Abstract
The present study sought to address gaps in knowledge concerning Australian Indigenous and non-Indigenous youth who commit sexual offenses. Developmental histories and onset sexual offense characteristics of Indigenous (n = 81) and non-Indigenous (n = 130) adjudicated male youth were compared. Results indicate that, in addition to problems affecting both groups, Indigenous youth in this sample were disproportionately exposed to systemic vulnerabilities (e.g., familial antisocial attitudes and incarceration, engagement with antisocial peers, poor school engagement and voluntary school dropout, low socioeconomic status) associated with onset of sexual offending. Differences in the circumstances and context surrounding the onset sexual offense (e.g., use of drugs/alcohol, relationship to person harmed, co-offending, age of person harmed, location, threats/force) were also found. When these analyses were stratified by age of person harmed, these differences were retained only for offenses against children below 16 years. Together, these findings highlight the need for more contextualized primary-, secondary-, and tertiary-level prevention efforts to reduce youth sexual offending in Australia and elsewhere.
Introduction
Young people (below the age of 18 years) are responsible for a significant proportion of officially reported sexual offenses in Australia (Australian Bureau of Statistics, 2015) and internationally (Centre for Effective Public Policy, 2008). In the United States, the National Incident-Based Reporting System indicates that up to one third of the sexual offenses against minors reported to police are committed by youth (Finklehor, Ormrod, & Chaffin, 2009). In Australia, national statistics show that the proportion of sexual offenses committed by youth has increased by 35% since 2008 to 2009 (Australian Bureau of Statistics, 2018).
General patterns of youth offending in Australia highlight a significant overrepresentation of Indigenous young people. Research from the Australian Institute of Health and Welfare (2017) indicates that the rate of youth justice supervision orders for Indigenous youth is 17 times that of non-Indigenous youth (184 per 10,000 compared with 11 per 10,000), and that incarceration rates are 25 times higher than for their non-Indigenous counterparts. This overrepresentation in general offending is also evident within youth sex offending populations. Although there is no national study that provides the prevalence rate of youth sex offending, state-level data indicate that Indigenous youth represent between 12% and 14% of known youth who sexually offend, despite only being 5% of the national youth population (Allan, Allan, Marshall, & Kraszlan, 2002b).
Concerning as they are, the prevalence rates of sexual offenses committed by Australian Indigenous youth are only one component of a complex, systemic problem more broadly affecting sexual violence and abuse in Indigenous communities. Numerous national and state inquiries (e.g., NSW Aboriginal Child Sexual Assault Taskforce, 2006; Robertson, 2000; Wild & Anderson, 2007) have described alarming rates of sexualization of Indigenous children and youth, especially in remote communities, with rates up to 2.5 times that of non-Indigenous children (NSW Aboriginal Child Sexual Assault Taskforce, 2006). Of additional concern is the assertion that up to 88% of sexual abuse within Indigenous Australian communities goes unreported (Robertson, 2000). In an examination of sexual assault against children and youth (below 18 years) by the NSW Ombudsman (2012), the majority of alleged perpetrators were Indigenous adults or youth, either known to the person harmed or residing within their neighborhood or community.
Several factors contributing to the sexual victimization of Indigenous children and youth have been identified in these reports, including the breakdown of Aboriginal culture and customary Lore; vulnerability of children associated with familial neglect, substance abuse, lack of education, and work skills; early exposure of children to sexual activity; and a reluctance to report abuse (Wild & Anderson, 2007). However, very limited research has been conducted to systematically evaluate the contributions these, or other factors, may or may not have on the involvement of Indigenous youth in sexual offending. Comparing the life experiences and circumstances surrounding sexual offending by Indigenous and non-Indigenous youth may identify similar and unique risk (or protective) factors that can help determine the best ways to prevent and respond to this problem.
The limited research that has been conducted in Australia points to important differences in the developmental backgrounds of Indigenous offending youth, particularly in the areas of maltreatment, mental health diagnoses, parental criminality, and separation, that may contribute to offending. Findings from these studies also indicate potential differences in the age at which Indigenous youth commence offending and suggest that Indigenous youth may be overrepresented in relation to sexual offenses. For example, Kenny and Lennings (2007) found that Indigenous offending youth were more likely to have a history of parental incarceration, parental separation or divorce, and a parent currently in prison than their non-Indigenous counterparts. Indigenous youth were also found to be more likely to present with conduct disorder and substance abuse as well as to begin offending at a younger age than non-Indigenous youth (Kenny & Lennings, 2007). Doolan, Najman, Mills, Cherney, and Strathearn (2013) found that incarcerated Indigenous youth were more likely to be subject to a child safety notification and have their maltreatment substantiated by child protection officials than their non-Indigenous counterparts. In their Western Australian study, Allan, Allan, Marshall, and Kraszlan (2002a) identified that Indigenous youth were overrepresented among all youth offenders, and although representing a smaller proportion of those who have committed sexual offenses, this overrepresentation was higher among youth who had committed sexual offenses. It should be noted that for one quarter to one third of cases in this study, the young person’s cultural heritage was not known. However, in a follow-up study (Allan et al., 2002b), no comparisons were made between Indigenous and non-Indigenous youth on demographic or offense characteristics. Although providing some insight into the risks associated with Indigenous versus non-Indigenous youth offending, these studies are limited by the number of Indigenous participants represented in the samples, a lack of replication, and their primary focus on non-sexual youth offending. Furthermore, with the known patterns of differential treatment of Indigenous peoples in the Criminal Justice System (Allan et al., 2002a; Weatherburn & Holmes, 2017) and lack of comparison data on life histories or offense characteristics, these patterns warrant more investigation. Existing studies thus provide a limited knowledge base from which to specifically understand how, and why, Indigenous youth commit sexual offenses and whether, and in what ways, this differs from non-Indigenous youth.
