CHILD-ON-CHILD SEXUAL ABUSE : RESULTS OF A SURVEY IN JOHANNESBURG

Child-on-child sexual abuse in South Africa has been recognised only recently as a significant social problem, reflected in the dearth of research on the topic. There is also a lack of evidence in South Africa on the extent of abuse and issues that relate to youth sex offending (Ehlers & Wood, 2001; Mbambo, 2002). It is difficult to establish the extent of either child-on-child or child-on-adult sex offending in South Africa, because not all of these cases are reported or recorded (Stout, 2003). It is estimated, however, that 42% of sexual offences reported to Childline, a national help-line providing crisis intervention services, are committed by other children (Vanzant, 2004) and the latest statistics available from the Department of Correctional Services (2007) show that on 31 January 2007 a total of 288 children were in prison for crimes of a sexual nature. It was reported in the Pretoria News that in South Africa a daily average of 82 children were charged for indecently assaulting or raping other children (Maughan, 2006:1). There is also a dearth of empirical international research and literature on the origins of child-on-child sexual abuse (Lightfoot & Evans, 2000).

offenders.This in turn could reduce the increasing burden being placed on law enforcement, legal, medical and social welfare resources, including non-governmental organisations (NGOs).

CONCEPTUAL FRAMEWORK FOR THE STUDY DEFINITION
Child sexual abuse is any sexual act between a person (adult and child, or child and child), perpetrated for the gratification of the one who is more powerful in terms of age, emotion, physicality and intellect.This definition includes both contact and non-contact sexual acts.Non-contact abuse refers to exhibitionism, voyeurism and exposure of children to pornography, or their involvement in its production.Contact abuse refers to oral sex, digital penetration (fingers or foreign objects inserted into the genitalia), and vaginal or penile penetration (Jacklin, 2000;National Department of Health Policy Framework for Child Abuse, 2003).Hence, a child sex offender is a child committing a sexual act on another child which includes any one or more of the above-mentioned acts.

THEORETICAL FRAMEWORK
The ecological model was used as a point of departure for the conceptual framework.However, it is critical that a brief critique on single-factor theory is discussed in order to appreciate the value of the ecological model.Previous research on single-factor theory identifies a range of factors that may lie within the individual or his or her environment.Many researchers isolate a single factor that may be a causal variable in explaining violent behaviour in adults and children.For proponents of social learning theory such as Bandura (1977) and Ross and Loss (1991), who are proponents of violence theory, contend that the development of violent behaviour patterns is a result of exposure to violent role models and a culture of violence in the society (Altbeker, 2007).Others such as Shalev, Yehuda and McFarlane (2000) are of the view that the mastery of a crisis or trauma and failure to master important life stages may result in maladaptive behaviour.Furthermore, socio-behavioural theories and behavioural theory take cognisance of the role of significant others, looking specifically at the family, whilst attachment theory focuses on the mother.Gender theory, meanwhile, is concerned with inequality of the sexes and how male power and control are explanatory factors in gender-based violence.
However, while each of the theories mentioned above provide a useful lens for understanding sexual violence among children, some of these may be somewhat limiting, reductionist and deterministic.For instance, the biological theory discounts the role of other contributory factors, namely psycho-social.Some psychological theorists do not consider biological traits, but only mental traits that make people prone to commit crime.Behavioural and socio-behavioural theory reduces an individual to a passive being and does not take into account individual differences or personality development.Psychoanalytic theory, on the other hand, could also be deterministic in that earlier experiences are perceived to determine future behaviour without taking account of later life events, and it does not consider the role of support systems and social interventions in mediating the impact of such early life experiences.
Similarly, while the sexual abuse cycle theory identifies the victim-victimiser pattern, it does not take into account why many victims do not become victimisers.The PTSD (Post-Traumatic Stress Disorder) theory refers to the need to re-enact a trauma but, again, no cognisance is taken of the impact of support and intervention.This summary illustrates a few of the single-factor theories directed at the individual in an attempt to explain the causality of sex offending by young adolescents largely on an individual level.The challenge is that no single theory can singlehandedly explain the reason for some children offending and not others (Pithers, Gray, Busconi & Houchens, 1998a, 1998b;Silovsky & Niec, 2002).According to these theories, all human behaviour is determined by biological, psychological or environmental factors, or a combination of these.Muncie (1999:98) comments that these theories focus on individual motivations and provide specific representations of the young offender as either "born bad" (genetic) or "made bad" (inability to develop a conscience).
Thus, the single-factor theories discussed in the literature review do not adequately explain the aetiology of sex offending.Instead, Bronfenbrenner's (2004) eco-systemic approach informed this study.In the former approach the child is the focal point or at the epicentre of the model, but for Bronfenbrenner the dyad, or two person system, is the basic unit of analysis.This dyad is usually between caregiver and child who come into contact with other people and settings that are influenced by and in turn influence them.This dyad is the connection between systems, thus making space for other systems.

