significantly higher proportion of individuals who were recorded as having non-deviant sexual experiences (27.4%), relative to the Detached (12.4%) and the Typical Teen/Restricted (18%) groups. Sixty-six percent of the Detached group reported academic or learning problems at school, which was significantly higher than the Offending Supportive (54.8%) and the Typical Teen/Restricted (46.1%) groups. The Detached and Offending Supportive groups had a significantly higher percentage of those who had engaged in self-harm or suicide attempts (14.4% and 16.4%, respectively), relative to the Typical Teen/Restricted group (5.5%). Furthermore, 34% of the Detached group had engaged in suicidal ideation, compared to 17.8% in the Offending Supportive group, and 7% in the Typical Teen/Restricted group.

Offence and victim characteristics

No significant relationships were found with profile groups on offence characteristics. Anal penetration (or attempted anal penetration) was the only offence-related variable to approach significance, with the Offending Supportive group having a lower percentage of individuals who had engaged (or attempted to engage) in anal penetration.

No significant relationships were found across the groups with victim characteristics. The Offending Supportive group had a lower percentage of individuals who had ever had two male victims as compared to the other two groups, and this approached statistical significance.

Discussion

This study aimed to pilot the use of a personality-based classification measure of AHSB from a community sample and to compare this to previously identified typologies. This study also aimed to investigate the relationships between these personality profiles and AHSB, offence and victim characteristics. A classification of three personality-based profile groups was identified, similar to those found in previous research (Everhart Newman et al., 2019; Fox & DeLisi, 2018; Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001). The three identified cluster groups were named Detached (n = 97), Offending Supportive (n = 73) and Typical Teen/Restricted (n = 128).

There were few statistically significant relationships between offender, offence or victim characteristics and the identified personality-based profile groups. Neither type of offence nor victim characteristics were significantly related to personality-based profiles. This is consistent with previous research in this area (Oxnam & Vess, 2008; Richardson et al., 2004; Worling, 2001).

Personality profiles of AHSB

Detached profile

The Detached group scored significantly higher than the other two personality profile groups on the Introversive, Inhibited, Doleful and Self-Demeaning Personality Patterns scales. The Detached group also scored significantly higher on the Oppositional and Borderline Tendency scales than the Typical Teen/Restricted group (similar to the Offending Supportive group). The Expressed Concerns and Clinical Syndromes scale elevations that contributed to the Detached group’s personality-based profile were Self-Devaluation, Peer Insecurity, Identity Diffusion, Family Discord and Childhood Abuse.

Given these results, it appears that the personality characteristics of the Detached group were isolation from others and chronic psychological helplessness and hopelessness. The Detached group scored significantly higher on Peer Insecurity and Self-devaluation than the other two groups. In addition, they scored higher on anxiety than the Offender Supportive group. These results suggest that their detached style of social isolation and interpersonal interaction difficulties may likely be due to low self-esteem and a fear of anticipated rejection and subsequent psychological pain (McCann, 1999; Millon, 1993). Of the three personality profiles, these adolescents are most likely to experience instability in self, relationships and emotional regulation (McCann, 1999). This is consistent with findings that anxiety and low self-esteem are more prevalent for AHSB than for their general offending counterparts (see Seto & Lalumière, 2010). Consistent with previous research (Everhart Newman et al., 2019; Oxnam & Vess, 2008; Richardson et al., 2004; Worling, 2001), this study therefore also found a group of AHSB with marked internalising difficulties and concerns around social isolation and feelings of hopelessness.

AHSB belonging to the Detached group may be at higher risk of reoffending than the Typical Teen/Restricted group. Worling (2001) found that the Unusual/Isolated typology, which was similar to the current study’s Detached group, was one of the personality-based groups of adolescents that had a higher rate of violent (including sexual offences) and non-violent recidivism (55%). Similarly, adolescent males belonging to the Depressed/Anxious group, which was also characterised by significant internalising difficulties, were found to have high rates of recidivism (48%) in Everhart Newman et al.’s (2019) study. These adolescents may also be more likely to offend against male victims. For example, Fox and DeLisi (2018) found that males belonging to the male victim offender group had high rates of depression, whereas those in their other identified groups did not.

Offending Supportive profile

The Offending Supportive group scored significantly higher on the Unruly scale than the other two groups. This typology also scored significantly higher on the Oppositional, Borderline Tendency and Doleful scales than did the Typical Teen/Restricted group (similar to the Detached group). Given these results, the personality characteristics of the Offending Supportive group are more likely to be antisociality, self-centeredness and impulsivity (McCann, 1999; Millon, 1993). These adolescents likely come from conflictual family environments (McCann, 1999; Millon, 1993).