The limited research on sexual offenses perpetrated by Indigenous Australian adults or youth necessitates a continued reliance by Australian policy makers and practitioners on international studies. Of particular interest to the present study, Rojas and Gretton (2007) found that Canadian Aboriginal youth who had sexually offended were more likely than their non-Aboriginal counterparts to have Fetal Alcohol Syndrome (FAS), were engaged in substance use, experienced unstable living environments and childhood victimization, and had struggled at school. Furthermore, they found that Canadian Aboriginal offending youth were more likely than non-Aboriginal youth to offend against girls below 12 years old who were known to them (peers or family), and had higher recidivism rates (specifically violent, sexual, and non-sexual reoffending) in their 10-year follow-up. Some of these findings were replicated by Bertrand, MacRae-Krisa, Costello, and Winterdyk (2013), whose examination of risk and protective factors in Canadian offending youth (including those who sexually harmed) identified significant differences between Aboriginal and non-Aboriginal Canadians. In that study, Aboriginal youth were more likely to also have experienced foster care, criminal involvement by family members, familial substance abuse, and gangs present within their community. These studies have highlighted several differences between the Canadian Aboriginal and non-Aboriginal youth who perpetrate sexual offenses and are now informing the development and implementation of prevention and treatment strategies for Canadian Aboriginal youth.
Although these international advancements are informative, questions have been raised about the applicability of such findings to the Australian context. For example, Smallbone and Rayment-McHugh (2013) suggested that clinical models developed overseas may not account for the unique geographical and cultural features of Australia’s youth offending population. Similarly, based on their examination of the predictive accuracy of risk assessment tools for adult male sexual offenders in Australia, Allan, Dawson, and Allan (2006) recommended that “instruments developed in other countries should preferably not be used with (Australian) Indigenous people without a proper investigation” (p. 65; see also Smallbone & Rallings, 2013). Therefore, it is important to investigate developmental and offense characteristics of non-Indigenous and Indigenous Australian youth who sexually harm to inform local responses in Australia.
Like Rojas and Gretton (2007) and Bertrand et al. (2013), the purpose of the present study is to compare developmental and offense characteristics of Indigenous and non-Indigenous youth who have committed sexual offenses in Australia. Drawing on ecological systems theory (Bronfenbrenner, 1979), we were particularly interested in the multiple systems within which youth are embedded and that influence their behavior. This approach is consistent with integrated theories of sexual offending (e.g., Smallbone & Cale, 2015; Smallbone, Marshall, & Wortley, 2008; Ward & Beech, 2006) and more recent theoretical frameworks approaching this problem through a local, contextual lens (Rayment-McHugh, Adams, Wortley, & Tilley, 2015; Rayment-McHugh, Smallbone, & Tilley, 2015), which acknowledges the diversity that individual characteristics and vulnerabilities have on behavior, but highlights how both ecological systems and the immediate offense environment can influence engagement in sexual offending behavior. This approach appears to be particularly relevant for Indigenous Australian offending youth, given the number of local inquiries that have highlighted the systemic issues associated with sexual abuse within Indigenous communities (Wild & Anderson, 2007). This is suggestive of a difference in the prevalence and impact of systemic factors on involvement by Indigenous youth in sexual offending.
The present study focuses specifically on the first (i.e., onset) sexual offense committed by youth. Recent theoretical discussions have emphasized the importance of focusing on offenders’ onset offense to avoid potentially confounding factors that may be related to offending progression (e.g., changes in individual motivations, impact and consequences of offending; McKillop, Smallbone, Wortley, & Andjic, 2012; Smallbone & Cale, 2015). There has tended to be concentrated research efforts on risks associated with reoffending by known offenders. Focusing on how, and why, an individual sexually offends for the first time provides opportunities to explore ways to prevent these incidents occurring in the first place, which is arguably the ultimate prevention goal.
Echoing Lim, Lambie, and Cooper’s (2012) sentiments, the focus of the present study on a cultural comparison is not to identify cultural weaknesses or to suggest that one cultural group is more criminogenic than another but rather to identify similar and unique factors associated with the onset of sexual offending by Indigenous and non-Indigenous Australian youth to advance current conceptualizations of this problem and inform prevention efforts.
In line with recent theoretical developments (Rayment-McHugh, Adams, et al., 2015; Rayment-McHugh, Smallbone, & Tilley, 2015) and patterns identified in the existing literature, we expected to find differences in systemic factors (e.g., family, peer, school, and community factors) between Indigenous Australian and non-Indigenous Australian youth who commit sexual offenses. Specifically, we expected that it would be the contextual and systemic factors, more so than individual factors, which would distinguish Indigenous from non-Indigenous youth in the onset of their sexual offending.
Method
As recommended by Simmons, Nelson, and Simonsohn (2012) for transparency, the “Method” section reports on how the sample size was determined, all data exclusions, data manipulations, and all measures in the study.
Participants
Participants were 211 male youth, drawn from a larger sample (n = 280) of youth who had been adjudicated for a sexual offense in Queensland, Australia, between 2001 and 2012, and who were subsequently referred to the Griffith Youth Forensic Service (GYFS). GYFS is a statewide service that provides specialist assessments of sexually offending youth to Queensland Courts and provides treatment services to youth and their families throughout Queensland, Australia.
Youth who were referred to the service but not assessed or treated (n = 36) were excluded from the study due to limited file information. Females (n = 5) were also excluded due to the small number being insufficient to examine separately, and a lack of research evidence to justify an assumption of gender similarity regarding developmental history and offense characteristics to confidently integrate the two groups. Youth who did not identify as either Indigenous or non-Indigenous Australian cultural heritage (e.g., Maori, Papua New Guinean, South Sea Islander; n = 28) were also excluded to ensure a reliable comparison between these two cultural groups.
The average age of participants at the time of committing their referred sexual offense was 14.5 years (SD = 1.2 years, range = 11-17 years). Most youth (n = 190, 90%) had been referred to the service for an aggravated sexual assault, and 87.2% (n = 184) had their referred sexual offense dealt with by a judge (Children’s Court of Queensland). Sixty (28.4%) had current non-sexual offenses at the time of referral to service. Indigenous Australian youth comprised 38.4% (n = 81) of the final sample.
Measures
Data regarding participants’ personal histories and onset offense characteristics were collated from official records and clinical files as part of a larger nationally funded project examining the onset and progression of youth sex offending. Coding was completed using the Development History and Office Situation Checklist (McKillop, Smallbone, Allard, & Leclerc, 2014), designed specifically for the project.