MICROSYSTEM
The interaction and reciprocity between the child and his/her environment is not a linear cause and effect process, but a circular and mutually reciprocal one.It can be inferred that the family plays a critical role in the development of the child as it is in this context that the child spends most time and, hence, where most profound learning is absorbed.The family is also part of the whole and cannot be seen to exist in isolation.

MESOSYSTEM
Other influential settings could include family members, healthcare facilities, neighbourhoods, early care and educational settings, playgrounds or recreational facilities.The frequency and quality of the association the child has with his/her family, school and neighbourhood may also influence development, as illustrated by the lack of communication between parents and educators, and disparities in their expectations of the child.

MACROSYSTEM
A child is not only affected by settings in which he/she spends time, but also by settings where no time is spent.This could include the parent's workplace or something more remote, such as policy and legislation (Bronfenbrenner, 2004).

MULTIVARIATE APPROACH
The systems thinking of Ward (2007) and Cole and Cole (2001) draw on Bronfenbrenner's ecological perspective, but both Ward (2007) and Cole and Cole (2001) are more comprehensive and multivariate.As Bronfenbrenner (1979) argues, there is no single, all-encompassing theory that can be employed to explain the behaviour of all offenders.Furthermore, the management of child-on-child sex offenders should always consider their individual differences but this alone is not sufficient, a multivariate approach to aetiology is needed.The researcher's approach to social work practice was informed by Kirst-Ashman and Hull's (2002) generalist practice approach, defined as the utilisation of an integrated theoretical paradigm synthesised with professional integrity, ethics and values, as well as an expansive knowledge base and intervention in systems of all dimensions.This approach recommends that social workers acquire a large range of skills at different levels, namely micro, mezzo and macro levels, and place an emphasis on the strengths of the client system.This perspective is broader than the social treatment approach in that it takes into account multiple levels of practice.Patel (2005) argues that the generalist practice model is concerned with building on the strengths of clients, groups and communities rather than with their inadequacies and pathologies.This model addresses the strengths perspective, prevention, rehabilitation and multiple causation and is strongly associated with the notion of empowerment.It forms an integral component of the social development practice model devised by Patel (2005) and is compatible with Bronfenbrenner's ecological model of the child.It therefore provided a sound frame of reference and a conceptual foundation for the study.