The Offending Supportive group is likely to be a subgroup of AHSB whose sexual offending may be committed in the context of a general offending lifestyle, in much the same way as previously identified Antisocial subgroups (Everhart Newman et al., 2019; Fox & DeLisi, 2018; Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001). Similar to the Detached profile, this subgroup may be at a higher risk of repeated offending behaviour. Indeed, of the three groups identified by Everhart Newman et al. (2019), the Antisocial group had the highest rate of general offence recidivism (69.1%). It is possible that the Offending Supportive group may also be distinguished by low rates of empathy. For example, Fox and DeLisi (2018) found that males belonging to the Impulsive/Unempathetic subtype, which was associated with high levels of antisocial traits, had low empathy for victims.

Typical Teen/Restricted profile

None of the Personality Patterns, Expressed Concerns, or Clinical Syndromes scales reached the clinical cut-off point presence of personality characteristics in the Typical Teen/Restricted group. This suggests that this group may exhibit fewer problematic personality characteristics than the other groups (Millon, 1993). Rather, their HSB may reflect the extent to which male HSB and offending-supportive attitudes are highly normative in many societies (Brown & L’Engle, 2009). However, two subscales, Submissive and Dramatising, were elevated above 60 BR points; thus, these subscales may reflect the typical characteristics of this profile. In addition, the Expressed Concerns and Clinical Syndromes scales that contributed to this group’s personality profile were Sexual Discomfort and Delinquent Propensity.

The mean score for the Typical Teen/Restricted group on the Disclosure Index suggests a response style of defensiveness and unwillingness to divulge personal qualities and concerns (McCann, 1999). The mean score for the Typical Teen/Restricted adolescents on the Desirability Index indicates that these adolescents may have responded to some extent for the purpose of positive impression management. Therefore, they may have under-reported their concerns or troubles (McCann, 1999). Due to their potentially dishonest response style, it is possible that more openness in responding would show elevated levels such that these personality characteristics would be at a clinically problematic level.

Given these results, it appears that the personality characteristics of the Typical Teen/Restricted group are more likely to be dependency on others, attention seeking, passivity, resentment, rebellion and manipulation (McCann, 1999; Millon, 1993). Due to lack of elevations on the Personality Patterns scales, it is possible that this group is a relatively psychologically ‘healthy’ group of young males. This finding is consistent with much of the existing research on personality-based classifications of AHSB in that, within this population, a relatively ‘normal’ or ‘healthy’ group emerges (Oxnam & Vess, 2008; Richardson et al., 2004; Smith et al., 1987; Worling, 2001). This group may be less likely to recidivate sexually or non-sexually, again which is similar to Worling’s (2001) Overcontrolled/Reserved group.

Relationship between AHSB, offence and victim characteristics and personality profiles

Of the AHSB characteristics that were investigated, only non-deviant sexual experience, academic/learning difficulties and suicidal and self-harm attempts/ideation were statistically different across the groups. The Offending Supportive group had a higher proportion of individuals who were recorded as having non-deviant sexual experience than did the Detached and Typical Teen/Restricted groups. However, given the inter-rater reliability for this variable, differences between groups should be interpreted cautiously. A significantly higher percentage of the Detached group reported academic or learning problems at school than did the Offending Supportive and Typical Teen/Restricted groups. A higher percentage of the Detached and Offending Supportive groups reported having engaged in self-harm or suicide attempts than did the Typical Teen/Restricted group. However, it is worth noting that the Typical Teen/Restricted group may have been under-reporting these difficulties, given that their average score on the reliability index indicated dishonest responding. Thus, the few offender variables that were found to have statistical relationships with the profile groups are not considered to be reliably meaningful for the purposes of the current study. No significant differences were found across the groups on offence or victim characteristics.

The fact that few statistically significant findings were found between the profile types and AHSB, offence and victim characteristics may be due to the nature of the adolescent developmental stage, whereby behaviour is fluid, and persistent offending patterns are yet to be fully established. Alternatively, relationships between AHSB, offence and victim characteristics with personality-based profiles may not exist, and/or are difficult to tease apart if they do.

Clinical implications

So, what could be the practice implications of the three personality-based profiles identified in the current pilot study? The Detached group’s major characteristic is detachment from others, which leaves them socially isolated and limited in appropriate social skills. As other researchers have suggested (Oxnam & Vess, 2006; Worling, 2001), this group may benefit from group-based interventions with a primary focus on addressing their social isolation and social skills deficits. Given their anxiety and depressive symptoms, this group is likely to require safety planning for risk to self, as well as anxiety and depression-related cognitive behaviour therapy (CBT) and more intensive mental health services as required (Ryan et al., 2012). Inclusion of psychoeducation and strengths-based strategies to cope adaptively with substance abuse problems is also likely to be useful. This group would also warrant interventions for the purposes of reducing their risk of recidivism.

