Public Policy Relevance Statement

The current study advances the similarities and differences between the predictors of cyberbullying and in-person violence, informed by socio-ecological framework for integrating various factors associated with youth violence. The study examines simultaneously the effects of factors at the individual level, family level, and contextual factors on youth involvement in cyberbullying. The findings indicated no significant gender differences with respect to involvement in cyberbullying, although with regard to in-person violence, it was found that males were more physically violent than were females. These findings are helpful for practitioners working with youth who perpetrate cyberbullying and violence in suggesting the need to provide them with skills and tools to intervene in preventing all types of violence.

Introduction

Over the past decades, we have seen an enormous increase in the individual use of information and communication technologies (ICTs). This development involves significant intrusion of online media into both adults’ and young people’s lives. Today’s children and youth have been born into this environment and are therefore referred to as “digital natives” (Palfrey & Gasser, 2008). ICTs provide them with unprecedented access to learning and self-exploration activities (Asterhan & Bouton, 2017; Blau, 2014), as well as to communication with others both within and outside their face-to-face social networks (Mesch, 2012; Mishna & Van Wert, 2016).

Despite the many benefits of cyber interactions, they also involve risks for youth, particularly cyberbullying (Khoury-Kassabri et al., 2016; Berson et al., 2002). Cyberbullying is defined as to the use of online platforms to bully others, whether individuals or groups (Bhat, 2008; Kowalski et al., 2014). Previous studies have indicated that “in-person bullying” or face-to face violence is significantly associated with involvement in cyberbullying, and that both bullying behaviors share common predictors (Arnarsson et al., 2020; Kim et al., 2017). Several researchers have suggested that cyberbullying is as an extension of face-to-face violence (McCuddy & Esbensen, 2017; Olweus, 2012). Mishna et al. (2012) found that cyberbullying perpetration is positively associated with youth involvement in “in-person” violence (see also, Kowalski & Limber, 2013; Mishna et al., 2016).

The important findings regarding the overlap between cyberbullying and in-person bullying in general notwithstanding, few studies have examined questions related to specific types of violence (Kim et al., 2017; Kowalski & Limber, 2013). To address this gap, the current study aims, first, to expand on previous work by exploring the association between cyberbullying and a wide array of violent acts: severe physical violence causing serious direct physical damage or threat of physical damage (CDC, 2007); moderate physical violence (violent acts such as pushing and shoving); and indirect violence (behaviors intended to damage others’ social relations and feeling of peer acceptance (Archer & Coyne, 2005).

Furthermore, cyberbullying has been recently conceptualized as sharing important characteristics with in-person bullying, with similar primary causes. Thus, the second aim of the current study is to conduct a comprehensive comparison of the predictive effects of a series on independent factors on cyberbullying and in-person violent behaviors by exploring the similarities and differences between them.

The present study is informed by Bronfenbrenner’s (1979) theory, which provides a socio-ecological framework for understanding human behavior, including youth violence. This theory considers human behavior as a combination of individual factors and contextual and environmental variables (social and physical) (Benbenishty & Astor, 2008). It examines individual factors (including gender, age, religiosity, and impulsivity), family factors (parental support, monitoring, and conflict), and contextual factors (commitment to school, exposure to community violence, and affiliation with delinquent peers) as they affect engagement in cyberbullying and the three types of in-person violence mentioned above (Espelage et al., 2023).

Individual Factors

Gender

The literature on the effects of gender on the involvement of children and youth in cyberbullying and in-person violence has produced inconsistent results. Many previous studies have found boys to be more likely to engaged in cyberbullying (Fanti et al., 2012; Floros et al., 2013; Heiman et al., 2014). However, other studies have failed to find any significant gender differences (Hinduja & Patchin, 2008; Mishna et al., 2012). Moreover, a few studies have reported higher cyberbullying rates among girls (e.g., Low & Espelage, 2013). Gender differences in indirect violence have also been reported inconsistently. Whereas many previous studies have reported girls to be engaged in indirect violence more frequently than are boys (Marshall, 2011), other studies have reported the opposite (Juliano et al., 2006; Ligthart et al., 2005), and still others have found no significant gender differences (Card et al., 2008). Contrary to these mixed findings, consistent results have been reported with respect to gender differences in physical violence in adolescence; specifically, boys engage in moderate and severe physical violence more frequently than do girls (Björkqvist, 2018; Cleverley et al., 2012; Kawabata et al., 2011).