METHODOLOGY
An exploratory descriptive research design was followed and included a mixed methods or two-phase approach.The quantitative method comprised 50 questionnaires administered to 50 boys aged 6 to 12 years and their mothers, who were referred to the Teddy Bear Clinic for treatment of sexual acting out behaviours.The qualitative method involved a document study of the case files of the children.However, this article focuses only on the quantitative component.
A non-probability, purposive sampling method was used to recruit research participants.Although the research focused on children who sexually abused other children, the criteria for the selection of suitable child sexual abuse cases were stipulated as follows:  The offending child needed to be a male;  The child needed to have sexually abused another child;  The child was required to be 12 years or younger;  The child was required to be involved in a form of intervention at the Teddy Bear Clinic at the time of the study;  The child had to have been referred by the court (external control).Child-on-child sexual abusers who were clients of the Teddy Bear Clinic and who fulfilled the selection criteria were included in the sample, selected from the total population at the research site for the year 2007, with a total of 505 children referred for diversion.Thus, the sample constituted 10% of the population.There were two sets of participants involved in the data collection, namely the 50 young sex offenders themselves, all of whom were referred into a diversion programme, and their mothers.The researcher is of the opinion that these respondents were representative of the population of young perpetrators referred to the clinic in 2007.
For the purposes of this study a self-administered questionnaire was applied.There were three parts to the questionnaire, the first of which was completed by the mother, if necessary with assistance from the researcher.The emphasis was on their socio-economic and demographic situation and the nature of the domestic environment.The second part was completed by the child, and once again, where requested, the researcher assisted in the completion of this section.The focus was on the characteristics of the child offenders.The third part was completed by the researcher, in which the child was classified according to O' Brien and Bera's (1986) typology and conceptual framework for assessment purposes, which informed the level of risk and the intervention which is applied at the Teddy Bear Clinic.These questionnaires were completed in English by the researcher, with the assistance of an auxiliary social worker who spoke the indigenous African languages.
The quantitative data were analysed by means of the Statistical Programme for the Social Sciences (SPSS, 2009), a computer software programme that enables researchers to conduct a range of statistical analyses.The researcher interpreted the data collected from SPSS in order to organise the findings into meaningful results.
The following limitations inherent in the research design need to be noted.First, the use of a social auxiliary worker as an interpreter may have resulted in the loss of some vital information as there was no forward or backward translation used in the study.Second, although the participants were informed that they could withdraw from the study at any stage, it is possible that they may have been afraid to withdraw and may have provided socially acceptable responses.A third limitation related to the research tools.The questionnaires elicited fixed responses which did not allow for elaboration.Finally, the study was gender-specific, with only male children, because very few females were referred to the research site.

RESULTS
The research attempted to develop a detailed picture of the respondents by presenting the research findings in relation to key characteristics of the child respondents in the study, with particular emphasis on their maturational stage of development, their family context (the role of the family in the development of the child as well as relationships between the child and his or her attachment figures), and identifying social and environmental influences and risk factors for child-on-child sex offending.In view of the fact that all the children were male, results are reported in the masculine gender.

CHARACTERISTICS OF THE CHILDREN ABUSING OTHER CHILDREN MATURATIONAL STAGE OF DEVELOPMENT
The majority of the boys (48 or 96%) ranged in age from 9 to 12 years, and a minority (4%) were between 6 and 8 years old.This finding was consistent with statistics generated in a study of over 1 000 children treated at the Teddy Bear Clinic in the previous eight years.It was found that the perpetrators' ages ranged between 6 and 12 years.The findings in this research appeared to confirm those reported by Delany (2005), Richter, Dawes and Higson-Smith (2004), and Graves, Openshaw, Asciane and Ericksen (1996), namely, that a large number of perpetrators of abuse against children are other children, and not adults.Puberty or pubescence is the time of sexual maturation and achievement of fertility, associated with the development of secondary sex characteristics and rapid growth (Sadock & Sadock, 2003).For the purposes of this study, boys between the ages of 12 and 16 were considered to be pubescent, with those aged between 8 and 11 years generally considered to be in the pre-pubescent phase, i.e. the transition from childhood to puberty (Sadock & Sadock, 2003).In this study, the majority of the young sex offenders were either in the pubescent phase, while only a few were in the pre-pubescent phase.This finding suggests that the boys were undergoing physiological, biological and physical changes in their bodies that were likely to impact on their emotional wellbeing.The feelings they were likely to experience could range from sexual arousal to confusion, and hence result in trauma resulting in sexualised behaviour.

Penetration
Figure 1 illustrates that the high-risk category ranged between 10 and 12 years, with the highest incidence of sexual abuse being perpetrated by 12 year olds, in the late pre-pubescent phase (10-12 years) and the entering of puberty (12-16 years).The offences described in Figure 1 consisted of fondling, attempted rape, being an accomplice or observer, and rape involving penetration.Figure 1 highlights that the majority of offenders committed these offences on one occasion only.

