Introduction
Bully victimization refers to the process by which an adolescent is repeatedly and over time exposed to intentional negative actions by their peers, and can include physical, verbal or relational aggression (Hamburger et al.
2011; Olweus
1996). Bully victimization can be distinguished from fighting or teasing by an imbalance in power between bully and victim (Olweus
1996). Bully victimization is distinguished from other forms of victimization because of a power difference between the perpetrator and the victim (Salmivalli and Peets
2009). Bully victimization is prevalent across countries worldwide, with an average of about 11 % of children reporting being bully victimized (Currie and Organization
2000). Bully victimization is a very powerful stressor in adolescence and can have long-lasting physical and psychological consequences (Arseneault et al.
2010). Recent work has shown that bullying and bully victimization should be understood as a group phenomenon (Salmivalli
2010). Besides the adolescents who bully and the victims of bullying, other peers are also involved in bullying by, for example, defending the victim or reinforcing the adolescents who bully (Salmivalli et al.
1996). Research also has begun to recognize that adolescents who belong to the same peer group might resemble each other in terms of how much they bully or are bully victimized by others (Espelage et al.
2007; Faris and Felmlee
2014; Huitsing et al.
2014; Sentse et al.
2013; Sijtsema et al.
2013).
Friends tend to be similar in a wide variety of behaviors (Brechwald and Prinstein
2011). These similarities can be due to selection or influence (Veenstra and Dijkstra
2011). Selection is the processes by which individuals choose friends who resemble themselves on certain characteristics. Influence is the processes that increase similarity between individuals once they have established a relationship (e.g., friendship) (Veenstra et al.
2013). Earlier research on children and adolescents examined selection and influence effects in the larger friendship network, taking into account all friendship ties within a school (Sentse et al.
2013; Sijtsema et al.
2013). One of these studies differentiated between relational victimization (e.g., being excluded) and overt victimization (e.g., being hit) (Sijtsema et al.
2013). Results indicated that adolescents who were relationally victimized tended to select friends who were similarly relationally victimized. Influence effects occurred for both relational and overt victimization. Other research found evidence for selection effects on overt victimization, and influence effects on relational victimization (Sentse et al.
2013). Thus, at the level of the larger friendship network, both selection and influence effects on bully victimization seem to occur, implying that not only individual characteristics of adolescents who are bully victimized are important for the development of bully victimization, but group processes play an important role (Salmivalli
2010). Indeed, many prevention and intervention programs aimed at reducing bully victimization work with peers, for instance, by creating a support group around children who are bully victimized, and show promising results (Ttofi and Farrington
2009).
Earlier research suggested that not all peers equally influence adolescents, and in some cases influence seems to be stronger in close friendships than more distant friendships in the larger friendship network (Giletta et al.
2012). If indeed peers who are closer to the adolescent have larger influence on their bully victimization, these peers may be important to target in bullying interventions. It is, therefore, important to examine whether selection and influence occur at different levels of the peer network, for instance by examining these effects at the level of the larger friendship network and at the level of closer friendships. Adolescents have multiple dyadic friendships that differ in their level of closeness (e.g., best friendships and close friendships). These friendships are interconnected in more complex friendship structures, such as groups, ultimately forming what can be referred to as the larger adolescent friendship network (Scholte and Van Aken
2006). Within these large networks of friendships, cliques can be defined as exclusive and relatively tight groups of friends with whom adolescents spend most of their time (Brown
2004; Brown and Klute
2006; Henrich et al.
2000). Compared to other peers that are more distantly connected to adolescents, clique members are considered to be among the most important sources of influence on adolescent development (Adler and Adler
1998; Brown and Klute
2006; Thompson et al.
2001; Witvliet et al.
2010b). A growing involvement in cliques occurs during the adolescent years (Thompson et al.
2001). Adolescents experience more affect, intimacy and self-disclosure with close friends, and during adolescence friendship groups such as cliques become more important for adjustment than in childhood (Giordano
2003). Because of the large role clique members play in adolescent development, researchers emphasized the importance of cliques when it comes to selection and influence processes (e.g., Conway et al.
2011; Ennett and Bauman
1994; Espelage et al.
2007; Paxton et al.
1999). Within the larger friendship network, cliques have their own social norms, and may thus have a different influence on adolescents compared to other peers in the larger friendship network (Urberg et al.
1995). Earlier research indicated that dyadic best friends may be more important compared to other peers in the larger friendship network (Giletta et al.
2012). However, because dyadic friendships usually do not occur in isolation, but are embedded within cliques, examining the added effect of all clique members is crucial (Bagwell et al.
2000; Espelage et al.
2007). Earlier research has not yet examined selection and influence processes involved in bully victimization at the level of the clique. But, earlier research showed that selection and influence processes at the level of the clique play a role for the perpetration of bullying (Espelage et al.