Age

The literature is similarly inconsistent with regard to age differences in perpetration of cyberbullying and in-person violence. Whereas Williams and Guerra (2007) found that rates of involvement in cyberbullying decreased with age, Mishna et al. (2012) found the opposite. In yet another study, which examined students reporting to have cyberbullied others, no age differences were found (Patchin & Hinduja, 2006; see also Wang et al., 2009). The same inconsistency is evident with regard to in-person violence, with some studies finding that aggressive tendencies increase with age (Pepler & Craig, 2005), and others finding the opposite (Steinberg & Morris, 2001).

Impulsivity

Another individual factor found significant in explaining child and youth involvement in cyberbullying and violence is the extent to which they behave impulsively (Moffitt, 1993; Vitulano et al., 2010). Impulsive individuals tend to act without forethought and to pursue their self-interest without regard to potential consequences. Ample research has consistently identified impulsivity as an individual risk factor for violence. In a large study of Greek students (N = 2,017), impulsivity was associated with cyberbullying (Floros et al., 2013). Similarly, Loeber et al. (2012) found that impulsive youth reported higher levels of involvement in violence.

Religiosity

Based on Hirschi’s (1969) social bond theory, the current study explored the effects of religiosity on youth involvement in violence. According to this theory, religious adolescents might avoid violence and delinquency because religiosity fosters attachment, helps nurture normative beliefs, and strengthens beliefs about the wrongfulness of antisocial behaviors. Research has supported this prediction by revealing that religious engagement serves to deter youth from involvement in violence (Massarwa et al., 2019; Pirutinsky, 2014). The few studies that have explored the association between cyberbullying and religiosity have also found the latter to be associated with lower levels of cyberbullying (Pradubmook-Sherera & Karansupamasb, 2020).

Family Factors

Family factors are pivotal in shaping children’s development. The present study focuses on three main family or parental aspects: parental support, monitoring by parents, and conflict with parents. According to Hirschi (1969), children who enjoy close relations with their parents, whose parents are involved in their lives, and who have good communication with and supervision by their parents may be involved less in delinquency as they do not want to jeopardize these relations. During adolescence, parent–child relationships undergo qualitative changes due to the children’s increased autonomy, but close relations with parents are crucial at this stage as well (Massarwa & Khoury-Kassabri, 2017; Simpkins et al., 2009). Finally, poor child-parent relations have been associated with in-person bullying (Farrington, 2005; Hamner et al., 2015; Loeber & Stouthamer-Loeber, 1986) as well as with cyberbullying (Hay et al., 2010).

Contextual Factors

School Commitment

Hirschi’s (1969) social bond theory suggests that commitment and motivation to succeed in conventional activities may decrease youth involvement in delinquency. This assumption has been empirically supported by many studies findings that school bonds and commitment reduce the likelihood of delinquent behavior and violence (Khoury-Kassabri, 2019; Benbenishty & Astor, 2008; Jang & Rhodes, 2012). Furthermore, poor academic performance and achievements were associated with negative peer relationships that could impact youths’ involvement in cyberbullying (Nansel et al., 2001). Benbenishty and Astor (2008) argued that the socio-ecological framework could be used to examine how a school’s external contexts interacted with internal school and student characteristics to affect levels of in-school victimization.