OFFENDERS' CHOICE OF VICTIMS
The great majority of victims were not family relations, but were known (88%) and had a relationship of some sort with the perpetrator, making it easier to gain access.Both males and females were victimised, and though female victims (82%) were far more frequently victimised, irrespective of their age, the potential risks of males also being victims cannot be overlooked.The majority of the victims were peers and younger than the offender, but very few were older.This finding can be attributed to the power exerted by the young sex offender over the victim.

EDUCATIONAL STATUS AND PERFORMANCE OF THE YOUNG OFFENDERS
All the children in the study were reported to come from mainstream schooling.They were not in bridging or remedial classes, and were not from any of the special needs schools.However, a limitation of the study was that the schools and educators were not interviewed to confirm or refute these findings.

Junior Primary School 20%
Senior Primary School 60%

Junior high School 18%
No Response 2% Figure 2 describes the level of education of the respondents.Nearly two thirds (30 or 60%) of the young sex offenders who participated in the study were in the senior primary school.A further 10 or 20% were in the junior primary school, while 9 or 18% of the boys were in junior high school, and 2% gave no response.This finding indicates that the children in the senior primary school fell into the highrisk cohort of respondents.These findings concur with those of Blankenship and Winokur (2010), who found that most of the sexual offences committed by the youths occurred mostly between seventh grade and high school, which corresponds to senior primary and junior high school in South Africa.

Failed a Grade 36% Never Failed a Grade 64%
Figure 3 illustrates the performance of the young offenders at school.Almost two thirds (32 or 64%) had not failed a grade, while just over a third (18 or 36%) had failed a grade.Contrary to expectations, the young sex offenders did not report being under-achievers at school and learning difficulties did not appear to be prevalent.However, this phenomenon may have been missed as schools in underprivileged areas frequently do not have the resources to identify and remediate difficulties experienced by scholars.
Although other authors (e.g.Concepcion, 2004;Graves et al., 1996) stipulate that children who sexually abuse others may have learning disabilities, this aspect was difficult to ascertain in the study because of the challenges in the schooling system.

THE CHILD AND HIS FAMILY
Family characteristics have been suggested as probable antecedents to adolescent sexual offending (Adams, Trachtenberg & Fisher as cited in Graves et al., 1996;Concepcion, 2004).For this reason the young sex offender was therefore viewed in relationship to parents, siblings and other family members as they can exert a significant influence on his development.