Developmental histories
Developmental risk factors were identified as those being present in the youth’s history prior to their onset sexual offense. Characteristics were prespecified, selected from known criminological factors associated with violent and antisocial youth (Andrews & Bonta, 2010; Loeber & Farrington, 1999) as well as previously identified characteristics of Indigenous youth who have sexually harmed (Bertrand et al., 2013; Rojas & Gretton, 2007). The impact of these experiences, attitudes, and behaviors on each participant’s individual, social, and academic functioning was considered in determining presence for each risk factor. Developmental risk factors were categorized as individual-, family-, peer-, school-, or community-level factors. Each risk factor was coded on a scale from 0 to 2. A score of 0 equated to the factor not being present, or not noted anywhere on the client’s case file; a score of 1 was defined as present, and was provided when the information given was not sufficient enough to warrant a score of 2; a score of 2 indicated the factor was definitely present and featured in the client’s file. For the present analysis, all variables were recoded dichotomously (0 = not present, 1 = present); all scores of 1 and 2 were recoded as present (1) for the purposes of this study. Previous examination of coder interrater reliability has identified fair to excellent reliability for most developmental history risk factors (κ = .41-.87; Dowling, 2016). In all, 13 developmental variables were examined.
Individual factors
Four individual-level risk factors were measured in the study.
Antisocial attitudes and behaviors
The antisocial attitudes and behavior variable comprised of several subtypes. These included (a) destruction of property, graffiti, and/or vandalism; (b) lying, theft, or dishonesty; (c) fighting, bullying, or intimidating others; (d) oppositional/defiance, non-compliance, or rule breaking; and (e) trespassing, loitering, or disturbing the peace. Scores were summed across these categories deriving a total score ranging from 0 to 5 for each participant, with higher scores indicating higher levels of antisociality.
Substance use
The substance use variable captured all self-reported use of alcohol and drugs, except cigarette smoking. Specific types of substance use included in the coding comprised alcohol, cannabis, controlled substances (e.g., speed, cocaine, heroin, ecstasy), and inhalants (e.g., glue, aerosols, petrol).
Mental health and behavioral problems
This variable measured any psychological concerns experienced by the participant. Although mental ill-health diagnoses were included in this measure, a formal mental ill-health diagnosis (Diagnostic and Statistical Manual of Mental Disorders [4th ed., text rev.; DSM-IV-TR]; American Psychiatric Association, 2000) was not required for this risk factor to be coded as present. Internalizing behaviors included nervousness, anxiety/depression, withdrawal, and somatic complaints (e.g., dizzy, tired, body aches). Externalizing behaviors included self-regulation problems, non-compliance, verbal or physical aggression, outburst or temper tantrums, and overactivity.
Self-harm and suicidal behavior
This variable comprised of three factors to measure self-harm and suicidality. First, engagement in self-harming behaviors was coded as present if there was any evidence that the participant had engaged in intentional infliction of physical harm as a method of coping with difficult or painful feelings. Second, suicidal ideation was coded as present where there was evidence that the participant had thought about suicide or planned to commit suicide. Third, self-reported or officially recorded (e.g., medical files, police information, prior psychological reports) suicide attempts were coded. A score of 1 was coded if there was a presence of one or more of the factors identified across the three factors.
Family-system factors
Three family-system risk factors were measured, as described below.
Abuse and neglect experiences
Experiences of abuse and neglect as indicated within official child protection records, participant self-report, and information from collateral sources (e.g., parents, family members, police, medical, and allied health reports) were all considered when determining the presence of this risk factor. The global child abuse and neglect risk variable measured whether the participant had experienced any intentional or unintentional acts of omission (e.g., neglect) or commission (e.g., physical abuse, emotional maltreatment, or sexual abuse [experienced contact sexual assault, for example, sexual touching, sexual assault with or without violence]) by any person responsible for their care. This was indicated by the presence of at least one type of incident, regardless of whether it had come to the attention of statutory agencies.
Formal family involvement with child protection agencies
Differing from the child abuse and neglect variable, this variable measured whether anyone in the participant’s immediate family had had formal contact with child protection agencies resulting from officially recognized incidents of child abuse or neglect. Two variables were coded dichotomously. First, contact with child protection agencies included any recorded contact due to concerns for the youth’s safety (e.g., youth is identified as a person harmed). The second variable measured the presence of a child protection notification, leading to investigation of suspected maltreatment or imminence of harm.
Criminal family involvement
Criminal family involvement variable was coded as present if the participant had any parental figure, caregiver, or siblings who had one or more of the following: (a) involvement in crime, (b) prior criminal records, (c) periods of incarceration, or (d) a positive attitude toward antisocial, criminal behavior.
Peer-system factors
Two peer-system risk factors were measured: antisocial peer networks and gang membership.
Antisocial peer networks
This was rated as present if there was evidence that the participant frequently identified and/or associated with delinquent peers (including peer-aged extended family members).
Gang membership
This was rated as present if there was any evidence that the participant identified and congregated with a specific group that shared a mutual identity, provided protection to group members, and engaged in aggression toward others outside the group.
School-system factors
Two school-system risk factors were measured: poor engagement in school and school dropout.
Poor engagement in school
Evidence from file and self-report information indicating lack of interest in school, a devaluing of education/school, consistent poor grades, non-completion of classroom or homework tasks, a lack of motivation regarding school or learning, no articulated future academic goals, and truancy from class or school were used to code this variable.
School dropout
This was coded as present if the participant had ceased going to school on their own volition (i.e., not due to suspension or expulsion) prior to their onset sexual offense.
Community-system factors
Two community-system risk factors were measured: remoteness and neighborhood socioeconomic status.
Remoteness
Remoteness was coded according to the Accessibility and Remoteness Index of Australia (ARIA), which is a widely accepted geographical measurement of location remoteness based on the availability, distance, and accessibility to community services (Commonwealth Department of Health and Aged Care, 2001). ARIA scores range from 0 (highly accessible) to 12 (very remote). Participants’ location was assigned as either non-remote (incorporating high accessible, accessible, and moderately accessible ARIA classifications) or remote (incorporating remote and very remote ARIA classifications) based on their residential location.