2007; Witvliet et al.
2010a). In addition, there is evidence that clique members resemble each other in their levels of bully victimization, but it is unknown whether this similarity is related to selection or peer influence processes (Salmivalli et al.
1997). Thus, despite the possible importance of cliques in selection and influence processes, research in this regard is limited to the level of the entire friendship network (Sentse et al.
2013; Sijtsema et al.
2013). Therefore, we examined selection and influence processes related to bully victimization within both the larger friendship network and at the level of the clique.
Selection and Influence Regarding Bully Victimization
Although victims of bullying are usually low in peer acceptance (de Bruyn et al.
2010; Scholte et al.
2009), there may be reasons why adolescents could select friends who are bully victimized. First, adolescents who are bully victimized themselves presumably have the need to form intimate relationships like anyone else (Baumeister and Leary
1995). Befriending others who are bully victimized might be their only option, and might be considered a default friendship choice (cf. Scholte et al.
2009; Sijtsema et al.
2013). That is, because adolescents who are bully victimized hold a relatively marginal position at school, the pool of possible peers that they can establish friendships with consists largely of peers with the same social position. Second, adolescents who are bully victimized might select friends that are also bully victimized by deliberate choice. Selecting friends who are bully victimized may be uniting (Salmivalli et al.
1997) as victims may feel understood and supported by a friend who has similar experiences. In addition, when a group perceives they are bully victimized, this may increase trust within that group, hence victims of bullying may feel connected to others who are also bully victimized, and joining their clique may be beneficial (Rotella et al.
2013). Another benefit of befriending others who are bully victimized may be that adolescents who are bully victimized are more willing to intervene against bullying, and adolescents who are bully victimized by the same perpetrators tend to defend each other (Batanova et al.
2014; Huitsing et al.
2014). Overall, we thus hypothesize that adolescents select their friends based on their level of bully victimization. We expect selection to occur both at the level of the larger friendship network and at the level of the clique.
On the one hand, selecting friends who are bully victimized may thus be beneficial. On the other hand, it can pose a risk as well. Because bully victimized adolescents tend to have poor social and emotion regulation skills as well as higher levels of psychopathological symptoms (both internalizing and externalizing), friendships between adolescents who are bully victimized might not be as beneficial as other friendships (see Prinstein and Giletta, in press). In this regard, research has shown that friendships of bully victimized adolescents have lower positive qualities and involve higher levels of conflict (Bagwell and Schmidt
2011), which eventually might lead these relationships to be short-lived (see Sijtsema et al.
2013). More importantly, these friendships might further maladaptive cognitions (e.g., self-blame, negative attribution styles) and symptomatology associated with bully victimization (Prinstein and Giletta, in press), and ultimately exacerbate the likelihood of experiencing bully victimization (via influence processes) as well.
Influence processes also may lead to similarities between friends in general, or clique members specifically, in their level of bully victimization. Friends who are bully victimized may not be adequate protectors against bullying, but rather serve as a risk factor for future bullying. One reason why friends who are bully victimized may be a risk factor for future bully victimization, is that having friends who are bully victimized may reduce opportunities to learn adequate social behavior and increase maladaptive behavior. Being bully victimized is related to low social skills (Schwartz et al.
1993). Friends who have low social skills may not be able to serve as role models of competent social behavior that might protect adolescents from bully victimization (Scholte et al.
2009). Instead, maladaptive behavior could be reinforced, creating a negative circle of maladaptive behavior (cf. cumulative continuity; Caspi et al.
1989).
Another reason why friends who are bully victimized could pose a risk for future bully victimization is that social contagion of bully victimization might occur. Joining a group with a certain social status may result in obtaining that social status as well (Peters et al.
2010; Witvliet et al.
2010a). This idea of social contagion may also apply to bully victimization. A group of adolescents who are bully victimized may hold a low social status, and have the reputation of not being able to defend themselves adequately. This may lead to acquiring a similar social position, and ultimately to an increased risk of bully victimization. Studies indeed suggest that having friends who are unable to protect against bullying, or receiving peer nominations from others who are bully victimized are risk factors for bully victimization (Hodges et al.
1997; Pellegrini et al.
1999). In addition, longitudinal friendship network analyses suggests that having a friends who is bully victimized and defending victims increase the likelihood of becoming bully victimized (Faris and Felmlee
2014; Huitsing et al.
2014). Thus, we expect that influence in terms of bully victimization would occur both at the level of the larger friendship network, and at the level of the clique.
Gender Differences
A large body of research suggests that bullying processes might be different for boys and girls (e.g., Bjorkqvist et al.
1992; Veenstra et al.