Currently 48% (43 fathers)
Figure 4 demonstrated that the great majority (43 or 86%) of the fathers were present when the child was born, and most (40 or 80%) continued to be present during the first five years of the child's life.However, this presence decreased, with only 24 or 48% of the fathers reported to be present in the life of the child at the time of the research, which can be a potentially significant factor in the developmental behaviour of children.
It is important to note that the lack of contact after the age of five could constitute paternal absence, which is considered to be a risk factor in adolescent sex offending, especially during the formation of a masculine identity after the age of five years.In view of the boys in this sample having little contact with their biological fathers during their childhood, pre-puberty and adolescence, it is likely that their masculine identity would have been largely formed by male role models in the community and the media particularly, and in areas where crime is valorised.
Furthermore, the literature asserts that the majority of youthful sexual offenders come from singleparent families, with the most prominent being female-headed households (Becker, Cunningham-Rathner & Kaplan, 1986; Loeber & Dishion as cited in Graves et al., 1996;Nanjundappa, De Rios, Mio & Verleur, 1987;Tingle, Barnard, Robbins, Newman & Hutchinson, 1986;Wedge, Boswell & Dissel, 2000).This finding was borne out only partially in the study, where more than half of the mothers (28 or 56%) consulted were de facto heads of their households.However, 44% (22) of the children came from nuclear families or those in which the mother was living with a partner.The relationship between the mothers and respondents was reported to be generally positive, and as evidenced by the fact that 88% of the participants turned to their mothers for support.
Table 1 illustrates that 42% of the mothers had a minimum of one and a maximum of nine other family members living with them, apart from their children and partners.This finding may suggest some form of overcrowding, as the participants lived in two-bedroom houses.
It was evident that very little quality time was spent on fun activities with family members.For example, 37 or 74% of children said they hardly ever played together with siblings and only half (27 or 54%) shared spiritual closeness by engaging in prayer together.In general, engagement in family activities seemed to be limited or absent.The vast majority (47 or 94%) of the children did not learn about sex from their families, but 46 or 92% described their acquisition of sexual information from television.The participants did, however, indicate a preference for learning about sex from their parents.The families did not appear to be dysfunctional.Only a third of the respondents (16 or 32%) described conflict between their parents and there was little evidence of substance abuse (alcohol or drugs) by the parents.Fear of disclosure of substance abuse may be because of other factors such as fear of being reported to "welfare" and their grants being terminated, as well as the need to furnish socially desirable responses, which is acknowledged as a possible limitation of the study.The respondents all came from poor families, where 72% supplemented their income with social grants with 58% receiving child support grants.A substantial number of the mothers were employed either full-time or part-time (62%).However, just over a third were employed in total.Fathers also contributed to family income (42%) and members of the family such as grandparents and other family members also contributed to household income.Although young sexual offenders come from all classes, these findings concur with that of other researchers who found that they come more frequently from lower socio-economic classes (Booysens, 2003).
High-density living arrangements in townships and inner city areas which are usually heavily populated giving rise to overcrowding and the blurring of boundaries between adults and children, with the latter inadvertently observing sexual intercourse between adults.Limited space and exposure to domestic violence, neighbourhood violence, gang sub-cultures, substance abuse and alcohol abuse have been cited by various authors as placing children and youths at risk of developing anti-social behaviour (Lee, Jackson, Pattison & Ward, 2002).This lack of privacy often leads to tensions between cohabitants and children, for whom crime may be an inevitable outcome (Anderson & Wild, 2007;Maree & Prinsloo, 2002).

RISK FACTORS
The early childhood experience of boys who later become male sex offenders has been widely regarded as pivotal by clinicians and researchers seeking to explain such behaviour (Concepcion, 2004;Smallbone & Dadds, 1998).This relationship was asserted by Finkelhor as early as 1990, and more recently by Finkelhor, Ormrod, Turner and Holt (2009).
A range of contributing factors was identified according to Bronfenbrenner's (1979) ecological perspective.Micro-level factors that were most commonly identified were: single headed households (56%), having more than one child (42%), history of victimisation (72%), conflict between parents (32%) and lack of consistent paternal presence (which decreased from 86% at birth to 48%).Although traumatic experiences may have been evident in the histories of these young sex offenders, there did not appear to be a strong correlation with the victim-victimiser theory of sexual abuse.Only 4% of the boys had been sexually abused in the previous year; however, it is known whether sexual abuse occurred prior to that date.In hindsight, it seems that this question should have been included in the questionnaire and its omission constitutes a limitation of the study.
The meso-level risk factors were schooling, lack of resources and support systems.There appeared to be few problems in the school environment, although a number of children reported behaviour such as name-calling (68%) and bullying (42%) from peers.However, many of the boys (47 or 94%) reported having friends at school and their accounts indicated that they enjoyed playing and socialising with them.As many of the children lived in the townships (74%) and in the inner city (14%), recreational opportunities were limited.
The exo-level risk factors were exposure to sexually explicit media (92%), highly sexualised environments and violent neighbourhoods.In violent communities where aggression is displayed as a means of resolving conflict, such behaviour may be normalised and become internalised in the behavioural repertoire of the child.
Macro-level risk factors were unemployment, limited financial resources and poverty.Although these factors were not specifically discussed, they relate to the legacy of apartheid inequalities and geographical locations in which poor families continue to live in communities where there is a lack of housing, infrastructure, safety and security.These communities are characterised by poverty, unemployment and overcrowding (associated with witnessing of sexual activity), partially explaining the increased incidence of child sexual offending among older children which emerged as further contributory factors.See Table 1 for data supporting overcrowding.