Neighborhood socioeconomic status
Participants’ primary residential community was coded according to the Socio-Economic Indexes for Areas (SEIFA; Australian Bureau of Statistics, 2011). SEIFA incorporates four indices quantifying relative socioeconomic advantage or disadvantage, including indices of economic resources and education and occupation, of particular geographical areas. SEIFA decile codes were utilized in coding each participant’s neighborhood, applying a scale from 1 (lowest 10% [most disadvantaged]) to 10 (highest 10% [most advantaged]) communities.
Onset offense characteristics
The onset sexual offense was defined as the first known incident of sexual offending behavior occurring from the age of 10 years (age of criminal responsibility in Queensland) with or without a formal charge. Good to excellent coder interrater reliability for onset sexual offense characteristics (κ = .79-1.00) have previously been found (Dowling, 2016).
Characteristics of person harmed in the offense
Characteristics of the person harmed in the offense included age, gender, and relationship to perpetrator. Age was measured both as a continuous variable and categorized in accord with the age of consent in Queensland (below 16 years and above 16 years), allowing for examination of child- versus peer-/adult-perpetrated offenses. Gender was coded dichotomously (female = 0, male = 1). For the bivariate analysis, the relationship was coded as known relative, known non-relative, acquaintance, or stranger (known for less than 24 hr). For the multivariate analysis, these categories were collapsed into a dichotomous variable (0 = established [relative, known non-relative, and acquaintance], 1 = non-established [stranger] relationship).
Characteristics of youth who committed the offense
Three characteristics of youth who committed the offense were examined. Age of the youth at the time of the onset sexual offense was coded as a continuous variable (e.g., chronological age). Consumption of drugs and alcohol immediately prior to the commission of the onset offense was coded dichotomously (0 = no, 1 = yes). Self-reported intent to commit the offense was also coded dichotomously (0 = no intent prior to offense, 1 = intent formed prior to the offense).
Offense characteristics
Four offense characteristics were examined. The presence of a co-offender during the onset sexual offense, the presence of an adult within the immediate offense location, and the presence of a weapon or violence during the commission of the offense were all coded dichotomously (0 = no, 1 = yes). For the bivariate analysis, offense location was coded into domestic, institutional, or public spaces; for the multivariate analysis, these were collapsed into public and non-public (i.e., domestic and institutional) locations.
Procedure
Data were obtained under the protocols approved by the University’s Human Research Ethics committee and in line with external agency research agreements. Data were coded directly from clinical files by a staff member of the service. File information regarding participants’ developmental histories, mental or physical health, prior and current (sexual and non-sexual) offending histories, and personal accounts of their offending behavior (including their onset sexual offense) was collated from a range of sources as part of the assessment process undertaken by clinicians of the service. These included official documentation (e.g., police, child protection agency), information from professional bodies (e.g., doctors, allied health professionals), and self-report information from the participant, their parents or caregivers, and other significant family members.
Comparisons between Indigenous and non-Indigenous youth on developmental risk factors and onset offense characteristics were first examined at the bivariate level (i.e., chi-square and t tests). The threshold for interpreting effect sizes for these analyses was set as small (.20), medium (.50), and large (≥.60; Ellis, 2009). Variables significant at the bivariate level were then included in a multivariate analysis (binary logistical regression, with Indigenous/non-Indigenous status as the dependent variable), to test some of our expectations regarding the potentially unique influence of systemic factors in the onset of Indigenous and non-Indigenous youth sex offending. The threshold for interpreting effect sizes for the regression analysis was set as small (.10), medium (.30), and large (≥.50; Ellis, 2009).
The recommended criteria for conducting logistic regression analyses were met (Hair, Anderson, Tatham, & Black, 1998) for the present study. Although our analyses included several simultaneous statistical tests from the one data set, due to the exploratory nature of this study, we followed Rothman’s (1990) advice and decided against adjusting significance thresholds, to avoid Type II error.
Results
Characteristics of Youth Who Committed the Offense
Bivariate analyses
Table 1 presents the comparisons between Indigenous and non-Indigenous youth on their developmental histories, prior to the first known (onset) sexual offense.
Bivariate Analysis of Developmental History Characteristics.
Note. Phi (Φ) and Cramer’s V (ΦC) are included for significant findings as an indication of the strength of the association between variables.
Independent-samples t-test values (means (SD)) are reported. Cohen’s d was used to calculate effect size; missing values (4-16 depending on variable).
Individual risk factors
Most participants in this sample were displaying antisocial attitudes and behaviors prior to the onset sexual offense, although the extent of antisocial behaviors was significantly higher among Indigenous youth. Approximately two thirds of the overall sample had a history of substance use prior to their onset sexual offense. However, Indigenous youth were significantly more likely to have a history substance use than their non-Indigenous counterparts. Regarding mental health and behavioral problems of participants, many youth, and similar proportions in each group, experienced some symptoms consistent with internalizing and externalizing mental health behaviors prior to their onset sexual offense. A smaller proportion had engaged in self-harm or suicidal behavior prior to their onset sexual offense, and this was also similar across the two groups.
Family risk factors
Results indicated that majority of youth within this sample experienced some type of child abuse or neglect prior to their onset sexual offense, and at similar rates to one another. Although nearly two thirds of the participant sample were also subject to a child protection notification (e.g., formal contact with a child protection agency) prior to their onset sexual offense, this was significantly higher for Indigenous youth. Overall, just less than half of the youth in this sample were exposed to a criminogenic familial environment prior to their onset sexual offense. However, for youth whose family did have criminal involvement, the significant majority were Indigenous.
Peer risk factors
Most youth within this sample reported engaging with an antisocial peer group or network; however, the proportion of Indigenous youth exposed to an antisocial peer group or antisocial associates was significantly higher than their non-Indigenous counterparts. Few youth, from either group, identified as belonging to a gang.
School risk factors
Overall most of the youth within this sample identified as having poor engagement with school, with about one third dropping out of school prior to their onset sexual offense. However, these patterns were significantly more pronounced for Indigenous youth, with almost double the percentage of non-Indigenous youth disengaged from formal schooling prior to their onset sexual offense.
Community risk factors
Almost all youth residing in remote locations were Indigenous. In addition, SEIFA scores (socioeconomic status of youths’ residential community during childhood development) for Indigenous youth were significantly lower (e.g., more disadvantaged) compared with their non-Indigenous counterparts.
