The frequency and predictors of poly-victimisation of South African children and the role of schools in its prevention

Abstract

Violence has become a characteristic feature of South African society, with women and children often bearing the brunt of this. Contemporary research suggests that the key to stemming the tide of child victimisation is understanding the complete inventory of victimisations that may co-occur during childhood. There is growing recognition that children in South Africa typically experience abuse in the context of other forms of maltreatment and victimisation. This article draws on the empirical data collected for a national prevalence and incidence study on child sexual abuse and maltreatment in South Africa and draws attention to the frequency of poly-victimisation amongst South African children and highlights why some children experience multiple co-occurring forms of victimisations while others do not. Understanding the complete victim profile of young children, and how the different forms of victimisation they experience intersect, is critical to ensuring that the most vulnerable South Africans are provided with the extensive and targeted interventions required to break free from their heightened vulnerability to victimisation.

Keywords:

• Poly-victimisation
• children
• South Africa
• schools
• violence prevention

Introduction

Violence has become a characteristic feature of South African society, with women and children often bearing the brunt of this. Contemporary research suggests that the key to stemming the tide of child victimisation is not focusing on individual types of child victimisation, but rather considering the complete inventory of victimisations that may co-occur during childhood. In so doing, the multiple underlying causes of child victimisation will be addressed, rather than merely the symptoms, and more appropriate interventions can be developed.

Nature and extent of child and youth victimisation

South Africa’s first national youth victimisation study found that for a large proportion of 12–22 year olds (41.4%), violence and crime was a common occurrence; with much of this violence occurring within their homes (21.8%) (Leoschut & Burton, 2006). Loeber and Stouthamer-Loeber (1986) state that ‘family factors never operate in a vacuum but take place against a backdrop of other influences’ which serves to amplify the effects of family variables on child victimisation. The frequency with which children are exposed to violence in their communities became apparent when 50.1% of participants in a national study reported having witnessed someone in their community using threat or force to physically harm another person (Leoschut, 2009).

Child abuse is also widespread in South Africa. According to a recent study, 35.4% of children are sexually victimised before the age of 17 (Artz et al., 2016). These figures were consistent with an earlier study that showed that 38% of women and 17% of men had been sexually victimised before the age of 18 (Jewkes, Dunkle, Nduna, Jama, & Puren, 2010). Although underreported, the official police statistics also attest to the widespread occurrence of child sexual victimisation; with between 18,000 and 20,000 child sexual abuse cases reported each year (Artz et al., 2016).

Physical abuse is pervasive, with 20.8% of children reporting physical abuse by a parent or caregiver (Artz et al., 2016). Similarly, Dawes, Kafaar, Richter, and De Sas Kropiwnicki (2005) found that 58% of parents reported having ever smacked their children, at times, using an object to do so (33%). Neglect (15%) (Seedat, Van Niekerk, Jewkes, Suffla, & Ratele, 2009) and emotional abuse (16.1%) are also endemic (Artz et al., 2016).

The identification of violence as one of the leading causes of child mortality in South Africa clearly demonstrates the vulnerability of children. A study analysing data from mortuaries in South Africa, found that in 2009, there were 1018 child murders (Mathews, Abrahams, Jewkes, Martin, & Lombard, 2013). A total of 44.5% of these murders occurred in the context of child abuse, and in 10% of these cases sexual abuse was suspected (Mathews et al., 2013).

Research studies also point toward the school environment as a common site for victimisation (Stevens, Wyngaard, & Van Niekerk, 2001) with school violence rates ranging from 15.1 to 22.2% nationally (Burton & Leoschut, 2013).