Association with Delinquent Peers

Sutherland’s differential association-reinforcement theory of criminal behavior (Burgess & Akers, 1966) suggests that adolescents’ behaviors and attitudes are shaped largely by their peer group (Brook et al., 1990; Trucco et al., 2011). This theory has been supported by empirical studies demonstrating that affiliation with delinquent peers is a leading cause of delinquent and violent behavior (Brook et al., 2011; Keijsers et al., 2012; Negriff & Tricket, 2011; Van Ryzin et al., 2012). Few studies, however, have examined whether and how peer affiliation is associated with cyberbullying specifically. An exception is Cappadocia et al. (2013), who found levels of cyberbullying to be higher the more the youth was associated with friends who exhibited antisocial behaviors.

Neighborhood Exposure to Violence and Crime

Exposure to crime and violence in one’s neighborhood predicted youth involvement therein (Benbenishty & Astor, 2008; Gorman-Smith et al., 2004; Hamner et al., 2015Selner‐O’Hagan et al., 1998). However, the effects of community violence on involvement in cyberbullying, as perpetrator or victim, have, to the best of our knowledge, never been studied. A recent meta-analysis concluded that cyberbullying research should focus also on community factors, rather than only on individual and family factors (Kowalski et al., 2014).

Arab Youth in Israel and their Socio-political Context

Arabs represent about one-fifth (21.1%) of the total Israeli population (Israel Central Bureau of Statistics, 2021). This ethnonational minority is characterized by significantly lower socioeconomic status and formal and informal discrimination in the allocation of state and social resources (Hammack, 2010), despite being formally recognized as citizens since the establishment of the State of Israel in 1948 (Kadan et al., 2017). On a psychological level, it is difficult for members of the Arab minority to feel at home in Israel because the state is essentially identified with Jewish identity and history, and its symbols and narratives are grounded in the perspective of its Jewish majority (Abu-Baker, 2016; Rabinovitch & Abu-Baker, 2002).

The gaps between the Arab minority and Jewish majority in Israel are keenly evident noticeable in the public education system, specifically. For example, the Arab school system is characterized by larger classes and fewer teachers (Tatar & Horenczyk, 2003). Despite some improvement in Arab students’ educational attainments over the years, a significant gap remains between them and their Jewish counterparts (Gharrah, 2015; Hadad Haj-Yahya et al., 2018).

The Arab population is characterized by traditional, patriarchal, and authoritarian family values (Kaufman et al., 2012). In recent decades, however, Arab society in Israel has been undergoing rapid socioeconomic modernization (Gharrah, 2015). Among other things, this process leads to an intergenerational gap, with adolescents adopting Western values more willingly than their parents (Sherer, 2009), potentially resulting in parent–child conflict and reduced parental supervision (Szapocznik & Kurtines, 1993). In turn, these factors place children and adolescence at a higher risk for engagement in violence (Lee & Stockdale, 2008).

To sum, the current study will explore the association between individual, family and contextual factors and Arab youth perpetration of violence. Moreover, it will address the similarities and differenced between cyberbullying and in-person violence in the prediction of these socio-ecological factors.

Method

Participants

The sample represented Arab public education students in grades 7–11 in northern and central Israel, excluding the Bedouin population in the south that has its unique characteristics. Stratified probability sampling was used, with the stratum based on the Israel Central Bureau of Statistics’ (2015) index of municipal socioeconomic status (SES). Commonly used in Israeli studies, this index is informed by indicators that include income, employment, education level, and social benefits (see Gharrah, 2015). From each SES cluster containing Arab municipalities, up to 20% of Arab localities were randomly sampled. Next, 21 schools from these municipalities were selected randomly out of a list provided by the Ministry of Education. From each grade level at each school, two classes were selected and all their students were asked to participate.

This sampling procedure resulted in 3,178 participants (59.6% girls; ~ 93% response rate). The participants’ ages ranged from 11 to 18, with a mean of 14.88 (SD = 1.40). About half (48.2%) of them attended junior high schools (grades 7–9). The respective education levels of their mothers and fathers were 3.46 (SD = 1.13) and 3.21 (SD = 1.20), on a scale from 1 (elementary school) to 5 (BA or above).