Onset Offense Characteristics
Bivariate analyses
Table 2 presents the comparison between the Indigenous and non-Indigenous youth on the characteristics of their onset sexual offense.
Bivariate Analysis of Onset Sexual Offense Characteristics.
Note. Phi (Φ) and Cramer’s V (ΦC) are included for significant findings as an indication of the strength of the association between variables.
Independent-samples t-test values (means (SD)) are reported. Missing values (1-10 depending on variable).
Characteristics of the person harmed
Most of the youth in this sample committed their first sexual offense against a child or peer-aged female. In less than a quarter of cases, the person harmed was aged above 16 years. In these cases, Indigenous youth were significantly more likely to be identified as the person committing the offense.
Although the majority of onset offenses within the sample were committed against a person known, or related, to the youth who committed the offense, some significant differences emerged within offenses committed against acquaintance and stranger individuals. Indigenous youth were almost 3 times more likely than non-Indigenous youth to offend against an acquaintance and offended against strangers at almost double the proportion of their non-Indigenous counterparts.
Characteristics of youth who committed the offense
Youth were on average 14 years of age at the time of their onset offense, and no significant difference in age was found between Indigenous and non-Indigenous youth. Prior intent to commit these offenses was indicated in just above one third of cases; this pattern was similar across the two groups. However, Indigenous youth were almost 3 times more likely to have used substances immediately prior to the onset sexual offense compared with non-Indigenous youth.
Offense characteristics
Almost three quarters of offenses were perpetrated within the domestic setting. More than half of the onset sexual offenses in this sample occurred while an adult was present within the offense location, and this was similar across the two groups. However, Indigenous youth were 3 times more likely to offend with a co-offender. In addition, Indigenous youth were significantly more likely to utilize threats, violence, force, or a weapon in the commission of the offense and were twice as likely to offend in a public setting compared with their non-Indigenous counterparts.
Multivariate analysis
A binary logistic regression analysis was conducted to examine unique predictors among the variables identified as significant at the bivariate level. Predictors were entered into the model using the block method, with developmental variables entered first, offense onset variables entered second, and Indigenous status as the dependent variable. Due to ad hoc missing data across some of the included variables, the multivariate analysis was limited to participants who had full developmental and offense information history across all predictor variables (n = 157, 74% of the sample).
Remoteness was removed from this analysis as too few non-Indigenous youth were identified residing in remote communities during their childhood development (1%), and thus, its usefulness as a predictor to distinguish between these two groups of offenders was limited. 1 Table 3 presents the results of the analysis. 2
Logistic Regression Analysis of Youth Who Sexually Harm as a Function of Developmental and Onset Sexual Offense Characteristics.
Note. Group membership (0 = non-Indigenous, 1 = Indigenous); substance use (0 = no, 1 = yes); child protection notification (0 = no, 1 = yes); criminal family involvement (0 = no, 1 = yes); antisocial peer group (0 = no, 1 = yes); poor engagement in school (0 = no, 1 = yes); school dropout (0 = no, 1 = yes); victim relationship (0 = established, 1 = non-established); drugs/alcohol immediately prior (0 = no, 1 = yes); presence of a co-offender (0 = no, 1 = yes); utilization of threats, violence, force, or a weapon (0 = no, 1 = yes); offense location (0 = non-public, 1 = public). CI = confidence interval; SEIFA = Socio-Economic Indexes for Areas.
The first block, containing eight developmental risk factors (antisocial attitudes and behavior, substance use, child protection history, criminogenic family environment, antisocial peer group, poor engagement in school, school dropout, and SEIFA scores), was significant, χ2(8, N = 156) = 41.25, p < .001. Overall classification accuracy was 70.7% (74.7% for non-Indigenous, 65.2% for Indigenous youth). Controlling for all variables in the equation, substance use was a unique predictor of Indigenous youth who sexually offend, Wald = 4.15, p = .042. Engagement with antisocial peers was also significant unique predictor of Indigenous youth sexual offending, Wald = 6.25, p = .012. Odds ratios (Exp(B)) for these variables indicated that Indigenous youth were 3.97 times more likely to have a history of substance abuse, and 3.24 times more likely to be associating with an antisocial peer group prior to their onset sexual offense than their non-Indigenous counterparts after controlling for other predictors. Furthermore, controlling for other predictors, being Indigenous decreased the odds of higher scores on the SEIFA, index indicating higher levels of socioeconomic disadvantage among this group.
Six offense onset variables (age of person harmed, relationship to the individual harmed, use of alcohol/drugs, co-offenders, threats/force, location) were entered into the second block. This block produced a significant result, χ2(6, N = 156) = 20.67, p = .002. The final model containing all predictors also yielded a significant result, χ2(14, N = 156) = 61.92, p < .001, and correctly classified 77.7% of youth (82.4% for non-Indigenous, 71.2% for Indigenous). Controlling for all other factors, SEIFA remained a unique developmental predictor in this model, Wald = 13.58, p < .001, with Indigenous status associated with lower SEIFA scores. Similarly, engaging with an antisocial peer group (Wald = 4.11, p = .043) using drugs or alcohol immediately prior to the offense (Wald = 4.42, p = .036) and the presence of a co-offender (Wald = 4.73, p = .030) were significant unique onset offense predictors associated with Indigenous youth sexual offending. Odds ratios (Exp(B)) indicated that, after controlling for other predictors in the model, Indigenous youth were 3.62 times more likely to associate with antisocial peers. They were also 3.62 times more likely to have consumed drugs or alcohol immediately prior to the onset sexual offense, and 4.60 times more likely to co-offend.
Post hoc bivariate analysis
Given some of the differences identified between Indigenous and non-Indigenous youth relating to the onset offense characteristics, additional bivariate analyses were conducted to examine whether the similarities and differences between the two cultural groups remained when the analysis was stratified by the age of the person harmed (below or above 16 years).