The child victimisation literature in South Africa have largely been concerned with documenting the magnitude of individual forms of victimisation (Ellonen & Salmi, 2011; Holt, Finkelhor, & Kantor, 2007) exploring the risk factors, as well as, identifying the deleterious outcomes associated with these specific forms of victimisation (Finkelhor, Ormrod, Turner, & Hamby, 2005a; Finkelhor, Shattuck, Turner, Ormrod, & Hamby, 2011; Ford, Wasser, & Connor, 2011; Voith, Gromoske, & Holmes, 2014). While these efforts have been lorded as useful for providing practitioners with an in-depth understanding of individual types of victimisation, many authors have argued that it ignores the greater spectrum of adversities that children are susceptible to during childhood (Hamby & Finkelhor cited in Finkelhor et al., 2005b; Price-Robertson, Higgins, & Vassallo, 2013) and underestimates the full burden of child victimisation (Holt et al., 2007).

Poly-victimisation in South Africa

There is growing recognition that children in South Africa often experience abuse in the context of other forms of victimisation. According to Finkelhor, Ormrod, Turner, and Holt (2009), victimisation types are often interconnected, and any one type of victimisation, breeds not only susceptibility to other forms of victimisation (Cole, Maxwell, & Chipaca, 2014), but also vulnerability across contexts. Thus, victimisation is rarely a one-off event, but instead a condition in which children may become ensnared (Finkelhor, Ormrod, & Turner, 2007) for years on end (Widom, Czaja, & Dutton, 2008).

In the South African context, risk factors for victimisation include, family and household composition, frequent exposure to violence in the home, living in a disorganised community, harsh and inconsistent parenting, poor parental supervision and monitoring, parental absence due to prolonged illness or hospitalisation, parental substance misuse, and child disability (Artz et al., 2016; Hawkins et al., 2000).

There is evidence to suggest that experiences of victimisation tend to cumulate for certain high-risk individuals or certain high-risk environments (Finkelhor et al., 2011). Finkelhor and colleagues use the term poly-victim to refer to those children who experience high levels of multiple forms of victimisation (Finkelhor et al., 2011). Not only do poly-victims experience high numbers of victimisations, but they also experience victimisation in different contexts (Finkelhor et al., 2009; Simmons, Wijma, & Swahnberg, 2015). For this reason, poly-victims have a higher likelihood of maladjustment given the more severe symptomatology associated with the co-occurrence of victimisation (Cyr, Clement, & Chamberland, 2014; Voith et al., 2014). Ellonen and Salmi (2011) found that poly-victimisation is associated with an increased level of psycho-social problems compared to children with no victimisation experience, as well as, those with fewer experiences.

Holt et al. (2007) found that poly-victims often behave aggressively towards others. Schools in South Africa already face a myriad of challenges including poor infrastructure, a lack of resources, and various safety-related concerns (Burton & Leoschut, 2013; Stevens et al., 2001). In addition, schools are tasked with having to teach large numbers of learners who experience a range of emotional and behavioural difficulties as a result of their extensive victimisation experiences (Burton & Leoschut, 2013; Holt et al., 2007). For this reason, schools, have an important preventative role to play.

Method

Aims and objectives

This article is based on the empirical data collected for a national prevalence and incidence study on child sexual abuse and maltreatment (Artz et al., 2016). The overall goal of this retrospective study was to provide an accurate estimation of the annual incidence and lifetime prevalence of child sexual abuse and to locate this abuse within the context of other forms of victimisation.

Participants

Data for this study was collected using a population-based survey that was targeted at households as well as schools. The sample frame for the population survey was based on the 2001 Census data of South Africa, adjusted according to the Statistics South Africa’s 2011 census population numbers and other district council estimates. A multi-stage stratified sample was used to achieve a nationally representative sample of 15–17 year olds at a household level. A school survey was also conducted at high schools that were clustered around the enumerator areas identified in the household survey; the analysis for this article draws on the household survey data only.

In the household survey, 5635 participants were recruited nationally, who described themselves as Black (80.4%), Coloured (9.5%), White (8.0%) and Indian (2.0%). Young people from KwaZulu-Natal (21.6%), Gauteng (18.6%), the Eastern Cape (14%) and Limpopo (12.5%) provinces comprised the greater part of the sample, followed by those from the Western Cape (10.0%), Mpumalanga (8.5%), North West (6.9%), Free State (5.3%) and Northern Cape (2.6%). There were more male (55.4%) than female participants (44.6%). Most of the participants were 16 years of age (36.5%), followed by those who were 15 (34.2%) and 17 (29.4%) years old.

Measures

The study used a combination of an interviewer-administered and self-administered questionnaire to collect the survey data. The questionnaires were designed to examine; the prevalence and incidence of child sexual abuse and maltreatment, the consequences of abuse, as well as the risk and protective factors associated with abuse.

In designing the questionnaires, the study drew on two instruments, namely the Juvenile Victimization Questionnaire (Finkelhor, Ormrod, Turner, & Hamby, 2005a) and the Trauma Symptom Checklist for Children (Briere et al., 2001). Minor revisions and additions were made to these instruments so that it was more appropriate to the South African context.

The result was a 38-item screener measure that explored a broad range of victimisations across several modules – conventional crime, cyber-bullying and online victimisation, child maltreatment, sexual victimisation, and witnessing and indirect victimisation. The questionnaires were comprehensively pilot-tested using cognitive interview techniques (Carbone, Campbell, & Honess-Morreale, 2002; Miller, Mont, Maitland, Altman, & Madans, 2011).

Procedure

The main questionnaire was administered by a trained enumerator, where after the participant was invited to also self-complete a short one-page version of the questionnaire; the analysis for this article was done using the data generated from the interviewer-administered questionnaire only. Interviews were conducted after informed parental consent and child assent was obtained. A number of measures were put in place to protect the privacy and confidentiality of the participants given the sensitive nature of the study. All questionnaires were stored in locked filing cabinets until it was captured electronically, all identifying information recorded on the first sheet of the questionnaire was removed and stored separately – ensuring that no information about child maltreatment and abuse could be traced back to any participant by name, and access to the password-protected data file was limited to core members of the research team only.

Participants were provided with the details for counselling services in their area in the event that they required any support following their interview. The research team was also legally obligated to report any cases of abuse that were disclosed during the course of the interviews, that had not previously been reported to a child protection agency. Prior to the interviews, the participants were informed of this in a manner they could understand. Enumerators were trained to flag reportable cases of sexual abuse, physical abuse and child neglect that were later reported to a child protection agency servicing the geographic location in which the participants lived. This reporting procedure as well as the research study more broadly was approved by both the Human Research Ethics Committee of the Faculty of Health Sciences, and the Research Ethics Committee of the Faculty of Humanities, of the University of Cape Town.

Limitations

The data for this article stems from a cross-sectional study. Since information on the different variables were all collected at the same point in time, it does not allow for the exploration of cause-and-effect relationships between poly-victimisation and the risk factors explored in this article. Instead, it merely points toward an association between the two.

Data analysis

This article follows the analysis of Finkelhor and his colleagues, to a large degree, by examining the clustering of different types of victimisation among a sample of 15–17 year olds. While this article does not explore the traumatic effects of poly-victimisation on children, it includes an examination of the factors associated with an increased risk for poly-victimisation.

Here, the term poly-victimisation referred to a situation where a participant had experienced several victimisations across different contexts ever in their lives. Calculating poly-victimisation using lifetime prevalence rather than last-year prevalence, provides a more holistic picture of the victimisation profile of young children (Finkelhor, Ormrod, & Turner, 2009) rather than focusing only on victimisation occurring in a one-year time period.

Descriptive analysis

Poly-victimisation was measured by items that asked participants whether they had experienced any of a number of victimisations. Items were scored 1 for Yes and 0 for No. A composite variable was created by summing all 38 items to create a Lifetime Poly-victimisation variable. The potential range of scores for poly-victimisation was thus from 0 to 38. Scores for the Lifetime Poly-victimisation variable ranged from 0 to 28, with a median of 3, a mean of 4.16 and a standard deviation of 4.0 events.

Inferential analysis

Predictor variables were created for the following constructs: sleeping density (i.e. the number of people with whom the participant shares a bedroom), accessing a social grant, which parent/s resided with the participant, parental absence due to physical ill-health, parental mental health, parental substance misuse, child substance misuse, parental incarceration, participant disability, child sexual risk behaviours, and whether the child lived in an urban or rural area.

Five binary logistic regression analyses were run with the same set of independent variables. The first regression compared no poly-victimisation to low poly-victimisation, the second compared low poly-victimisation to high poly-victimisation, the third compared high poly-victimisation to very high poly-victimisation, the fourth compared no poly-victimisation to any form of poly-victimisation and the final binary logistic regression compared no poly-victimisation to very high poly-victimisation.

Results

In order to run a binary logistic regression, the lifetime poly-victimisation variable was changed from an interval scale to an ordinal scale. Four categories were created using the mean and SD: (1) No or 1 victimisation, i.e. no poly-victimisation; (2) Low poly-victimisation (below the mean, i.e. 2–4 events); (3) High poly-victimisation (between the mean and 1 SD above the mean, i.e. 5–8 events); and (4) Very high poly-victimisation (more than 1 SD above the mean, i.e. 9–28 events). The frequencies for these categories are listed in Table 1.

Table 1. Frequency of poly-victimisation.

Poly-victimisation category	Frequency	Per cent
No (0–1 events)	2033	36.1
Low (2–4 events)	1605	28.5
High (5–8 events)	1288	22.9
Very high (>8 events)	709	12.6
Total	5635	100.0

Table 2, lists the frequency of the different forms of victimisation. The forms of victimisation are arranged from the most frequently to the least frequently occurring. The most frequent form of victimisation was theft, with 2133 participants (37.9%) reporting that they had had an item stolen from them at least once in their lifetime. While it is comforting to note that only 37 participants (.7%) reported that they had been sexually abused by an adult known to them, it is concerning to note that 357 participants (6.3%) report having had a sexual experience with an adult.

Table 2. Frequency of types of victimisation.

Type of victimisation (total sample size N = 5635)	Poly-victimisation category
	No	Low	High	Very high
Had something stolen (N = 2133, 37.85%)	96	691	818	528
Seen anyone attacked without a weapon (N = 1859, 32.99%)	43	504	750	562
Seen anyone attacked with a weapon (N = 1774, 31.48%)	28	495	712	539
Had something forcefully removed (N = 1318, 23.39%)	46	375	511	386
Item stolen from home (N = 1109, 19.68%)	53	286	446	324
Threaten to hurt (N = 1072, 19.02%)	26	240	406	400
Bullied (N = 965, 17.13%)	10	179	396	380
Physical victimisation (N = 962, 17.07%)	43	228	356	335
Attacked without an object (N = 939, 16.67%)	9	146	404	380
Attacked with an object (N = 831, 14.75%)	17	174	336	304
Heard shots, bombs or riots (N = 818, 14.52%)	14	146	349	309
Malicious damage to property (N = 698, 12.39%)	6	158	255	279
Emotional abuse (N = 688, 12.21%)	16	130	253	289
Seen parent hurt siblings (N = 633, 11.23%)	17	123	211	282
Escape attack (N = 537, 9.53%)	2	68	197	270
Parent threatened to hurt other parent (N = 524, 9.30%)	9	76	170	269
Murder of friend/neighbour/family member (N = 436, 7.74%)	16	65	177	178
Member of household assaulted other member (N = 416, 7.38%)	24	112	127	153
Parent pushed other parent (N = 383, 6.80%)	1	25	108	249
Sexual experience with an adult (N = 357, 6.34%)	33	89	122	113
Neglect due to physical living conditions (N = 351, 6.23%)	12	76	133	130
Parent hit or slapped other parent (N = 350, 6.21%)	4	23	99	224
Hit or attacked by an adult (N = 332, 5.89%)	3	46	98	185
Hit on purpose other than mentioned (N = 266, 4.72%)	7	28	69	162
Parent damaged other parent’s property (N = 262, 4.65%)	1	13	62	186
Neglect due to fear of parents’ visitors (N = 259, 4.60%)	5	51	77	126
Parent kicked, choked or beat other parent (N = 217, 3.85%)	1	9	43	164
Sexual exposure abuse (N = 192, 3.41%)	4	29	51	108
Attacked due to prejudice (N = 131, 2.33%)	2	13	45	71
Forced sexual intercourse (actual or attempted) (N = 128, 2.27%)	7	23	27	71
Written or verbal sexual harassment (N = 121, 2.15%)	2	15	39	65
Sexual abuse by child or teen (N = 114, 2.02%)	2	22	27	63
Sexual abuse by known adult (N = 110, 1.95%)	1	24	30	55
Neglect due to alcohol or drugs (N = 100, 1.78%)	3	14	22	61
Attempted kidnapping (N = 86, 1.53%)	2	8	37	39
Neglect due to abandonment (N = 80, 1.42%)	1	8	15	56
Neglect of physical cleanliness (N = 75, 1.33%)	3	9	17	46
Sexual abuse by unknown adult (N = 37, .66%)	1	9	8	19

Binary logistic regression analyses predict membership of one of the categories of the binary dependent variable. Table 3 below lists the number of cases per analysis, the Wald statistic, statistical significance, β and Exp β as well as model significance values. The category of membership being predicted is denoted by an asterisk.

Table 3. Binary logistic regression analyses.

Lifetime poly-victimisation	N	Wald	Sig	B	Exp (B)	Model sig
No vs. Low poly-victimisation^*	347	20.393	.000006	.5	1.649	.000004
Low vs. High poly-victimisation^*	524	15.984	.000064	.355	1.426	.001
High vs. Very high poly-victimisation^*	532	13.152	.000287	−.138	.727	.000157
No vs. Any poly-victimisation^*	879	338.363	<.000001	1.742	5.71	<.000001
No vs. Very high poly-victimisation^*	355	23.787	.000001	.536	1.71	<.000001

^* Category of membership being predicted.

Table 4 lists the significant predictors of poly-victimisation for all five regression analyses, followed by detailed explanations of the predictors of poly-victimisation.

Table 4. Significant predictors for poly-victimisation.

Predictors	No poly-victimisation vs. Low poly-victimisation			Low poly-victimisation vs. High poly-victimisation			High poly-victimisation vs. Very high poly-victimisation			No poly-victimisation vs. Any poly-victimisation			No poly-victimisation vs. Very high poly-victimisation
	Odds ratio	95% CI for odds ratio		Odds ratio	95% CI for odds ratio		Odds ratio	95% CI for odds ratio		Odds ratio	95% CI for odds ratio		Odds ratio	95% CI for odds ratio
		Lower	Upper		Lower	Upper		Lower	Upper		Lower	Upper		Lower	Upper
Parental substance misuse	11.852^*	1.325	106.01	3.454^***	1.790	6.665	1.796^*	1.128	2.861	24.392^***	3.109	191.345	47.611^***	4.729	479.374
Parental absence due to physical ill-health				2.303^***	1.548	3.428	1.527^*	1.050	2.220	2.987^***	1.830	4.876	4.393^***	2.311	8.354
Child substance misuse	2.008^**	1.222	3.298				1.739^**	1.170	2.586	2.823^***	1.844	4.323	4.845^***	2.646	8.871
Child sexual risk behaviour	1.98^*	1.176	3.322							2.074^**	1.323	3.253	2.393^**	1.3	4.405
Parents residing with child										2.260^*	1.231	4.148
Urban/rural residence													2.545^*	1.144	5.65

^* p < .05

^** p < .01

^*** p < .001.

Of the 5635 young people interviewed, 2033 had experienced no or only one type of victimisation ever in their lives, while 3602 – the majority of the sample – had experienced two or more different forms of victimisations ever in their lives. Furthermore, 35.4% had experienced five or more types of different victimisations by the time they turned 17 years of age.

Parental substance misuse

Parental substance misuse consistently significantly predicted higher poly-victimisation in all five regression analyses. Participants whose parents abused substances were 11.852 (95% CI; 1.325–106.010) times more likely to have experienced low poly-victimisation (between one and four victimisation events) compared to no poly-victimisation (p < .05), 3.454 (95% CI; 1.790–6.665) times more likely to experience high poly-victimisation (between five and eight victimisation events) compared to low poly-victimisation (p < .001), and 1.796 (95% CI; 1.128–2.861) times more likely to experience very high poly-victimisation (more than eight victimisation events) compared to high poly-victimisation (p < .05) when compared to participants whose parents did not misuse substances.

When we compare no poly-victimisation to any poly-victimisation and very high poly-victimisation, the odds ratios increase substantially. Participants whose parents abused substances were 24.392 (95% CI; 3.109–191.345) times more likely to be members of the poly-victims category than the no poly-victims category (p > .001), and 47.611 (95% CI; 4.729–479.374) times more likely to be very high poly-victims than participants whose parents did not misuse substances (p < .001).

Parental absence due to illness

Parental absence due to illness significantly predicted poly-victimisation in all the regression analysis excluding the regression analysis that compared no poly-victimisation to low poly-victimisation. Children whose parents were absent for prolonged periods due to physical health problems were 2.303 (95% CI; 1.548–3.428) times more likely to experience high poly-victimisation compared to low poly-victimisation (p < .001), and 1.527 (95% CI; 1.050–2.220) times more likely to experience very high poly-victimisation compared to high poly-victimisation, than children whose parents had not been absent (p < .05).

When we compare no poly-victimisation to any and very high poly-victimisation, we see that children whose parents were absent due to ill-health were 2.987 (95% CI; 1.830–4.876) times more likely to experience any poly-victimisation compared to none (p < .001), and 4.394 (95% CI; 2.311–8.354) times more likely to experience very high poly-victimisation compared to none, than children whose parents had not been absent (p < .001).

Child substance misuse

Child substance misuse significantly predicted poly-victimisation in all the regression analyses except the regression analysis that compared low to high poly-victimisation. Children who abused substances were 2.008 times (95% CI; 1.222–3.298) more likely to experience low poly-victimisation (p < .01), and 1.739 (95% CI; 1.170–2.586) times more likely to experience very high poly-victimisation (>8 events) than children who did not misuse substances (p < .01).

When we compare no poly-victimisation to any poly-victimisation and very high poly-victimisation, we see that children who misuse substances are 2.823 (95% CI; 1.844–4.323) times more likely to experience any poly-victimisation (p < .001) and 4.845 (95% CI; 2.646–8.871) times more likely to experience very high poly-victimisation (>8 events) than children who do not abuse substances (p < .001).

Child sexual risk behaviour

Children who engaged in sexual risk behaviour significantly predicted poly-victimisation in three of the five regression analyses. When predicting membership of the low poly-victimisation category, children who engaged in risky sexual behaviour were 1.98 (CI 95%; 1.176–3.322) times more likely to experience low poly-victimisation than children who did not engage in sexual risk behaviour (p < .05). Similarly, when predicting membership of the any poly-victimisation category, children who engaged in risky sexual behaviour were 2.074 (95% CI; 1.323–3.253) times more likely to experience poly-victimisation than children who did not engage in risky sexual behaviour (p < .01). Finally, when predicting membership of the very high poly-victimisation category (>8 events), children who engaged in risky sexual behaviour were 2.392 (95% CI; 1.3–4.405) times more likely to experience very high poly-victimisation than children who did not engage in risky sexual behaviour (p < .01).

Stays with parents

Which parent/s resided with the child was only significant when predicting membership of any poly-victimisation. Children who lived with one parent were 2.260 (95% CI; 1.231–4.148) times more likely to experience any poly-victimisation than children who lived with both parents (p < .05).

Urban/rural

Whether the children resided in an urban or rural area was only significant when predicting membership of the very high poly-victimisation category (>8 events) compared to no poly-victimisation. Children in urban areas were 2.545 (95% CI; 1.144–5.65) times more likely to experience very high poly-victimisation than children from rural areas.

Discussion and conclusion

Eaton, Flisher, and Aarø (2003) argue that there are three levels at which risk behaviours are influenced: the personal; the proximal (the physical environment and interpersonal relationships); and the distal level (cultural and structural factors). It would seem that experiencing poly-victimisation is influenced by factors at all levels. At the personal level (sexual risk behaviour and substance misuse), in the proximal context (parental substance misuse, parental absence due to physical health reasons, and the number of parents the child resides with) as well as in the distal context (urban vs. rural).

This article draws attention to the frequency of poly-victimisation amongst South African children and highlights why some children experience multiple co-occurring forms of victimisations while others do not. Understanding the complete victim profile of young children, and how the different forms of victimisation they experience intersect, is critical to ensuring that the most vulnerable South Africans are provided with the extensive and targeted interventions required to break free from their heightened vulnerability to victimisation. This is essential, given that poly-victims are likely to remain highly victimised as they get older (Finkelhor et al., 2007; Finkelhor et al., 2011).

Poly-victims are significantly more likely to be urban children, living with one rather than both their biological parents, whose parents abuse substances, and are absent from the home due to prolonged illness and are children who themselves use substances and engage in risky sexual behaviours. For each of the victimisation types, participants were asked whether they had been under the influence of alcohol and/or drugs at the time of the incident. In the vast majority of cases, the participants were found to have been sober at the time of the incident. Although, this research can merely point toward an association between these descriptors and poly-victimisation and cannot make any causal claims about these variables, it may suggest that participant substance abuse specifically, may have been a consequence of the poly-victimisation.

Given the persistent nature of poly-victimisation across the life-span of children, early intervention is key. Identifying children most at risk of poly-victimisation and intervening early on may buffer children from experiencing continued victimisation later on in life (Finkelhor, Ormrod, & Turner, 2009). Schools are in an ideal position to do this, given that children spend a large amount of time there.

The identification of schools as an important site for violence prevention is further underscored by Ozer and Weinstein (2004) cited in Ozer (2005) who argue that there are generally two types of protective factors for adolescents. The first, are supportive relationships with significant others, while the second, is growing up in physically safe social environments; of which the home and the school are most important (Ozer & Weinstein, 2004 cited in Ozer, 2005).

There are also other reasons why schools are ideal sites for preventing child victimisation. Firstly, schools provide a social context that comes along with established infrastructure and resources that could support violence prevention initiatives (Stevens et al., 2001). When school personnel can effectively identify high risk learners, they can ensure that the available resources are targeted at those children who are most prone to multiple victimisations. Secondly, schools have a captive audience and can implement carefully targeted interventions for a sustained period of time. This will ensure that poly-victims who are attending schools, can access continued support services during the years that they’ll be attending school. Thirdly, schools are attended by children and youth who are at critical developmental stages in their lives. Carefully targeted interventions can positively influence their developmental trajectories (Ozer, 2005).

Although schools provide an important entry-point for prevention (Holt et al., 2007), other interventions that fall outside the ambit of schools are required to address poly-victimisation including substance abuse treatment and prevention initiatives, and parenting support programmes (Finkelhor et al., 2011).

Funding

This work was supported by Know Violence in Childhood.

Disclosure statement

No potential conflict of interest was reported by the authors.