Data Collection and Ethics

Guided by a research assistant, the participants completed a structured self-report survey in their classrooms. All were ensured anonymity and confidentiality. The survey questions and instructions and the students’ consent forms were approved by the Ministry of Education and by the internal review board of the Author’s university. Prior to the study, school principals sent letters and consent forms to the participants’ parents, informing them of the objectives of the study goals and giving them the option to decline participation on behalf of their children (their refusal rate was around 2%). The participants themselves were free to withdraw at any stage of the study, for any reason (their refusal rate was around 5%).

Measures

Cyberbullying

Involvement in cyberbullying was examined by eight items referring to online behaviors, without defining them explicitly as “bullying” (Mishna et al., 2012) (α = 0.91). The participants were required to indicate whether, over the past month, they were engaged in acts such as “Spreading harmful comments about someone in the internet,” on a scale from 0 (never) to 4 (everyday). One point was assigned for each specific behavior in which they had been engaged. The cyberbullying score was based on an accumulation of the items, and it ranged from 0 to 8.

Physical Violence

Involvement in physical violence was assessed by an adapted Arabic version of the California School Climate and Safety Survey (CSCSS; Furlong et al., 2005; Rosenblatt & Furlong, 1997), previously used in two Israeli national studies (Benbenishty, 2003; Benbenishty & Astor, 2005). The original survey was designed to measure reports of school violence victimization by students, including moderate and severe and moderate physical violence and verbal violence. In the present study, the items were adapted to assess perpetration of moderate and severe physical violence against others, not only students. The questionnaire included two subscales. In the first, five items measured perpetration of severe physical violence (e.g., “I attacked someone with the intent of severely hurting him/her”; “I attacked someone with a knife”) (α = 0.80). In the second subscale, four items were used to measure perpetration of moderate physical violence (e.g., “I threatened to hurt somebody”; “I attacked someone with my leg or hand”) (α = 0.74). In the current study, response scale was changed to a range of 0 (never) to 4 (more than ten times). The composite score for each type of violence was the mean of the item scores.

Indirect Violence

The perpetration of indirect violence was measured by four items (α = 0.67) taken from the Direct and Indirect Aggression Scale (DIAS; Björkqvist et al., 1992). This scale had previously been used among Jewish and Arab students in Israel (Benbenishty et al., 2000). The original scale measured perpetration of indirect violence against others (e.g., “I asked others not to play or communicate with someone”). In the present study, participants indicated the frequency in which they had perpetrated indirect violence in the past month, ranging from 0 (never) to 4 (more than ten times). The composite score was the mean of all items.

Impulsivity

Impulsivity was assessed by four items taken from the Teen Conflict Survey (Bosworth & Espelage, 1995), measuring the frequency of impulsive behaviors, including difficulty siting still, lack of self-control, and trouble completing tasks (α = 0.73). Respondents indicated how often they engaged in given behaviors on a scale ranging from 1 (never) to 5 (always). The final score was the mean of the item scores.

Religiosity

The scale developed by Pickering et al. (2011) was used to measure religiosity. It included 15 items addressing three sub-measures of the construct with five items each. Responses to the items of the first sub-measure, relationship (e.g., “Reading my faith’s book of truth helps me develop a bond with God”) ranged from 1 (never) to 5 (very often) (α = 0.96). The other second sub-measure was request (e.g., “I pray as often as I can”) (α = 0.90). The third was retribution (e.g., “God will make my life difficult if I misbehave”) (α = 0.88). For the latter two subscales, participants indicated their agreement with the items on a scale from 1 (strongly disagree) to 4 (strongly agree). In the adapted version used in the current study, questions referring to Christianity were rephrased to suit the Muslim faith. As all subscales loaded on a single factor, a composite religiosity measure was created by standardizing their values and calculating their mean score.

Parent-Adolescent Relationship

Three aspects of the parent-adolescent relationship were measured using the Adolescent Family Process scale (AFP; Vazsonyi et al., 2003): (1) Parental monitoring: five items, rated from 0 (strongly disagree) to 4 (strongly agree); e.g., “When I am not at home, my mother/father knows my whereabouts” (α = 0.76); (2) Conflict: three items, likewise rated; e.g. “How often do you get angry at your parents?” (α = 0.81); (3) Parental support: four items, rated from 0 (never) to 4 (very often); e.g., “My parents put me down in front of other people” (α = 0.70). An overall score of each aspect was derived by calculating the mean of its items.

Commitment to School

The commitment-to-school scale developed by Hirschi (1969; α = 0.78) included five items (e.g., “How important to you personally is getting good grades?”). The participants rated their responses on a scale ranging from 0 (never) to 4 (to a large extent). The scale score was based on the mean of the items.

Peer Delinquency

Peer delinquency was measured using an adapted version of the Social Network Assessment Questionnaire (Henry et al., 2001). Eleven items were used to measure the extent to which the adolescent’s close friends were involved in delinquency (α = 0.87), such as “My friend attacked someone with the idea of hurting her/him.” The items were measured on a five-point Likert scale ranging from 0 (never) to 4 (very often). The scale score was based on the mean of the item scores.

Neighborhood Violence

The participants’ exposure to violence in their neighborhood was measured using Selner-O’Hagan et al.’s (1998) My Exposure to Violence Scale (MYETV). We used its Arabic version, by Haj-Yahia et al. (2011). The scale provides six descriptions of violent events with regard to which the participants indicated whether they had experienced them in their community over the past month (e.g., “Someone beat you”), on a scale from 0 (never) to 3 (many times) (α = 0.83). The overall scale score was calculated based on the items’ means.

Data Analysis

Using SPSS 25, we examined the descriptive data on the participants’ involvement in each of the violence types examined in the study: cyberbullying, moderate and severe physical violence, and indirect violence. We also analyzed descriptive statistics of the independent variables. Next, bivariate analyses were used to test relationships between engagement in violence and the independent variables. The correlations of all the other variables were calculated as well (see Table 1). Third, we estimated a series of hierarchical multivariate regression models to predict each dependent variable: cyberbullying, severe and moderate physical violence, and indirect violence (see Table 2). We added the predictors to the regression sequentially, following a hierarchy from the most general context: demographics and individual characteristics to family factors and lastly included contextual factors.

Table 1 Correlations among the study variables (N = 3.178)
Full size table
Table 2 The findings of a multivariate hierarchal regression
Full size table

Results

Descriptive Statistics

The type of violent behavior most frequently reported by the participants was indirect violence (65.7%). Almost half (47.3%) reported perpetrating moderate physical violence at least once over the month prior to the study. More than a quarter (28.4%) reported having perpetrated serious physical violence. The least frequent violent behavior was cyberbullying, reported by 14.1% of the participants.

The mean of participants’ reports of impulsivity was 2.32 (SD = 0.92), on a scale from 1 (never) to 5 (always). The means of conflict with parent and perceived parental monitoring were 2.39 (SD = 0.99) and 3.50 (SD = 0.63) and, respectively, on a scale from 0 (strongly disagree) to 4 (strongly agree). The mean of perceived parental support was 2.83 (SD = 0.92), on a scale from 0 (never) to 4 (very often). The mean commitment to school was 3.16 (SD = 0.80), on a scale from 0 = never to 4 = to a large extent. More than half (52.5%) of the participants reported that their peers were involved in at least one delinquent act and almost half (46.6%) experienced violence in their neighborhood during the previous month.

Bivariate Analyses

As shown in Table 1, the bivariate relationships between the dependent and independent variables were in the same direction for all types of violence, except for age that was significantly and positively associated only with indirect violence. Also, while males reported cyberbullying and severe and moderate physical violence significantly more than did females, no gender differences were found with respect to indirect violence.

The results reported in Table 1 show that religiosity is associated negatively with all violence types. The more religious the participant, the lower their involvement in all violence types. It was also found that the higher the participant’s impulsivity the higher their reports of cyberbullying and violence. Stronger parent-adolescent relationships (higher monitoring and support and lower conflict) are associated with lower reports of cyberbullying and in-person violence. A similar trend was found with respect to commitment to school, which was significantly and negatively correlated with cyberbullying and violence. Finally, affiliation with delinquent peers and neighborhood exposure to violence were positively and significantly associated with all violence types.

Multivariate Regression

Table 2 presents the findings of a multivariate hierarchal regression for predicting in-person violence and cyberbullying. With respect to the individual factors included in the first stage of the model, it was found that while gender was insignificantly associated with cyberbullying, males reported higher involvement in severe and moderate physical violence and lower indirect violence. As expected, the more impulsive the participant, the higher their involvement in all types of violent behavior.

Family factors were included next in the model. Differences in the association of the three parental factors and the various violence types were found. While monitoring was significantly and positively associated with all violence types, except for indirect violence, lower conflictual relationships with parents were associated with lower in-person violence, but not cyberbullying. Furthermore, parental support was associated, significantly and negatively, only with moderate and indirect violence.

With respect to the contextual factors, the more committed the participants were to school the lower their involvement in both physical violent types, but not in cyberbullying and indirect violence. The results show that both affiliation with delinquent peers and exposure to community violence were risk factors for involvement in cyberbullying and in-person violence.

Note that the contextual factors explain the largest share of variance in participant violence (between 17.1% for moderate physical violence and 20.8% for cyberbullying), except for indirect violence, for which the largest share of variance was explained by the demographics and individual characteristics (12.6%). Overall, the predictors in this study explained considerably high shares of the variance in youth involvement in severe (40.5%) and moderate (41.2%) physical violence and cyberbullying (31.7%). A lower share of the overall variance was explained for indirect violence (22.4%).

Discussion

The current study was based on a large sample representative of Arab high-school students in Israel. It explored similarities and difference in the effects of a series of socio-ecological factors on both cyberbullying and in-person violence. The results showed that cyberbullying was significantly and positively associated with all in-person types of violence. The strongest association was between cyberbullying and adolescents’ reports of involvement in severe physical violence. These two behaviors were also the most similar in terms of the rates of youth reporting being involved in them.

One the one hand, these results support previous works showing that youth are still more involved in in-person violence than in cyberbullying (Mishna et al., 2012). On the other hand, the strongest association of cyberbullying was with severe violence. This finding might explain the perception of cyberbullying as severe, as reported by both students (Chapin & Coleman, 2017; Sharma et al., 2017) and parents (Mishna et al., 2020)—indicating that cyberbullying may be far more harmful than in-person bullying.

Individual Factors

Our present results support previous studies indicating that males are more physically violent than females (Björkqvist, 2018; Cleverley, et al., 2012). Furthermore, we found that females reported more frequently than did males on involvement in indirect violence (Marshall, 2011). However, no significant gender differences were found with respect to involvement in cyberbullying (see Hinduja & Patchin, 2008; Mishna et al., 2012). This finding might be explained by gender difference in Arab youth involvement in violence. Jeries-Loulou (2023) argued that in Arab culture, girls are subject to significantly more restrictions and control compared to boys—as well as compared to girls in non-traditional cultures—which might increase boys’ comparative involvement in violence and antisocial behavior. Conversely, indirect forms of violence that do not involve physical contact with others might attract less attention and be subject to less restrictions, possibly leading girls to use it similarly or even more than boys. This interpretation should be further examined in future studies.

As expected and in line with many previous works, it was found that the more impulsive the adolescents the higher their involvement in all types of in-person violence, including cyberbullying. As suggested by Moffitt (1993), impulsivity affects youths’ ability to control their behavior and to consider the consequences of their antisocial actions; therefore, impulsive youth who act with no forethought might be involved more than might others in cyberbullying and violence.

Family and Contextual Factors

The quality of parent-adolescent relationships is a key factor that may contribute to preventing involvement in cyberbullying (Khoury-Kassabri et al., 2016) and violence (e.g., Hamner et al., 2015; Massarwi, 2017; Moitra & Mukherjee, 2012; Pickering & Vazsonyi, 2010; Vazsonyi et al., 2003). The results of this study show that higher parental support and mentoring and lower conflictual relationships with parents are associated with lower involvement in moderate physical violence. Furthermore, lower levels of youth involvement in indirect violence were reported by youth who had more support and less conflict with their parents. Interestingly, and consistently with what was argued earlier with respect to the similarity between cyberbullying and severe physical violence, only these types of violence were associated with parental mentoring. These results may be understood in light of Hirschi’s (1969) social bond theory, whereby parental monitoring and supervision are among the factors considered helpful in developing parental control and restraining children from committing crime (Svensson, 2003). Still, the results emphasize that various violence types might be affected differently by parenting methods. These findings should be considered by professional who aim to develop an involvement program to deal with youth involvement in violence. These programs should consider the variation in the effects of parental factors on different violent types.

Interestingly, commitment to schools was only associated with direct violence, rather than with indirect violence and cyberbullying. These finding may also be interpreted in light of Hirschi’s (1969) social bond theory that states that youth commitment and motivation to succeed in conventional activities may decrease involvement in delinquency. As physical violence is easier to identify, especially by school staff, students might be warier of using it, while the other two violet types might be less captured by school personnel or even not viewed as the school’s responsibility, especially cyberbullying that often occurs outside school (Boulton et al., 2014). These results emphasize the need to increase the awareness of school staff of their responsibility. This includes providing school staff with the skills and tools to intervene in preventing all violet types, including those who are harder to capture, such as indirect violence and cyberbullying. The results also emphasize the importance of developing policies to distinguish between the different types of violence and address each risk factors in developing prevention and awareness programs.

One of the most consistent risk factors that increases the likelihood of involvement in violence is affiliation with delinquent peers. The results of the current study supported this direction, showing that adolescents with delinquent friends are more likely to be involved in both cyberbullying (Pepler et al., 2008) and in-person violence (Brook et al., 2011; Rios et al., 2020; Van Ryzin et al., 2012). These results are in line with social learning theories (Burgess & Akers, 1966) indicating that delinquent peers affect tendencies to violence and delinquency by reinforcing youths’ antisocial behavior, providing violent and delinquent models, and fostering beliefs conducing to these antisocial acts. This effect, as seen from the study’s results and previous works (e.g. Brook et al., 2011; Rios et al., 2020; Van Ryzin et al., 2012), is evident across different types of delinquent and violent acts, including cyberbullying.

A similar trend was found with respect to the effects of exposure to community violence and perpetration of violence toward others. Although only a few studies have focused on these effects with regard to cyberbullying, our results are consistent with previous studies on in-person violence, which found exposure to community violence to be associated with higher risk of adolescents’ involvement in antisocial behavior (Borofsky et al., 2013; Gorman-Smith et al., 2004; Hamner et al., 2015; Romero et al., 2015).

Limitations and Future Directions

This study has contributed to the understanding of factors related to cyberbullying among Arab youth in Israel. Nevertheless, it has three main limitations. Firstly, since a cross-sectional design was used, causal inferences can only be suggested. Future studies should use a longitudinal design or at least measure the variables of interest at several points in time. Secondly, adolescents’ self-reports were used to measure all variables. Future studies should collect data from additional sources, including parents and school data on violent incidents. Thirdly, as the present findings are based on a large representative sample of Arab students in Israel, they may be generalized to the population of Arab youth in Israel, but not to all Israeli adolescents, and definitely not to adolescents in other countries. Future studies should examine whether our findings can be generalized further by replicating this study in other populations in Israel and beyond.

Conclusion

Overall, our findings emphasize the similarities between cyberbullying and in-person and even severe physical violence. Both are similarly affected by many of the same socio-ecological factors including the adolescents’ characteristics, their relationship with their parents, and contextual factors, including delinquent peers and community violence.