Other than geographical remoteness, differences originally identified in the developmental histories and onset offense characteristics between Australian Indigenous and non-Indigenous youth were not observed in offenses committed against persons aged above 16 years. However, as shown in Tables 4 and 5, for persons harmed aged below 16 years, several differences were observed. Higher scores on antisocial attitudes and behavior were found. Moreover, a significantly higher proportion of substance use, criminal family involvement, school disengagement and dropout, as well as engagement with antisocial peers was found in the background of Indigenous youth who perpetrated offenses against this younger cohort. Furthermore, for the Indigenous group, offenses against the younger cohort were similar to those perpetrated against persons above 16 years. They were more likely to include a co-offender, occur in a public setting, and perpetrated with force, threats of violence, or presence of a weapon (see Table 5). A logistic regression was subsequently performed but did not yield any significant differences when stratified by age of the person harmed.
Bivariate Analysis of Developmental History Characteristics (Persons Harmed <16 Years).
Note. Phi (Φ) and Cramer’s V (ΦC) are included for significant findings as an indication of the strength of the association between variables.
Independent-samples t-test values (means (SD)) are reported. Cohen’s d was used to calculate effect size.
Bivariate Analysis of Onset Sexual Offense Characteristics (Persons Harmed <16 Years).
Note. Phi (Φ) and Cramer’s V (ΦC) are included for significant findings as an indication of the strength of the association between variables.
Independent-samples t-test values (means (SD)) are reported.
Discussion
The aim of the present study was to compare developmental history and onset offense characteristics of Australian Indigenous and non-Indigenous youth who had committed sexual offenses. Drawing on ecological systems theory (Bronfenbrenner, 1979) and other more recent contextual frameworks (e.g., Rayment-McHugh, Adams, et al., 2015; Rayment-McHugh, Smallbone, & Tilley, 2015), the characteristics chosen for analysis represented the multiple systems within which youth are embedded and that have the potential to influence their behavior leading up to and at the onset of their offending. Based on prior research in Australia and Canada, it was expected that differences would be found on a range of system-level factors rather than individual-level factors. In addition to problems affecting both groups, Indigenous youth in this sample were disproportionately exposed to systemic vulnerabilities (e.g., familial antisocial attitudes and incarceration, engagement with antisocial peers, poor school engagement and voluntary school dropout, low socioeconomic status). The differences in the circumstances and context of the onset sexual offense (e.g., use of drugs/alcohol, relationship to person harmed, co-offending, age of person harmed, location, threats/force) also highlighted system-level influences on the nature and dynamics of the sexual offenses these youth committed.
Commonalities Between Youth Who Commit Sexual Offenses
Developmental experiences
There were several similarities in the developmental histories of Indigenous and non-Indigenous youth in this sample. Of particular interest was the overwhelming majority of youth, in both cultural groups, who had exhibited a general history of antisocial attitudes and behaviors prior to their onset sexual offense. That this history was evident prior to the onset sexual offense is consistent with previous studies of youths who have committed sexual offenses and reinforces the versatility argument (e.g., Cale, Smallbone, Rayment-McHugh, & Dowling, 2016; McCuish, Lussier, & Corrado, 2014). That is, very few adolescents commit sexual offenses exclusively; rather, they engage in a range of antisocial acts, and their engagement in sexual violence and abuse is an extension of this general antisocial and rule-violating behavior (Lussier, 2017).
Experiences of childhood abuse and neglect were also similar between the two groups. This contrasts with previous findings relating to Canadian Aboriginal (Rojas & Gretton, 2007) and Australian Indigenous (Kenny & Lennings, 2007) sexual offending youth. These prior studies found higher rates of child abuse, specifically sexual abuse, physical abuse, and neglect, among Indigenous youth. The difference in the present findings may be attributed to the different methods used to examine this construct. In previous studies, experience of abuse and engagement with child protection agencies have been incorporated into the one variable, or alternatively, formal child protection history has been used as an exclusive measure of abuse experienced by the youth (Doolan et al., 2013; Rojas & Gretton, 2007). The present study examined youths’ experiences of abuse and neglect separate to the formal notification process and involvement of child protection agencies, an important distinction in Australia given the concerns of overservicing by child protection agencies in relation to Indigenous children (Tilbury, 2009). It should be noted that the present findings did not account for the severity of abuse (e.g., frequency or impact) experienced, which may have also influenced child protection agency involvement. It is recommended that this be considered in future research. Nevertheless, adverse childhood experiences appear to be a common vulnerability factor for many Australian youth who commit non-sexual and sexual offenses.
Nature and dimensions of the onset sexual offense
Regarding the features of the onset sexual offense, some striking and important similarities were also identified. For example, we found that around two thirds of youth in both groups (67.5% non-Indigenous and 60% Indigenous) did not form the intent to commit their onset sexual offense prior to the commission of the offense. This suggests many of these offenses were opportunistic, or perhaps that the situation (e.g., the physical environment, intoxication, presence of peers) itself triggered their offending motivation and behavior (Wortley & Smallbone, 2006). These findings lend support to Wortley’s (2001, 2008) situational precipitators theory that emphasizes the dynamic influences of situational cues on human (and criminal) behavior, even in the absence of prior intent.
Further supporting the situational perspective, in many instances (62.9% non-Indigenous and 51.9% Indigenous), adults were reportedly present at or near the location where the first sexual offense occurred. This suggests that the mere physical presence of an adult at an offense location may not be a sufficient deterrent for youth sexual offending behavior to occur, and suggests the need for more active guardianship and supervision of both potential vulnerable people and those who may harm, particularly in locations of higher risk (e.g., domestic settings), to decrease the likelihood of sexually abusive behavior occurring.
Areas of Divergence Between Youth
Developmental experiences
As expected, our findings highlighted that the most significant differences present in the developmental histories of Indigenous and non-Indigenous youth were located at all levels of the young person’s social ecosystem (familial [antisocial attitudes and incarceration], peer [engagement with an antisocial peer group], and school [poor engagement and dropout]) rather than at an individual level. Potentially compounding these risk factors are community-level factors (i.e., lower socioeconomic status, community isolation, and remoteness), with almost half of the Indigenous youth in our sample living in these conditions. These differences paint a familiar picture of pronounced systemic vulnerabilities for justice-involved Indigenous youth during their childhood and adolescent development. Taken together, these findings situate risk for sexual offending by Indigenous youth within disproportionate, multilayered risks linked to later violent and antisocial behavior, and the potential lack of access to community resources and services to effectively target these risks prior to the onset of their offending (Loeber & Farrington, 1999).
Nature and dimensions of the onset sexual offense
Substance misuse appears to be a chronic and acute risk factor particularly for Indigenous youth. Although many of the youth in our sample reported engaging in some form of substance use prior to the onset of their sexual offending, Indigenous youth were significantly more likely to be utilizing alcohol, cannabis, and inhalants, both prior to and at the time of their offending. These findings are consistent with previous research involving Indigenous offending youth in Australia and Canada (Kenny & Lennings, 2007; Rojas & Gretton, 2007) and reinforces the argument made by Weatherburn, Snowball, and Hunter (2006) regarding alcohol as a significant predictor in Indigenous offending. The present findings help to advance the current understanding of how alcohol or drug consumption may differentially affect Indigenous versus non-Indigenous offending, and suggest it is a key marker for prevention and intervention for Indigenous youth. They also highlight the role of situational precipitators (i.e., disinhibiting effects of intoxication) for sexual violence and abuse perpetrated by youth.
Other differences in the characteristics of the onset sexual offense were apparent between the two groups. In the present study, Indigenous youth were more likely to commit their first sexual offense against individuals who were peer aged or older, compared with their non-Indigenous counterparts, who tended to offend against younger people. This finding is in contrast with findings from other youth sexual offending research, and studies of Canadian Aboriginal youth who sexually offend, which found that these youth are more likely to sexually offend against those younger than themselves (Langstrom & Lindblad, 2000; Rojas & Gretton, 2007). Given our results were also found by Allan and colleagues (2002b) in their examination of Western Australian youth who have sexually harmed (of which 30% of their sample identified as Indigenous Australians), this significant difference in age of person harmed might be suggestive of a different pattern for Indigenous Australian offending youth than that typically attributed to youth who commit sexual offenses.
This difference in the age of the person harmed may have also influenced the significant differences found in the relationship between the individuals involved in the offending behavior, as well as the offense location. In this study, most non-Indigenous youth (81%) committed sexual offenses against known (relative or known but non-relative) individuals, compared with just more than half of the Indigenous youth (59.2%). Rather, 21% of Indigenous youth committed their onset sexual offense against a stranger, compared with only 12.4% for non-Indigenous youth. Furthermore, two thirds of Indigenous youth (63%) committed their offense within a domestic setting, whereas almost one third (29.6%) of Indigenous youths’ onset offense occurred in a public setting—double that of their non-Indigenous counterparts.
Other points of difference found between these two groups included the proportion of incidents that involved the utilization of threats, violence, force, or a weapon during the onset sexual offense, as well as the presence of a co-offender. Overall, neither force nor co-offending was a hallmark feature of sexual offending by youth in this sample. However, these offense characteristics featured more prominently among Indigenous youth. Of particular interest, our post hoc analyses indicated that these differences were most pronounced when the analyses were stratified by age of person harmed. In contrast to expectations, offenses committed against children by the Indigenous youth in this sample were more likely to include a co-offender, the use of force or weapons, and occur in a public setting. Differences in these offense patterns were not observed between Indigenous and non-Indigenous youth when the person harmed was a peer or adult. Neither of these findings have been previously identified in cross-cultural research to be a feature of Indigenous youth sexual offending. As such, these patterns require further theoretical and empirical investigation, as part of a more comprehensive understanding of factors associated with the onset of sexual offending by youth.
Interpretation of the Findings
Overall, the present findings lend tentative support to the assertion that unequal exposure to systemic criminogenic factors (e.g., family, school, peers, and community) account for some of the differences found between Indigenous and non-Indigenous youth who commit sexual offenses. Multivariate analyses indicated that certain developmental and onset offense characteristics significantly distinguished Indigenous versus non-Indigenous onset sexual offending. Of particular note was the finding that higher socioeconomic disadvantage, which is typically also linked to a number of individual- and system-level vulnerability markers found here at the bivariate level, was a unique predictor of group membership. These reflect vulnerability factors already established in the broader youth offending literature and serve to reinforce recommendations for the development of primary- and secondary-level, community-based prevention initiatives as part of a multilayered approach to prevention of youth offending generally, which may also reduce risk and prevent the onset of sexual offending behavior by youth.
From this perspective, it may also be worth considering whether the developmental differences identified between these two groups may not be solely attributable to differences in cultural heritage. Rather, they may be a reflection of the social and economic conditions of the “place” in which they reside that affect individual and social development and that are more conducive to involvement in antisocial and offending behavior, and by extension, sexual offending behavior.
This may also help to explain differences found in the onset offense characteristics (e.g., utilization of drugs/alcohol, co-offending) of Indigenous and non-Indigenous youth, which could perhaps also be understood as a reflection of “place.” The context in which young people interact with others, as part of their social development, as well as their access to, and use of, public and private spaces may take a different form, dependent on the place (urban, rural, or remote) in which they reside. Thus, the level of risk, and contexts for risky behavior, may differ according to location. Such examination of “place” in the prevention of youth sexual violence and abuse is already occurring (see Rayment-McHugh, Adams, et al., 2015; Tilley et al., 2014), and it is recommended that future research consider more nuanced approaches that attend to some of the possible contextual risks and varying socioeconomic advantage, specific to the community itself, to extend existing offender-focused prevention initiatives.
Based on our findings, engagement with an antisocial peer group appears to be particularly relevant for Indigenous young people. Although the risk of antisocial peer associations is not a newly identified risk factor, nor unique to Indigenous youth, in our study these peer associations are hypothesized to be contributing to the higher rates of co-offending in our Indigenous youth sample. Even so, within the Indigenous sample, this was not typical in the perpetration of sexual offenses; only a fraction (21%) had an accomplice for their sexual offenses; yet, 85% of these youth were associating with criminal peers at the time of their onset offense. This raises the question as to what extent, these variables uniquely discriminate Indigenous and non-Indigenous youth who have perpetrated sexual offenses from those who perpetrate non-sexual offenses. Whether antisocial peer involvement is incidental to the onset of sexual offending, and thus a better predictor of general offending behavior (of which sexual offending becomes a component of this antisocial behavior), is difficult to determine from this study. The inability to distinguish co-offending and antisocial peer associations as unique correlates of youth sexual offending is a limitation of this study and requires further investigation.