2005). Selection and influence of bully victimization in cliques may be different for boys and girls as well. Female victims of bullying usually have a broader friendship network than male victims of bullying, in the sense that their network does not just consist of victims or otherwise rejected adolescents (Salmivalli et al.
1997). Also, girls typically show more willingness to intervene against bullying than boys, and take up the role of defender of victims more often (Batanova et al.
2014; Salmivalli et al.
1996). In addition, boys are more similar than girls in their level of bully victimization (Hodges et al.
1997). This indicates that girls may not take victim status into account in forming friendships as much as boys do. Girls also may have the opportunity to practice a larger range of social skills as their cliques are more diverse and include peers who might be able to defend against bullying. Therefore, we hypothesize that both selection and influence of friends based on the level of bully victimization will be stronger for boys than for girls.
Discussion
Earlier research emphasized the importance of group processes in bully victimization (Salmivalli
2010). Adolescents reporting bully victimization may actively select friends who are also bullied, because they are the default choice (cf. Scholte et al.
2009; Sijtsema et al.
2013), or by deliberate choice (Huitsing et al.
2014; Salmivalli et al.
1997). In addition, adolescents may influence their friends’ levels of bully victimization over time, because friends who are bullied may not provide opportunities to practice social skills needed to defend against bullying (Scholte et al.
2009), or because of social contagion of bully victimization status (Faris and Felmlee
2014; Huitsing et al.
2014). Indeed, earlier research showed that selection and influence processes play a role in bully victimization at the level of the larger friendship network (Sentse et al.
2013; Sijtsema et al.
2013). Earlier research also indicated that different types of friendships exist within the larger friendship network, and that, in some instances, closer friends may be of larger influence than more distant relations (Giletta et al.
2012). Cliques may be especially important, because they encompass close friendships, and clique members are amongst the most important peers for adolescents (Bagwell et al.
2000; Conway et al.
2011; Espelage et al.
2007; Thompson et al.
2001).
Our study was the first to examine selection and influence processes involved in bully victimization at the level of the larger friendship network, and at the level of friendship cliques. We hypothesized that selection and influence of bully victimization would occur both levels, and that selection and influence of bully victimization at the clique level would be stronger for boys than for girls. Indeed, in the larger friendship network there was evidence that adolescents select friends based on their level of bully victimization, and influence the degree to which their friends’ levels of bully victimization over time. Contrary to expectations, at the clique level adolescents did not select their clique members on the basis of these members’ levels bully victimization. For boys, we found evidence suggesting that the average level of victimization in a clique influences the level of individual bully victimization over time. This influence effect was not found for girls at the level of the clique. In addition, whereas we found that average clique levels of bully victimization influenced future individual levels of bully victimization, we did not find that adolescents’ individual levels of bully victimization influenced average clique levels of bully victimization.
Our findings for the larger friendship network replicated earlier findings indicating that both selection and influence processes account for similarities between friends’ levels of bully victimization (Sentse et al.
2013; Sijtsema et al.
2013). Thus, in general adolescents tend to befriend others with similar levels of bully victimization, and they tend to become more alike in bully victimization over time. Regarding cliques, cross-sectional studies suggest that both selection and influence processes are responsible for similarities between clique members’ level of bully (Salmivalli et al.
1997). Our findings, using two time points, indicate that similarity between clique members’ bully victimization may not be due to selection. Thus, although selection was observed within the larger network that included all friendship ties, including for instance less close relations with friends who were not in the same clique, such selection effects did not hold for cliques. As cliques consist of relatively close friendships with whom adolescents spend most of their time (Brown
2004; Brown and Klute
2006; Henrich et al.
2000), selection of friends who are bully victimization to the same extent may thus not be due to the selection of one’s closest friends, but rather seem to reflect a tendency to select friends from a larger pool of friends that hold a similar social status. Thus, whereas adolescents’ selection of any friend within their grade may be influenced by their bully victimization levels, this does not necessarily apply to the selection of their closest friends. This is an important finding as it counteracts the idea that adolescents who are bullied may actively and deliberately select very specific social niches that pose risks for prolonged bullying. Future research could explore this idea by examining selection and influence effects at the level of the best friend, and by combining research on selection and influence processes involving bullying and social status.
We found evidence for peer influence regarding bully victimization in the larger friendship network, in line with earlier findings (Sentse et al.
2013; Sijtsema et al.
2013). At the clique level, our findings indicated peer influence regarding bully victimization only for boys. For boys, the average level of clique members’ bully victimization predicted predicted individual levels of bully victimization over time. Thus, for boys social contamination processes seem to occur that increase the likelihood of bully victimization by associating with other victims. A reason for this may be that friends of adolescents high in bully victimization acquire a similar social position and are seen as easy targets who are not likely to retaliate successfully against harassment (Hodges et al.