DISCUSSION AND CONCLUSIONS
These findings need to be considered in relation to limitations inherent in the research design and analysis of the study mentioned previously.Nevertheless, despite these limitations, important conclusions emerged from the study.Both male and female children were victims of abuse, with females being more frequently abused.The victims were mostly known to the abusers.The child and mother relationship was secure and the children relied on their mothers in times of distress, which is a positive factor.Moreover, the children were in mainstream education and seemed to enjoy positive interactions with their peers.The findings do not confirm the contributing factors identified by Bronfenbrenner's (1979) ecological perspective, and from this small sample it was not evident that children came from dysfunctional families.The findings suggest that different sets of influences coalesce or interact in the lives of the boys to create conditions under which they may offend.An understanding of how these influences are interconnected may aid our understanding of the nature of the phenomenon in South Africa today.These influences are: first, the maturational stage of development of the boys; second, the particular environmental influences; and third, the family.
In terms of maturation, the stage of development may predispose children to sexual curiosity and exploration in that pubescent children were found to be in the high-risk age cohort.
In terms of environmental influences, family situations were characterised by poverty, unemployment, overcrowding and lack of privacy, and home environments tended to be located in violent, criminogenic communities.The children's acquisition of information about sex and sexuality came mostly from exposure to television and the media rather than from their parents.
In terms of the family, the strong presence of the mother in the lives of the respondents is certainly a strength.However, the social dynamics of single parenthood and the absence of fathers in single-parent families were important factors that need to be taken into account.Just over half of the boys were growing up in single-parent families, where parents were either separated, divorced or no longer cohabiting, which in turn coincided with poverty or other stressors.Contrary to the findings of other studies (Turner & Peck, 2002), the fathers were found to be present in the lives of the respondents in the early years, i.e. in the first five years and to a lesser extent thereafter or at the time of the study.It seems that the current absence of the fathers in the family and in the daily lives of the respondents may have had some bearing on the psychosocial development of the boys.However, even though fathers may have been absent, the children lived in extended families with their mothers, which can be perceived as a protective factor.More research is needed on the situation and stressors of single parents and the role of fathers in the lives of these children.
In conclusion, these influences point to risk factors being associated with the maturational stage of the child, his family and the community, and the unique ways in which these interact in the lives of children abusing other children.Single-factor theories are not sufficient; hence a combination of expansive multi-factor theories may be more useful in understanding this complex phenomenon.This also points to the need for a more integrated and generalist practice framework that will allow the social worker to draw on specialist knowledge on sexual offending, strengths and empowerment perspectives to inform assessment and intervention on multiple levels involving the child, his family and the wider community (Kirst-Ashman & Hull, 2002).More specifically, strategies are needed to assist parents and teachers in responding appropriately to the maturational needs of children, especially boys in the pubescent phase, as a preventive intervention.There is also need for engagement with the media about what constitutes appropriate social programming for children, and for public education for parents and children about sex and sexuality are also important.

Age of young offenders Touching Genitals Mutual Touching of Genitals by Accomplice (including watching friends rape another) Attempted Rape
WEDGE, P., BOSWELL, G. & DISSEL, A. 2000.Violence in South Africa: key factors in the background of young serious offenders.Acta Criminologica, 13(2):31-38.The authors wish to acknowledge the input of Prof Jean Triegaardt of the Department of Social Work at the University of Johannesburg.Prof Eleanor Ross, Visiting Professor at the Centre for Social Development in Africa at the University of Johannesburg provided valuable editorial advice on the article.Dr Shaheda Omar, Deputy Director at the Teddy Bear Clinic for Abused Children, Transvaal Memorial Institute, Parktown, Johannesburg; Prof Leila Patel, Social Development, the Centre for Social Development in Africa, University of Johannesburg, Auckland Park, Johannesburg, South Africa. ACKNOWLEDGEMENT: