Adolescents’ Reports of Physical Violence by Peers in Residential Care Settings

Abstract

Physical victimization by peers was examined among 1,324 Jewish and Arab adolescents, aged 11 to 19, residing in 32 residential care settings (RCS) for children at-risk in Israel. Hierarchical Linear Modeling (HLM) was used to examine the relationships between physical victimization and adolescents’ characteristics (age, gender, self-efficacy, adjustment difficulties, maltreatment by staff, and perceived social climate) as well as institution-level characteristics (care setting type, size, structure, and ethnic affiliation). For this study, we define physical violence as being grabbed, shoved, kicked, punched, hit with a hand, or hit with an object. Over 50% (56%) of the adolescents surveyed reported having experienced at least one form of physical violence by peers. Boys and younger adolescents were more likely to be victimized than girls and older adolescents. The results show that adolescents with adjustment difficulties or low social self-efficacy, and adolescents who perceive an institution’s staff as strict and/or had experienced maltreatment by staff, are vulnerable groups for peer victimization. Lower levels of victimization were found in RCS with a familial element than in traditional group settings. Institutions with high concentrations of young people with adjustment difficulties and violent staff behaviors had higher levels of violence among residents. Applying an ecological perspective to an investigation of peer victimization in RCS enables the identification of risk factors at adolescent and institution levels. This type of examination has implications for child welfare practice and policy that can help in the development of prevention and intervention methods designed to tackle the involvement in violence of youth in care.

Keywords

residential care victimization by peers adjustment difficulties climate staff maltreatment

Children and youth in out-of-home care are considered to be a vulnerable group, subject to many risks (Kendrick, 2011). Many of these children experience abuse and neglect before they are placed in care (Ellonen & Pösö, 2011), so there is a particular need to protect them from further victimization while in care. The examination of their victimization is critical in helping child welfare professionals develop policies and programs to protect these young people.

Little research has been conducted on peer-to-peer violence among young people in care (Barter, 2008; Monks et al., 2009). The few studies that examine this issue are based mainly on convenience, small-scale samples (e.g., Barter, Renold, Berridge, & Cawson, 2004; Gibbs & Sinclair, 2000), and on adult reports (e.g., Attar-Schwartz, 2008; Freundlich, Avery, & Padgett, 2007). These studies usually draw a worrying picture, indicating that adolescents living in residential care settings (RCS) are highly involved in violence (Barter et al., 2004; Lees, Morris, & Wheatley, 1994; MacLeod, 1999). For instance, in a study conducted by Barter et al. (2004) among 71 young people in 14 children’s homes in England, over half of the young people (53.5%) reported they had been physically victimized during their stay in residential homes. A similar trend was found in another British study (Gibbs, & Sinclair, 2000), where a considerable share of the young people (43.9%) had been bullied in their current RCS. The latter study, however, used a general subjective definition of bullying, and did not differentiate between various types of bullying.

These numbers are disturbing because of the possible detrimental short- and long-term consequences of peer-to-peer violence among young people residing in care institutions (e.g., Barter et al., 2004; Gibbs & Sinclair, 1999; Sinclair & Gibbs, 1998). A follow-up study of 141 young people in children’s homes in the United Kingdom found, for example, that residents who reported being bullied while in care were more distressed and less able to adjust to various aspects of life in the care setting and after (Gibbs & Sinclair, 1999).

Most children in Israel’s care system are placed in RCS that belong to the welfare system and are specifically designed for at-risk placements. Placement is made by an interdisciplinary decision committee appointed by the Israeli Welfare Ministry. The committee considers the needs and family background of the child before placing him or her in a care setting.

There are four types of RCS: rehabilitative, therapeutic, post-hospitalization, and educational settings (Dolev, Benbenishty, & Timar, 2001). Rehabilitative care is provided for about 40% of young people in welfare RCS. These settings serve young people with satisfactory developmental potential who have been removed from their homes because of their parents’ inability to cope with their emotional and educational needs. Therapeutic care is provided for approximately 30% of young people placed in welfare RCS. It is designed to treat young people with extreme family problems, personal needs, and adjustment difficulties. Post-hospitalization care and educational care were not considered in this study. Post-hospitalization placements serve youth who suffer from severe psychiatric problems; educational settings serve youth from underprivileged backgrounds whose parents choose this option because they believe it will provide them with education of good quality (see Attar-Schwartz, 2009).

Little is known about peer-to-peer violence among young people in RCS in Israel. There is a growing body of research examining violence among large-scale representative samples of Israeli school students (Benbenishty & Astor, 2005), as well as violence among young people in correctional facilities (e.g., Davidson-Arad & Golan, 2007). These studies reveal a high rate of physical, emotional, and sexual violence among young people in those settings.

Previous studies addressing violence and bullying among youth in RCS have mainly focused on the adolescents’ characteristics and pre-care experiences to explain violence in institutions (e.g., Herrera & Stuewig, 2011). These studies tend to overlook young people’s experiences of their care settings, such as their evaluation of the social climate in the institution, as explanatory aspects of victimization. They also frequently overlook institution-level factors that might be associated with victimization.

Studies examining school violence have incorporated student-level and school-level factors to explain school violence (e.g., Gottfredson, Gottfredson, Payne, & Gottfredson, 2005; Khoury-Kassabri, Astor, & Benbenishty, 2009). This type of multi-level investigation is rare in RCS (for an exception, see Attar-Schwartz, 2011, regarding staff maltreatment in RCS). In the current study, we adopt Bronfenbrenner’s (1979) ecological perspective, which offers a framework for integrating the various factors associated with youth violence. It describes peer-to-peer violence as an interplay among several relevant subsystems (e.g., child, family, institution, and culture). Along these lines, Astor, Pitner, and Duncan (1996) argued that child development can be understood only within the total context in which the child lives and when development is considered on a variety of levels and in a multitude of natural settings. They use this framework to show how victimization and perpetration in school can result from the interactive effects of subsystems such as students’ families, communities, and the larger societal context.

Guided by this ecological multi-level perspective, the current study examines physical victimization by peers in RCS by both child-level and institution-level correlates. The following review describes the findings of existing research on both levels’ correlates included in the current study.

Previous studies have shown that boys in substitute care tend to exhibit more externalizing problems, such as aggression, delinquency, acting-out, and anti-social and violent behaviors, than girls (e.g., Attar-Schwartz, 2008; Glisson, Hemmelgarn, & Post, 2002; Schiff & Benbenishty, 2006). For example, Attar-Schwartz (2008) found that, according to social workers’ reports of 4,420 children and youth in residential care in Israel, boys showed significantly higher levels of aggressive behavior than girls. A similar trend was reported by Barter (2008), who showed that more boys than girls were involved in high-level physical violence. In addition to the direct gender effects on peer violence, it will be interesting to explore whether the effects of risk factors at the individual level (e.g., self-efficacy, adjustment difficulties) or the contextual level (e.g., Arab vs. Jewish institutions) are similar or different for boys and girls. Some indications for the moderating role gender plays in the relationship between risk factors and violence and delinquency have been reported in previous studies of school students, but not among children in RCS. For instance, higher levels of social control and commitment to school have been found to be associated with lower levels of delinquency among males but not among females (Ellonen, 2008; Özbay & Özcan, 2008). Furthermore, Khoury-Kassabri et al. (2009) found that the differences between males and females in reports of violent acts toward peers and teachers are greater in Arab schools than in Jewish schools.

Studies have also shown that, compared with older children, younger children in care are at greater risk of being victimized by peers (e.g., Gibbs & Sinclair, 2000). These findings are in keeping with studies that address physical victimization by peers at school (e.g., Khoury-Kassabri, 2011).

Risk factors for victimization might also result from behavioral, emotional, and psychological characteristics of the child. Previous studies have shown that children in substitute care are a high-risk group for psychological, psychiatric, educational, social, health, and behavioral problems (e.g., Attar-Schwartz, 2008, 2009; Heflinger, Simpkins, & Combs-Orme, 2000; Mitic & Rimer, 2002; Vinnerljung, Oman, & Gunnarson, 2005). These factors have been found to be associated with higher levels of victimization by peers (Smith, Talamelli, Cowie, Naylor, & Chauhan, 2004) and victimization by RCS staff (e.g., Attar-Schwartz, 2011). It seems, therefore, that children with more adjustment difficulties are more vulnerable to violent experiences.

Self-efficacy is an important contributing factor to victimization by peers in school settings and to the ability to cope with it (Singh & Bussey, 2009). The concept of self-efficacy was introduced by Bandura (1977, 1982) and was influenced by his Cognitive Social Learning Theory. It refers to the “beliefs in one’s capabilities to organize and execute the course of action required to produce given attainments” (Bandura, 1997, p. 3). Singh and Bussey (2009) found that self-efficacy helps individuals develop a belief in their ability to deal with threatening circumstances. A well-developed sense of self-efficacy can empower individuals by fostering a sense of control based on the belief that they are proactive agents who can shape their life circumstances.

Bandura, Pastorelli, Barbaranelli, and Caprara (1999) examined children’s social, emotional, and academic self-efficacy to assess their general level of self-efficacy. The current study focuses on social and emotional self-efficacy. Social self-efficacy pertains to children’s ability to deal with social challenges. Emotional self-efficacy pertains to their perceived ability to cope with negative emotions. Kokkinos and Kipritsi (2012) found that all aspects of self-efficacy were negatively associated with victimization in schools. However, to the best of our knowledge, none of these studies has explored the association between victimization among children in care, adjustment difficulties, and self-efficacy.

In addition to the child’s personal characteristics, children’s experiences and their evaluation of the environment in the RCS they reside in might also be associated with exposure to peer-to-peer violence. For example, Barter et al., (2004) found that levels of physical and sexual violence were lower in residential homes that encouraged positive relationships between staff and residents. In a different study, it was shown that children who related poorly to staff experienced higher levels of bullying in residential homes (Gibbs & Sinclair, 2000). The findings emerging from ongoing studies of violence in schools show similarly that supportive relationships with school staff are associated with lower levels of physical victimization of students by peers (Gottfredson et al., 2005; Khoury-Kassabri, 2011).

While positive relationships with staff seem to act as buffers to victimization, negative and abusive relationships with staff might be correlated with greater exposure to violence by peers. Adolescents who are maltreated by staff might be at greater risk of peer victimization, as their peers might interpret the staff’s behavior as a legitimate way to solve problems and conflicts. Support for this assumption is found in school violence studies where physical and emotional victimization by teachers was positively correlated with students’ physical, emotional, and indirect forms of victimization by peers (Khoury-Kassabri, 2011).

There is also evidence that child outcomes and levels of peer bullying vary according to the RCS factors (Barter et al., 2004), such as the structure of the RCS. In Israel, RCS may use different institutional structures to accommodate at-risk youth. Traditional group institutions, which are the most common RCS structure, care for large numbers of children who reside in small groups. Each group of children has a social worker, as well as social-educational carers in rotating shifts. There is also the familial setting in which a married couple with its biological children cares for a small number of children at risk (up to 10). All family units in the facility share the same administrative, cleaning, and general support services. Other types of settings include institutions that combine residential groups and family-like units in the same structure (hybrids) and other settings (Attar-Schwartz, 2009; Children and Youth Service, 2005).

As far as we are aware, the institution structure has not been examined as an explanatory variable of peer violence, but it has been studied with regard to other outcomes. There is some evidence, for example, that children’s psychosocial status varies across different institutional structures (e.g., Ajdukovic & Franz, 2005; Heflinger et al., 2000). For example, Mosek, Taus, and Shomodi (1997) described an improvement in the well-being of Israeli children who were transferred from a traditional group setting to a cluster of family homes. Another study in Israel found a correlation between the type of care promoted in the setting and levels of staff violence. In more intensive, therapeutic settings, with higher concentrations of children with more severe difficulties, there were higher levels of staff physical maltreatment than in less intensive, rehabilitative settings (Attar-Schwartz, 2011).

Some studies have found a positive correlation between institution size and peer violence (Barter et al., 2004; Sinclair & Gibbs, 1998). In a qualitative study by Barter et al. (2004), several staff members from children’s homes argued that large institutions prohibited effective supervision of peer violence. Similarly, another study estimated a positive association between the institution size and physical maltreatment by staff (Attar-Schwartz, 2011). However, other studies found no association between institution size and children’s outcomes. For example, Attar-Schwartz (2008) found that the number of children in Israeli RCS was not associated with children’s emotional and behavioral difficulties. That study was based on social workers’ reports rather than on reports from the children themselves.

In Israel, Arab RCS serve Arab children and Jewish RCS serve Jewish children (Attar-Schwartz, 2011). For this reason, the current study examines levels of victimization among children belonging to the same ethnic group. There are no studies in Israel examining the differences in levels of peer victimization between youth in Arab and Jewish residential care facilities. However, Israeli national school violence studies found that students in Arab schools experienced higher levels of severe physical victimization by peers (e.g., being stabbed, or receiving a beating that resulted in needing medical treatment) than students in Jewish schools. Several researchers attributed these differences to the many sociocultural and sociopolitical differences between Jews and Arabs in Israel that might contribute to youth involvement in violent behavior (see Khoury-Kassabri, Benbenishty, Astor, & Zeira, 2004). However, it should be noted that these assumptions were not empirically tested. Furthermore, those studies found no significant differences between Jewish and Arab children in their levels of moderate physical victimization by peers (e.g., being kicked or punched).

Level 1 Hypotheses

1. Boys will report higher levels of physical victimization by peers than girls.

2. Younger children will report higher levels of physical victimization by peers than older children.

3. Adolescents with higher levels of emotional and social self-efficacy will report lower levels of physical victimization by peers.

4. Higher levels of child adjustment difficulties will be associated with higher levels of physical victimization by peers.

5. Positive perception of the RCS’ social climate (higher caretaker support and lower perception of strictness) will be associated with lower levels of victimization by peers.

6. Physical maltreatment by staff will be positively correlated with children’s physical victimization by peers.

7. Gender will moderate the association between individual-level factors (age, self-efficacy, adjustment difficulties, social climate, and staff violence) and children’s reports on peer physical violence.

Level 2 Hypotheses

8. Higher levels of physical victimization by peers will be found in traditional group institutions than in settings with familial elements.

9. Children in rehabilitative institutions will be exposed to higher levels of victimization by peers than children in therapeutic RCS.

10. Positive association will be found between the number of children in the RCS and children’s victimization by peers.

11. There will be differences in physical victimization by peers between Jewish and Arab institutions.

12. The differences between males and females in their reports of victimization by peers will differ in Jewish and Arab institutions (interaction between a Level 1 factor, gender, and a Level 2 factor, nationality).

Method

Sample

The sampling frame is derived from a list provided by the Israeli Welfare Ministry of 45 RCS with more than 20 residents in the relevant age group (11 to 19). Three RCS in remote villages were excluded from the sample, because their inclusion would have demanded special resources. As a result, 42 RCS were contacted for this study. Of these, seven opted not to participate in the study (17.9%), and three Ultraorthodox Jewish RCS were excluded because their directors asked for significant revisions to the questionnaire (7.14%). The response rate of the RCS contacted was 76.19%. The sample is broadly representative of the age, gender, ethnic affiliation, and treatment type of adolescents residing in rehabilitative and therapeutic care settings in Israel (Attar, 2006; Attar-Schwartz, 2009; National Council for the Child, 2010). The study’s sample includes 1,324 adolescents aged 11 to 19 from 32 RCS.

The mean age of the sample was 14.06 years (SD = 3.11); 54% of respondents were male. Most of the RCS included in this study were Jewish (75%) and the rest were Arab. Approximately 60% of the RCS were group institutions; the rest were settings with familial elements (see definitions above). About two thirds (62.5%) of the RCS were rehabilitative and the rest were therapeutic. Each RCS serviced an average of 102 youth of all ages (SD = 65.34). On average, 41.37 questionnaires, completed by respondents in the 11 to 19 age group, were returned (SD = 24.18). About 21 adolescents (1.56%) did not take part in the study, either because they or their parents refused, or because they decided to withdraw from the study even though they and their parents had given their consent.

Data Collection

Information was collected from the adolescents through an anonymous, structured, self-report, questionnaire. Letters were sent to the directors of the sampled RCS requesting assistance in collecting data. Each director also received a letter to be sent to parents or guardians outlining the purpose of the study and providing contact information in the case of questions. Each parent or guardian was asked to return a slip if they did not want the adolescent to take part in the project. The adolescents were told that their participation was strictly voluntary. They were told that they were free to withdraw from the study at any time for any reason.

The questionnaire was administered by specially trained research assistants, who made one or two visits to each facility to obtain maximum participation. At each visit, the questionnaire was administered to all residents between the ages of 11 and 19 who were on the premises and available at the time. The adolescents completed the questionnaires in the RCS. Those who completed the questionnaire at the first visit were asked not to complete it again. Given the sensitive nature of the questions, researchers asked the staff not to be present in the room when the youth were filling out the questionnaires, and they assured participants that their responses would remain confidential. On completing the questionnaire, the adolescents received a sheet listing child welfare agencies and services they could contact if they felt distress or in need of help from outside the RCS. In addition, the directors were asked to provide some organizational details about their settings through structured self-report questionnaires. These questionnaires included questions about the organizational characteristics and ethnic affiliation of the institution as detailed below. Questionnaires, procedures, consent forms, and instructions were reviewed by the Ethics Committee of the Hebrew University of Jerusalem.

Measures

Arab adolescents completed the questionnaires in Arabic; Jewish adolescents completed them in Hebrew. The questionnaire included several measures that originated from English-language scales (e.g., the Revised Social Climate Scale, Colton 1989). These were translated into Hebrew and Arabic. To enhance the compatibility of the Hebrew and Arabic versions, the authors followed common procedures, enlisting experts familiar with Hebrew and Arabic languages, as well as the two cultures, to back-translate and discuss the surveys. Furthermore, in the pilot phase of the study, the questionnaire, including the translated scales, were tested among approximately 20 children in care, and items that were found to be unclear were rephrased.

Dependent variable: Physical victimization by peers

Adolescents were presented with a list of aggressive physical behaviors and were asked to indicate whether they had been victims of any of those behaviors perpetrated by a peer in their RCS in the preceding month (responding options: 1 = never, 2 = once or twice, 3 = three times or more). The physical victimization by peers measure was drawn from the Hebrew and Arabic translation and adaptation by Benbenishty, Zeira, and Astor (2000) of the California School Climate Survey (Furlong et al., 2005). The measure included four items indicating whether in the previous month adolescents had been (1) grabbed and shoved, (2) kicked or punched, (3) seriously hit by others, or (4) hit with an object (stick, stone, chair, etc.) deliberately to hurt (α = 0.79). The subscale was based on the mean of the items, therefore ranging from 1 (never physically victimized) to 3 (victimized more than three times by all listed physical violence behaviors).

Individual-level correlates

Background characteristics

To assess the sampled youth’s personal characteristics, adolescents were asked to report their age and gender (0 = girl, 1 = boy).

Adolescent adjustment difficulties

Adolescents’ adjustment difficulties were assessed by a total difficulties measure derived from the self-report version of the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997) for 11 to 17 year olds, which was translated into Hebrew and Arabic by professionals. The translated questionnaires are available at the official website of the SDQ (www.sdqinfo.com). The SDQ is a 3-point Likert-type scale (0 = not true, 1 = somewhat true, 2 = certainly true) measuring four difficulties (hyperactivity, emotional symptoms, conduct problems, and peer problems), as well as prosocial behavior (Goodman, 1997). In this study, we used the total difficulties score that was calculated by summing the average scores of hyperactivity, emotional symptoms, conduct problems, and peer problems, therefore ranging from 0 (no reported adjustment difficulties) to 8 (high level of adjustment difficulties) (α = .75).

Self-efficacy

The current study included two subscales derived from the Self-Efficacy Questionnaire for Children (SEQ-C; Muris, 2001): (a) social self-efficacy (α = .64) that includes eight items relating to the perceived capability for peer relationships and assertiveness (e.g., “How well can you tell other children that they are doing something that you don’t like?”) and (b) emotional self-efficacy (α = .77) that includes eight items pertaining to the perceived capability of coping with negative emotions (e.g., “How well do you succeed in cheering yourself up when an unpleasant event has happened?”). Each item was scored on a 5-point scale (1 = not at all; 5 = very well). Each measure was created by computing the mean score of the items. A high score indicated a high level of self-efficacy.

Social climate

Adolescents’ perceptions of the social climate at their RCS were examined using the RSCS (Revised Social Climate Scale; Colton, 1989). The original Social Climate Scale was devised by Heal, Sinclair, and Troop (1973) to assess how delinquent boys in community homes perceive their social environments. Colton (1989) modified the scale so that it can be applied to residential and foster settings with at-risk youth. The scale involves simple statements about the social environment in the institution. Respondents rate these statements as either true (1) or false (0). The RSCS consists of five subscales. The current study included the two subscales that seemed most relevant. The caretaker support subscale includes 10 items (α = .79) that examine the extent to which respondents perceive staff to be interested, warm, and supportive (e.g., “You usually talk to the staff if you have a personal problem”). The summative score of the subscale ranges from 0 (not supportive at all) to 10 (very supportive). The strictness subscale includes 8 items (α = .63) that examine young people’s perceptions of their authority relationship with the staff (e.g., “You get into serious trouble if you don’t obey the rules here”). The summative score of the subscale ranges from 0 (not strict at all) to 8 (very strict).

Physical maltreatment by staff

Adolescents were presented with a list of aggressive behaviors and asked to indicate whether, within 1 month of the survey, they had been victims of such behavior perpetrated by a staff member in their RCS. The questionnaire defines staff as any adult who works or volunteers in the institution. This can include social workers, direct caregivers, home parents, directors, psychologists, volunteers, and administrative staff. Four items (α = .74) drawn from Benbenishty and associates’ (2000) translation and adaptation of the California School Climate Survey into Hebrew and Arabic were used (Furlong et al., 2005): (1) being grabbed or shoved, (2) pinched, (3) slapped, and (4) kicked or punched by a staff member with the intention to hurt. Respondents were assigned one point for each behavior they reported. Scores ranged from zero to four physically violent behaviors by a staff member toward the adolescent.

RCS-Level Correlates

To assess whether the care environment is linked to victimization by peers, this study included four RCS-level correlates as reported by the directors: type of care, institutional structure, institution size, and ethnic affiliation. Type of care indicates whether the RCS was rehabilitative or therapeutic. RCS structure indicates whether the RCS was arranged as a collection of familial groups, as a mixed familial-group setting, or as a traditional group setting. Institution size was measured as the overall number of youth residing in an RCS. Ethnic affiliation identifies the RCS as either an Arab or Jewish institution.

Analytic Plan

The present study used Hierarchical Linear Modeling (HLM), a technique that allows the investigation of data organized hierarchically on more than one level (Raudenbush & Bryk, 2002). Adolescents were nested within RCS. The Level 1 model represents the relationships of the individual-level characteristics to physical victimization by peers; the Level 2 model captures the relationships of RCS-level variables to physical victimization by peers. Adolescent-level variables were centered on the group mean. However, this approach overlooks the variation of these centered variables between the RCS participating in the study. To overcome this, the group means of the centered variables were reintroduced as second-level (RCS-level) variables (for further explanation see Hox, 2002; Kreft & De Leeuw, 1999).

Results

Descriptive Statistics

The results show that adolescents in RCS are highly exposed to victimization by peers. Most participants (56%) reported being exposed to at least one form of physical victimization during the previous month. Approximately 45% reported having been grabbed or shoved and 38.6% reported having been kicked or punched at least once during the last month. More than a quarter of the participants (29.1%) reported that they had been hit with an object deliberately to hurt, and one in five adolescents reported they had been seriously hit. Another important finding concerns the high rate of reports of being victimized by staff. One-quarter of the adolescents (24.7%) reported being exposed to at least one form of physical maltreatment by staff during the previous month.

Multi-Level Model

The results of the fully unconditional model—a model with no independent variables—indicate that a significant amount of the variance in physical victimization by peers (11.48%) lies between RCS, χ²(31) = 188.175, p < .001. Hence, a two-level analysis was conducted.

Individual-Level Factors

Table 1 presents estimates of the relationships among individual-level correlates and adolescents’ reports of physical victimization by peers in the month prior to the survey. The results, based on the HLM analysis, suggest that, as expected, boys and younger adolescents reported statistically significantly more victimization than girls and older adolescents. In addition, again consistent with our hypotheses, both higher levels of adjustment difficulties reported by adolescents and higher levels of maltreatment by RCS staff are positively associated with victimization.

Table 1.

The Relationships Between Physical Victimization by Peers and Adolescent-Level Factors.

	Coefficient	SE	t ratio
Background characteristics
Gender (male)	0.158	0.032	4.817***
Age	−0.067	0.020	−3.266***
Adjustment difficulties	0.103	0.025	4.068***
Self-efficacy
Social self-efficacy	−0.056	0.023	−2.371***
Emotional self-efficacy	−0.004	0.025	−0.190
Perception of the RCS social climate
Caretaker support	0.018	0.025	0.706
Strictness	0.079	0.025	3.179*
Physical maltreatment by staff	0.083	0.026	3.232***
Interaction effects
Gender × Age	−0.037	0.030	−1.240
Gender × Adjustment difficulties	0.101	0.034	3.493**
Gender × Social self-efficacy	0.014	0.034	0.396
Gender × Emotional self-efficacy	−0.001	0.035	−0.032
Gender × Caretaker support	−0.048	0.034	−1.411
Gender × Strictness	−0.063	0.033	−1.839*
Gender × Physical maltreatment by staff	0.003	0.34	.076
Variance component^a	0.19348
Explained variance	32.85%

Note. N = 1,324 adolescents in 32 RCSs. RCS = residential care setting.

Variance component (r) null model = 0.28812.

p < .05. **p < .01. ***p < .001.

The results supported some of the hypotheses at the individual level. Social self-efficacy was found to be positively and significantly associated with victimization, but no significant link was found between emotional self-efficacy and levels of victimization by peers. Adolescents who perceived the social climate at the RCS to be stricter were more likely to report being physically victimized by their peers, but perception of caretaker support was insignificantly linked with victimization.

We explored the interaction effects of gender and each of the independent factors in predicting victimization by peers. As shown in Table 1, gender was found to moderate the relationship between adjustment difficulties and children’s reports of victimization by peers (G10 = .101). This finding indicates that the association between children’s reports of adjustment difficulties and physical victimization by peers is stronger among males than females. The opposite trend was found in the association between perception of strictness and physical victimization by peers (G10 = −.062). The association between adolescents’ perception of the staff’s strictness and physical victimization by peers was weaker among boys than girls. These were the only significant interaction effects found between gender and other child-level factors in predicting victimization by peers in RCS.

Overall, as shown in Table 1, adolescent-level variables explained 32.85% of the variance within RCS in adolescents’ reports of physical victimization.

RCS-Level Factors

Table 2 presents, based on the HLM analysis, the relationships among factors at the institution level and levels of victimization by peers. As expected, levels of physical victimization by peers were lower in RCS with a familial element than in traditional group settings. The type of care, institution size, and cultural affiliation were insignificantly associated with victimization.

Table 2.

The Relationships Between RCS-Level Characteristics and RCS’ Means of Physical Victimization by Peers.

Fixed effects	Coefficient	SE	t ratio
Adolescents aggregative variables
Percentage of males	0.103	0.058	1.751
Average age of adolescents	−0.030	0.025	−1.93
Average adjustment difficulties	0.043	0.014	3.212**
Average social self-efficacy	0.015	0.177	0.086
Average emotional self-efficacy	0.078	0.173	0.450
Average caretaker support	−0.005	0.037	−0.127
Average strictness	−0.052	0.034	−1.536
Average physical maltreatment by staff	0.262	0.066	3.962***
Cultural affiliation (Jewish)	0.159	0.109	1.460
Organizational factors
RCS structure (group RCS)	0.108	0.051	2.148*
Setting type (rehabilitative)	−0.072	0.065	−1.103
RCS size (# children)	0.000	0.000	0.640
Variance component^a	0.00774
Explained variance	79.29%

Note. N = 1,324 adolescents in 32 RCSs. RCS = residential care setting.

Variance component (U₀) null model = 0.03738.

p < .05. **p < .01. ***p < .001.

We explored whether aggregated factors from the individual level were associated with reports of victimization at the RCS level. We found that higher levels of physical victimization by peers were positively correlated with the percentage of children with adjustment difficulties at the RCS and also with levels of reports of physical victimization by RCS staff. The effects of the aggregated factors were above and beyond their contribution at the individual-level factors.

As mentioned above, the variance in physical victimization by peers between RCS is 11.48%. Table 2 shows that RCS-level variables explain 79.29% of this unexplained variance.

To explore our hypothesis regarding the interaction between gender and Level 2 factors, we first examined, as required by the HLM, whether the slopes of the association between gender and physical victimization vary significantly between RCS. We found that none of the slopes of Level 1 factors (including gender) had a significant amount of variance. These findings indicate that the association between victimization and Level 1 factors (such as gender, age, and adjustment difficulties) is not affected by institution level factors. Therefore, all slopes in this study were fixed.

Discussion

The study shows that adolescents in RCS are exposed to high levels of peer violence: More than 50% of the participants reported that they had been physically victimized by at least one act of physical violence by a peer in the previous month. This high level of victimization is consistent with findings in previous studies (see, for example, Barter, 2008; Gibbs & Sinclair, 2000).

However, it may be misleading to compare levels of victimization reported in previous studies and the levels found in our study because of methodological differences in the time frame and events the children were asked to report on. In our study, children reported on violent acts that had occurred in the last month, while other studies referred to violent acts that had occurred at any time during their stay in residential homes (Barter, 2008). We chose a limited time frame following the recommendation of Benbenishty and Astor (2005), who argue that asking a child about a period that spans many years may lead to information processing that is more vulnerable to retrospective interpretation and screening. They recommend using shorter time frameworks, such as the one used in this study, because recent events are more easily recalled.

The study’s results highlight the importance of applying the ecological-contextual perspective to the research of correlates of peer-to-peer physical violence in residential care. They show that adolescent victimization by peers is explained simultaneously by the adolescents’ individual characteristics and by the social context in which the adolescents’ development occurs (Kuppens, Grietens, Onghena, Michiels, & Subramanian, 2008). Given the vulnerability of these young people, many of whom experienced maltreatment prior to care, it is critical to find ways to protect them from further harm while in the care setting. Knowing which factors may increase a child’s risk of being victimized by peers will be an essential part of designing and implementing interventions to deal with violence in out-of-home care facilities.

Individual Risk Factors for Victimization Among Adolescents

As expected, more males reported being exposed to physical victimization by peers than females. These results are consistent with previous studies that found that males in care are more involved than females in physical aggression both as victims and as perpetrators (Attar-Schwartz, 2008; Glisson et al., 2002). This trend might be explained by differences in the socialization processes of boys and girls. While physical aggression might be perceived as a legitimate way of expressing masculinity, it might be considered to be unfeminine among females (Barter, 2008). In school violence studies it has been shown that girls are less exposed to direct violence, but more exposed to relational and indirect forms of aggression (Attar-Schwartz & Khoury-Kassabri, 2008; Bjorkqvist, 1994). Future studies of young people in care should further examine the correlation of gender to relational, verbal, and sexual forms of violence.

In keeping with previous studies of victimization by peers of children in residential care (e.g., Gibbs & Sinclair, 2000) and in schools (e.g., Khoury-Kassabri, 2011), our study found that younger children have higher levels of victimization by peers than older children. Younger children might be less capable of protecting themselves from violence perpetrated by other children, especially those who are older than them (Khoury-Kassabri et al., 2009). Smith, Madsen, and Moody (1999) also suggest that younger children have not yet acquired the social assertiveness skills needed to effectively deal with and discourage bullying.

As predicted, adolescents with adjustment difficulties experienced higher levels of victimization by peers. Previous studies found that adjustment difficulties that represent internalizing behaviors (such as emotional symptoms) and externalizing behaviors (such as conduct problems) increase a child’s vulnerability to victimization by RCS staff (e.g., Attar-Schwartz, 2011) as well as by peers (Smith et al., 2004; Wolke, Woods, & Samara, 2009). We explored the moderating role gender plays in the association between these two factors and found that the link between adjustment difficulties and victimization is stronger among males than females. The implications of such findings to practice and policy should be further explored.

As hypothesized, adolescents with lower levels of social self-efficacy were victimized more often. Adolescents with little belief in their social abilities might be less assertive when faced with a violent act. They may have lower belief in their ability to be accepted by others, and may suffer from feelings of social maladjustment and loneliness. These characteristics could account for the increased risk of these adolescents to be victimized (Andreou & Metallidou, 2004; Crick & Grotpeter, 1995; Kokkinos & Kipritsi, 2012).

Exposure to violence by RCS staff also increases an adolescent’s probability of experiencing violence from peers. Bandura’s (1982) Cognitive-Social Learning Theory indicates that adolescents will model the social interactive behavior of the adults in their environment. Assuming this theory, children might turn to violence to solve problems and conflicts when they observe the adults in their environment doing so, especially toward those who are seen to be the targets of adult violence (Geving, 2007; Lewis, Romi, Qui, & Katz, 2005). The young perpetrators of violence might also assume that their attacks on the victims of staff violence might not result in punishment from staff—a circular situation in which violence toward victimized adolescents might increase.

The perception of staff as strict was associated with both males’ and females’ reports on peer violence but the association was stronger among females. This might be because females tend to seek more formal and informal help than males (Grinstein-Weiss, Fishman, & Eisikovits, 2005). When girls perceive staff to be strict, their tendency to seek help, including in cases of victimization by peers, might decrease, resulting in an increase of exposure to violence. This hypothesis should be further tested in future research.

Contextual Risk Factors for Victimization by Peers

The current study makes an important contribution by exploring the role of various factors in different levels of adolescents’ physical victimization by peers. For example, we found that levels of peer victimization are higher in traditional group settings than in familial groups or mixed familial-group settings. It might be argued that in family-like placements the likelihood of nurturing strong relational bonds is higher. Adolescents in such settings may be encouraged to report victimization and seek help more strongly than their counterparts in other settings, which in turn may reduce the probability of being victimized. It might also be that settings with familial elements allow for more effective supervision of peer violence than traditional settings (Barter et al., 2004).

The results reveal that RCS with a high percentage of children with adjustment difficulties have higher levels of physical victimization, regardless of the adolescent’s own level of difficulties. A high percentage of children with adjustment difficulties may pose a challenge for the institution and its staff, especially when the staff is undertrained and unable effectively to handle behavioral and emotional difficulties and protect youth from being victimized by their peers. Residential care staff in many countries, including Israel, often have insufficient training and are therefore potentially unable to deal effectively with problematic behaviors (see review in Attar, 2006). Placement of youth in settings with undertrained staff may increase their risk of being victimized (Groze, 1990; Hobbs, Hobbs, & Wynne, 1999). This finding calls for further investment of the welfare system in training RCS staff.

We found that RCS with higher levels of staff physical violence toward adolescents also have higher levels of peer-to-peer violence. These effects were found above and beyond their influence at the individual level. The relationship between staff and adolescents is a major contextual factor in the social atmosphere of the RCS. Because many of the children in RCS were removed from their homes following violent interactions within their families, placement in environments where the adults in positions of authority are perceived to be overly strict, or are possibly even abusive, might damage the children’s ability to form attachment relationships and, in turn, their chances of rehabilitation. It might also diminish the adolescents’ trust in others and increase their sense of alienation (Dwyer, Osher, & Hoffman, 2000). These circumstances could be true for victims and perpetrators of violence. Perpetrators may feel that violence is a legitimate way to handle difficulties, and may therefore turn to violence more frequently. Victims may not feel encouraged to seek help from staff members who are also perpetrators and thus perceived as unable to help finding ways to avoid violent interactions.

The study’s findings emphasize the need to address victimization in RCS from a comprehensive perspective. In addition to intervening at the personal level with the victimized children and the perpetrators of violence, there is a clear need to provide RCS staff with support and tools for intervention. Some RCS staff might be facing a group of residents with diverse needs that they are unable to meet with the resources and training currently available (Barter et al., 2004). Further training of RCS staff is essential if they are to understand the impact these factors have on youth entering care. Such training should provide them with the tools necessary to help children with adjustment difficulties and other needs resolve conflicts without resorting to violence.

Limitations and Recommendations for Future Research

The current study is the largest of its kind to investigate adolescents’ reports of physical victimization by peers in RCS and the multilevel correlates of individual and contextual factors. However, because of the study’s cross-sectional design, our ability to derive causal inferences was limited. To establish a direct causal effect of RCS and child factors on victimization longitudinal studies would be required.

Moreover, although the individual- and institution-level characteristics included in our study clearly contribute to an explanation of the variance between and within institutions, some variance remains to be explained. For example, factors related to the adolescents’ placement, such as the reason for their placement and the number of placements they have had, should be explored in the future. Further adolescent variables (such as sexual orientation, previous exposure to violence, and involvement as perpetrators in violence) as well as institutional factors (such as staff training, needs, supervision levels, and RCS policies regarding violence) should also be examined in future studies.

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) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Mona Khoury-Kassabri PhD is an Associate Professor at the School of Social Work and Social Welfare at the Hebrew University of Jerusalem. Her research evolves around issues related to the welfare of children and youth. She is especially interested in the area of children and youth at risk for becoming perpetrators and victims of violent and delinquent acts.

Shalhevet Attar-Schwartz PhD is a Senior Lecturer at the School of Social Work and Social Welfare at the Hebrew University of Jerusalem. Her research focuses on the well-being and welfare of children and youth. Particularly, she has published extensively on following issues: children’s quality of life in out-of-home settings, youth violence, children’s rights, and intergenerational relationships in the family.

References

Ajdukovic

Franz

B. S.

(2005). Behavioral and emotional problems of children by type of out-of-home care in Croatia. International Journal of Social Welfare, 14, 163-175.

Andreou

Metallidou

(2004). The relationship of academic and social cognition in behavior in bullying situations among Greek primary school children. Educational Psychology, 24, 27-41.

Astor

R. A.

Pitner

R. O.

Duncan

B. B.

(1996), Ecological approaches to mental heakth consolation with teachers on issues related to youth and school violence. Journal of Negro Education, 65, 336-355.

Attar

(2006). Child, family, and institutional characteristics as predictors of the psychosocial and educational functioning of children at-risk in residential care settings (Unpublished doctoral dissertation). Hebrew University of Jerusalem (Hebrew).

Attar-Schwartz

(2008). Emotional, behavioral and social problems among Israeli children in residential care: A multi-level analysis. Children and Youth Services Review, 30, 229-248.

Attar-Schwartz

(2009). School functioning of children in residential care: The contributions of multilevel correlates. Child Abuse & Neglect, 33, 429-440.

Attar-Schwartz

(2011). Maltreatment by staff in residential care facilities: The adolescents’ perspectives. Social Service Review, 85, 635-664.

Attar-Schwartz

Khoury-Kassabri

(2008). Indirect versus verbal forms of victimization at school: The contribution of student, family, and school variables. Social Work Research, 32, 159-170.

Bandura

(1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215.

10.

Bandura

(1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122-147.

11.

Bandura

(1997). Self-efficacy. New York, NY: W. H. Freeman.

12.

Bandura

Pastorelli

Barbaranelli

Caprara

G. V.

(1999). Self-efficacy pathways to childhood depression. Journal of Personality and Social Psychology, 76, 258-269.

13.

Barter

(2008). Prioritising young people’s concerns in residential care: Responding to peer violence. In Kendrick

A. J.

(Ed.), Residential child care: Prospects and challenges (pp. 137-151). London, England: Jessica Kingsley Publishers.

14.

Barter

Renold

Berridge

Cawson

(2004). Peer violence in children’s residential care. Basingstoke, UK: Palgrave.

15.

Benbenishty

Astor

R. A.

(2005). School violence in context: Culture, neighborhood, family, school, and gender. New York, NY: Oxford University Press.

16.

Benbenishty

Zeira

Astor

(2000). Violence in Israel education system: A summary report. Jerusalem, Israel: The Hebrew University (Hebrew).

17.

Bjorkqvist

(1994). Sex differences in physical, verbal, and indirect aggression: A review of recent research. Sex Roles: A Journal of Research, 30, 177-183.

18.

Bronfenbrenner

(1979). The ecology of human development. Cambridge, MA: Harvard University.

19.

Children and Youth Service. (2005). Children and youth service: Children in residential care settings, 2003–2004. Jerusalem, Israel: Welfare Ministry and Brookdale Institute (Hebrew).

20.

Colton

(1989). Foster and residential children’s perceptions of their social environments. British Association of Social Workers, 19, 217-233.

21.

Crick

N. R.

Grotpeter

J. K.

(1995). Relational aggression, gender, and social-psychological adjustment. Child Development, 66, 710-722.

22.

Daly

D. L.

Dowd

T. P.

(1992). Characteristics of effective, harm-free environments for children in out-of-home care. Child Welfare, 71, 487-496.

23.

Davidson-Arad

Golan

(2007). Victimization of juveniles in out-of-home placement: Juvenile correctional facilities. British Journal of Social Work, 37, 1007-1025.

24.

Dolev

Benbenishty

Timar

(2001). Decision committees in Israel: Their organization, work processes, and outcomes. Jerusalem, Israel: JDC-Brookdale Institute (Hebrew).

25.

Dwyer

K. P.

Osher

Hoffman

C. C.

(2000). Creating responsive schools: Contextualizing early warning, timely response. Exceptional Children, 66, 347-365.

26.

Ellonen

(2008). Adolescent delinquency and social control in Finnish schools: A multilevel analysis. Journal of Scandinavian Studies in Criminology and Crime Prevention, 9, 47-64.

27.

Ellonen

Pösö

(2011). Violence experiences in care: Some methodological remarks based on the Finnish child victim survey. Child Abuse Review, 20, 197-212.

28.

Freundlich

Avery

R. J.

Padgett

(2007). Care or scare: The safety of youth in congregate care in New York City. Child Abuse & Neglect, 31, 173-186.

29.

Furlong

M. J.

Greif

J. L.

Bates

M. P.

Whipple

A. D.

Jimenez

T. C.

Morrison

(2005). Development of the California School Climate and Safety Survey—Short form. Psychology in the Schools, 42, 137-149.

30.

Geving

A. M.

(2007). Identifying the types of student and teacher behaviors associated with teacher stress. Teaching and Teacher Education, 23, 624-640.

31.

Gibbs

Sinclair

(1999). Treatment and treatment outcomes in children’s homes. Child & Family Social Work, 4, 1-8.

32.

Gibbs

Sinclair

(2000). Bullying, sexual harassment and happiness in residential children’s homes. Child Abuse Review, 9, 247-256.

33.

Glisson

Hemmelgarn

A. L.

Post

J. A.

(2002). The shortform assessment for children: An assessment and outcome measure for child welfare and juvenile justice. Research on Social Work Practice, 12, 82-106.

34.

Goodman

(1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581-586.

35.

Gottfredson

G. D.

Gottfredson

D. C.

Payne

A. A.

Gottfredson

N. C.

(2005). School climate predictors of school disorder: Results from a national study of delinquency prevention in schools. Journal of Research in Crime & Delinquency, 42, 412-444.

36.

Grinstein-Weiss

Fishman

Eisikovits

(2005). Gender and ethnic differences in formal and informal help seeking among Israeli adolescents. Journal of Adolescence, 28, 765-779.

37.

Groze

(1990). An exploratory investigation into institutional mistreatment. Children and Youth Services Review, 12, 229-241.

38.

Heal

Sinclair

Troop

(1973). Development of a social climate questionnaire for use in approved schools and community homes. British Journal of Sociology, 24, 222-235.

39.

Heflinger

C. A.

Simpkins

C. G.

Combs-Orme

(2000). Using the CBCL to determine the clinical status of children in state custody. Children and Youth Services Review, 22, 55-73.

40.

Herrera

V. M.

Stuewig

(2011). Impact of child maltreatment and domestic violence. In Barter

Berridge

(Eds.), Children behaving badly: Peer violence between children and young people (pp. 155-166). Chichester, UK: Wiley-Blackwell.

41.

Hobbs

G. F.

Hobbs

C. J.

Wynne

J. M.

(1999). Abuse of children in foster and residential care. Child Abuse & Neglect, 23, 1239-1252.

42.

Hox

(2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum.

43.

Kendrick

(2011). Peer violence in provision of children in care. In Barter

Berridge

(Eds.), Children behaving badly: Peer violence between children and young people (pp. 71-84). Chichester, UK: Wiley-Blackwell.

44.

Khoury-Kassabri

(2011). Student victimization by peers in elementary schools: Individual and teacher-class-level predictors. Child Abuse & Neglect, 35, 273-282.

45.

Khoury-Kassabri

Astor

R. A.

Benbenishty

(2009). Middle Eastern adolescents’ perpetration of school violence against teachers: A cross-cultural and ecological analysis. Journal of Interpersonal Violence, 24, 159-182.

46.

Khoury-Kassabri

Benbenishty

Astor

Zeira

(2004). The contribution of community, family and school variables on student victimization. American Journal of Community Psychology, 34, 187-204.

47.

Kokkinos

C. M.

Kipritsi

(2012). The relationship between bullying, victimization, trait emotional intelligence, self-efficacy and empathy among preadolescents. Social Psychology of Education, 15, 41-58.

48.

Kreft

De Leeuw

(1999). Introducing multilevel modeling. London, England: Sage.

49.

Kuppens

Grietens

Onghena

Michiels

Subramanian

S. V.

(2008). Individual and classroom variables associated with relational aggression in elementary-school aged children. Journal of School Psychology, 46, 639-660.

50.

Lees

Morris

Wheatley

(1994). Time to listen. London, England: Child-Line.

51.

Lewis

Romi

Qui

Katz

Y. J.

(2005). Teachers’ classroom discipline and student misbehavior in Australia, China and Israel. Teaching and Teacher Education, 21, 729-741.

52.

MacLeod

(1999). The abuse of children in institutional settings: Children’s perspectives. In Stanely

J. M. N.

Penhale

(Eds.), Institutional abuse: Perspectives across the life course (pp. 44-65). London, England: Routledge.

53.

Mitic

Rimer

(2002). The educational attainment of children in care in British Columbia. Child & Youth Care Forum, 31, 397-414.

54.

Monks

C. P.

Smith

P. K.

Naylor

Barter

Ireland

J. L.

Coyne

(2009). Bullying in different contexts: Commonalities, differences and the role of theory. Aggression and Violent Behavior, 14, 146-156.

55.

Mosek

Taus

Shomodi

(1997). The implications of an organizational change in residential care setting for children: A transition from group setting to family group homes. Welfare and Society, 18, 303-325. (Hebrew).

56.

Muris

(2001). A brief questionnaire for measuring self-efficacy in youths. Journal of Psychopathology and Behavioral Assessment, 23, 145-149.

57.

National Council for the Child. (2010). Children in Israel: Statistical abstract. Jerusalem, Israel: Author (Hebrew).

58.

Özbay

Özcan

Y. Z.

(2008). A test of Hirschi’s Social Bonding theory. International Journal of Offender Therapy and Comparative Criminology, 52, 134-157.

59.

Raudenbush

S. W.

Bryk

S. A.

(2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.

60.

Schiff

Benbenishty

(2006). Functioning of Israeli group-homes alumni: Exploring the differences and in-care correlates. Children and Youth Services Review, 28, 133-157.

61.

Sinclair

Gibbs

(1998). Children’s homes: A study in dicersity. Chichester: Wiley.

62.

Singh

Bussey

(2009). The development of a peer aggression coping self-efficacy scale for adolescents. British Journal of Developmental Psychology, 27, 971-992.

63.

Smith

P. K.

Madsen

K. C.

Moody

J. C.

(1999). What causes the age decline in reports of being bullied at school? Towards a developmental analysis of risks of being bullied. Educational Research, 41, 267-285.

64.

Smith

P. K.

Talamelli

Cowie

Naylor

Chauhan

(2004). Profiles of non-victims, escaped victims, continuing victims and new victims of school bullying. British Journal of Educational Psychology, 74, 565-581.

65.

Vinnerljung

Oman

Gunnarson

(2005). Educational attainments of former child welfare clients: A Swedish national cohort study. International Journal of Social Welfare, 14, 265-276.

66.

Wolke

Woods

Samara

(2009). Who escapes or remains a victim of bullying in primary school? British Journal of Developmental Psychology, 27, 835-851.