Study Limitations
The present findings should be considered in light of some methodological limitations. First, the group comparisons were limited to cultural heritage only; this was done simply as a stepping stone to understanding the similarities and differences between these populations of youth. Furthermore, the statistical methods we employed within this study (bivariate analysis and correlations) make it difficult to identify and understand the causes of the differences between these two groups. Future investigations would benefit from including more comprehensive proxy measures to reflect the complexities of cultural heritage to extend the present investigations. We also focused on risk factors for the present study. Examining potential protective factors for both Indigenous and non-Indigenous youth would help to advance the conceptualization of this problem, and inform its prevention.
The findings are based on a sample of youth adjudicated for sexual offenses, and as such might not be representative of the whole population of youth who commit sexual offenses. Focusing on adjudicated youth, and in our sample those who were sentenced in higher courts, may be selective of more serious youth sexual offenders, and so, the results of the present study may be more representative of the development and onset sexual offense of serious offending youth rather than the wider population of youth who sexually harm. The multivariate analysis was also affected by some missing data across the study variables, reducing the sample size for these analyses, and may have affected the true number of potential predictors in this analysis. Therefore, we remain tentative in our interpretations of these findings.
Current and Future Directions
Despite these identified limitations, this study has highlighted some important and meaningful distinctions in the developmental histories and onset offense characteristics of Indigenous and non-Indigenous Australian youth who have engaged in sexually harmful behavior. These lead to several theoretical, practice, and research directions.
Theoretical directions
It is hoped that the findings of the present study will contribute to theorists’ conceptualization of sexual offending perpetrated by Indigenous and non-Indigenous Australian youth. The differences in developmental history risk factors provide some tentative support for recent theoretical propositions highlighting the level of systemic vulnerabilities that may influence the onset of youth sexual offending, and perhaps will encourage new theorizing on their specific applicability for Indigenous Australian youth (see Smallbone & Cale, 2015). Furthermore, the significant differences found in the onset offense characteristics may encourage more critical examination and theoretical discussion of how and why Indigenous youth’s onset offenses differ from that of their non-Indigenous counterparts and characteristics typically identified in youth sexual offenses internationally.
Practice directions
For practitioners, this study highlights the significant role that antisocial peers play in the development and onset sexual offense particularly for Indigenous youth, as well as the proximal and dynamic impact that drugs and alcohol play in the onset sexual offense. These findings are suggestive of the need for primary-level and secondary-level prevention initiatives that are locally informed and driven, targeting factors associated with antisocial peer associations and substance misuse, not only to reduce general offending but also for the benefit of potentially reducing the occurrence of youth sexual violence and abuse by Indigenous youth in community. We are aware of several government-led initiatives currently focused on the prevention of youth sexual violence, including within regions covered in the present study. Future investigation and evaluation of these initiatives will help determine the utility of these measures for preventing youth sexual violence.
At a tertiary level, these findings will also assist practitioners in focusing on these broader antisocial factors in the assessment and treatment of those youth at risk of developing sexual offending behaviors. The identification of a differing offense pattern for Indigenous youth may assist in the assessment of youth who offend and lead practitioners to examine their assessment processes and instruments to ensure they adequately capture any unique characteristics associated with the onset of Indigenous sexual offending behavior.
Research directions
Further to this, the focus of the present study was the onset sexual offense to enhance our current understanding of why youth engage in this behavior in the first place. It would be remiss, however, not to question whether, and how, these same factors potentially influence recidivism outcomes across these groups. In our previous study (Allard, Rayment-McHugh, Adams, Smallbone, & McKillop, 2016) examining recidivism outcomes for Indigenous and non-Indigenous youth, we found that field-based interventions addressing sexual violence from an individual and ecosystemic perspective, undertaken within the youth’s community, significantly reduced the level of sexual recidivism among Indigenous youth (including for those in remote communities). This suggests that targeting individual, ecological, and situational risks (such as those subject to investigation in the present study) in assessment and treatment may be particularly well suited for reducing sexual recidivism in youth populations. However, we did not observe the same results for non-sexual (violent and non-violent) recidivism outcomes. Acknowledging that the program was designed specifically to target sexual offending behavior, our findings do raise questions as to the types of responses best suited to addressing general versus sexual offending by youth.
Similarly, Olver et al. (2018) found differences in static and dynamic risk factors that predicted observed rates of sexual recidivism between Indigenous and non-Indigenous adults convicted of sexual offenses; yet, this was not so much the case for broader (e.g., violent and general) recidivism outcomes. This, like our findings, indicates other likely contextual factors unique to Indigenous peoples may need to be considered. Further investigation is thus warranted to tease out factors unique to sexual versus general offending, in both adults and youth, to inform prevention, risk assessment, and treatment responses. We hope that our findings continue to stimulate discussions and research on this issue.
Finally, one of the key contributions of this research is in the extension of thinking and consideration given to risks associated with “place” and the detrimental impacts on the systems that comprise it, as this may differentially influence pathways to, and commission of, youth sexual offenses. This is in line with recent developments in scholarship around the importance of widening the scope of prevention to include more contextualized and place-based approaches to sexual violence and abuse (Firmin, 2017; Rayment-McHugh, Adams, et al., 2015; Rayment-McHugh, Smallbone, & Tilley, 2015) that can be tailored to the unique risks that comprise those places, in addition to interventions that already consider other important individual and cultural factors. Finally, it draws attention to the dynamic and powerful role of the environment and situational cues in facilitating the commission of sexual offending behavior. These theoretical and research developments are likely to have a significant influence on future sexual violence prevention and intervention responses in Australia and internationally.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported under an Australian Research Council’s Discovery Projects funding scheme (Project: DP110102126) with the support and assistance from the Department of Justice and Attorney-General (Youth Justice) and the Queensland Police Service. The views expressed herein are solely those of the authors, and do not necessarily reflect the views or policies of these organizations.
Statistical Significance Statement
The authors take responsibility for the integrity of the data and the accuracy of the data analyses, and have made every effort to avoid inflating significant results.