The Offending Supportive group’s major characteristic is an offending supportive lifestyle, which includes antisocial attitudes, antisocial associates and antisocial behaviours (e.g. substance abuse and impulsivity). Therefore, intervention at the point of any offending and subsequent to detection of their HSB is warranted. This group may also benefit from group-based interventions with a primary focus on addressing general antisocial attitudes, behaviours and associations, such as cognitive restructuring and increasing skills and rewards for prosocial peer networks (Higgins, 2008; Oxnam & Vess, 2006, 2008; Worling, 2001). In addition, problem-solving and emotional regulation skills training are also likely to address impulsivity and mood-related offending behaviours.

Given their typical disregard for social laws or authority as well as their desire for power and control, the Offending Supportive group is likely to benefit from a focus on the therapeutic relationship (Oxnam & Vess, 2006) and Motivational Interviewing (Miller & Rollnick, 2002), encouraging authenticity, transparency and self-autonomy, as well as working with any current cognitive barriers to change. This group may also benefit from systemic family therapy-based interventions to address the areas they feel most troubled about regarding adverse childhood experiences and family dysfunction such as Multi-Systemic Therapy (Letourneau et al., 2013).

The Detached and Offending Supportive groups reported significantly higher levels of self-harm or suicide attempts than the Typical Teen/Restricted group. This suggests that suicidality may be more common in these two groups and would need to be emphasised in assessment and intervention, and may require the engagement of more intensive mental health services.

The Typical Teen/Restricted group is a subgroup of AHSB who did not present with abnormally elevated levels of personality-based characteristics, due either to these characteristics not being present, or to under-reporting. As noted, this group’s higher levels of dependency on others for direction, as compared to the other two groups, may allow for responsiveness to facilitator-led interventions due to their ability to follow the direction of others. This group is likely to benefit from addressing their need for independent thinking and behaving, particularly with respect to sexuality and other interpersonal interactions.

In this pilot study, few AHSB, offence and victim characteristics had statistically significant relationships with the three personality profile groups. More research is clearly required to understand this more fully.

Limitations

Limitations of this study included the reliance on retrospective self-report. Responses may be either over-reported, due to a motivation to justify HSB, or under-reported, motivated by shame of self-disclosure and/or distrust of others. Information from additional sources, such as caregivers, may have provided valuable insight not gained in the present study. Further, the normative data used for the MACI are from North American samples, as there are no NZ normative data for this test to date. Although the MACI has been used previously in NZ research, the applicability within NZ populations has not been well established. It is well documented that there are unique cultural differences between young people of European descent and youth of Māori ethnicity (the indigenous population of NZ). Given that approximately one third of the current sample were of Māori ethnicity, these characteristics are not captured by the MACI, and hence any findings should be treated with caution, and the findings may have limited relevance.

Conclusion

This pilot study did not provide conclusive support for the classification of subgroups of AHSB on personality-based characteristics. It did, however, provide some partial support for the heterogeneity of AHSB on personality and psychopathology characteristics. Consistent with previous research, this study found three personality-based profiles, Detached, Offending Supportive and Typical Teen/Restricted, which are practical for application to clinical settings and may be used to guide therapeutic interventions. On the other hand, the lack of statistically significant relationships for the majority of offender, offence and victim characteristics may be a consequence of the fluid nature of male adolescents with harmful sexual behaviours with regard to their developmental stage of identity formation, sexuality and related behaviours.

The results of this study could suggest that therapeutic interventions may be tailored according to the specific needs of AHSB, with a focus on personality styles rather than specific offence and victim-related characteristics. More research is clearly needed but it may be that with more conclusive research findings, there could be a benefit in administering personality scales such as the MACI in the assessment phase of treatment to assist in guiding therapeutic interventions according to the individual needs of the young person. More research in this area will help facilitate our growing understandings of AHSB and whether routinely administering personality measures during the assessment phase of treatment is warranted for this commonly considered heterogeneous population.

Clinical impact statement

The current pilot study indicates that there may be benefit in administering personality scales such as the MACI in the assessment phase of treatment to guide therapeutic interventions. However, we recommend more research is undertaken to establish the benefit, or not, of doing so before routine administration of the scale in therapeutic programmes.

Ethical standards

Declaration of conflicts of interest

Maria Purcell has declared no conflicts of interest.

Ian Lambie has declared no conflicts of interest.

Heather McDowell has declared no conflicts of interest.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of The University of Auckland Human Participants Ethics Committee and 2007/350 and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study

Acknowledgements

We would sincerely like to thank the SAFE network for their ongoing support with research on children with harmful sexual behaviours.