1997; Witvliet et al.
2010a,
b). In addition, peers with high levels of bully victimization may be inadequate role models who cannot help to acquire the social skills needed to defend against bully victimization and may instead reinforce socially maladaptive behavior (Scholte et al.
2009).
One crucial point to be addressed in future research is the question when peer groups become a risk for bully victimization and when they serve a protective function. Our findings indicate that if a clique consisting of adolescents who are not bullied are joined by one adolescent high in bully victimization, this does not seem to increase the likelihood that the other clique members will become the victim of bullying as well. At the same time, if a adolescent who reports low levels of bully victimization is part of a clique with high levels of bully victimization this does not provide protection for the entire clique. This finding further stresses the importance of incorporating groups in our understanding of adolescent bully victimization. Group factors (i.e., whether adolescents’ clique members are bully victimized of not) are of great importance for the future bully victimization status of individuals, whereas individuals do not influence the clique as much.
We only found evidence for influence of clique members in stable cliques, that is, in cliques that were still present in the same composition at T2, and not in cliques that were dissolved at T2. This is in line with findings on dissolved friendships in earlier research (Laursen et al.
2012). A reason why cliques break up might be the extent to which some of their members are bully victimized, as bully victimization is related to de-selection of friends (Sijtsema et al.
2013). Adolescents may be aware of the risk of being part of a clique characterized by high levels of bully victimization, and decide to diminish this risk is by leaving the clique. In addition, clique members may exclude specific others in their clique who they perceive are high in bully victimization. As Bukowski and Sippola (
2001) suggested, peer groups have goals such as group cohesion and homogeneity. Adolescents high in bully victimization may jeopardize these group goals, for example by threatening cohesion because other members experience increased risk of bully victimization. Excluding this clique member may thus be beneficial for the clique. Although such processes have been proposed for aggression and the perpetrators of bullying (Garandeau and Cillessen
2006), development of cliques in relation to bully victimization needs to be addressed in future research.
Regarding gender differences, we confirmed our hypothesis that influence effects are stronger for boys than for girls. In fact, we found no evidence for peer influence processes relating bully victimization in adolescent cliques for girls at all. This might be due to the differences in networks between boys and girls. Girls have more diverse networks in terms of bully victimization than boys; for instance, it is possible for girls to have both perpretrators and victims of bullying in their network (Salmivalli et al.
1997). This indicates that the proposed processes might not be as apparent for girls as for boys. Girls may have different role models in their clique (i.e., not just adolescents with low social status or low social skills), so the negative cycle of maladaptive behavior might not occur. Moreover, girls’ clique members might be more able to defend each other than boys’ clique members. Indeed, research suggests that girls are more likely to take up the role of defender in bullying situations than boys (Batanova et al.
2014; Salmivalli et al.
1996).
This gender difference also has implications for prevention and intervention programs against bullying. In recent programs, peers have been used to prevent or intervene against bullying, for instance by providing support groups for victims of bullying (Ttofi and Farrington
2009). For boys, clique members may only be effective against bully victimization if these clique members are low in bully victimization themselves. If boys have clique members who are all high in bully victimization, selecting other peers from the larger friendship network in a support group may be a better strategy. For girls, clique members do not seem to influence the degree to which adolescents are bullied. For girls, having clique members who are high in bully victimization is not a risk factor, but having clique members who are low in bully victimization is not a protective factor either. Future research could examine whether it is beneficial for girls to train clique members to intervene against bullying, or whether only peers from the larger friendship network, such as popular classmates, are effective in reducing bullying (Faris and Felmlee
2014). Moreover, because we found evidence for selection and influence at the level of the larger friendship network, our results imply that adolescents may select friends from a larger pool of friends that hold similar social positions, and that they may be influenced by others in the same general social group as well. This emphasizes the importance of social position for interventions against bullying.
A major strength of the present study is that we examined selection and influence regarding bully victimization at the level of the entire friendship network and at the level of cliques. This allowed us to obtain a more in depth view of where selection and influence processes involved in bully victimization for adolescents. Moreover, our study is the first to examine selection and influence processes for bully victimization in the context of cliques using multiple time points.
Despite its strengths, this study also had some limitations. First, we did not include peer reported bully victimization in our design. One of the mechanisms we proposed behind clique members’ influence on individual levels of bully victimization is social contamination, which means that adolescents might become perceived by peers as victims of bullying when their clique members are bullied. Future research should include peer reported bully victimization to explore this option. Second, we did not differentiate between various forms of bully victimization, whereas earlier research suggested that selection and influence processes may differ for overt and relational bully victimization (Sijtsema et al.
2013). Future research should examine whether such differences hold for selection and influence processes in cliques as well. Third, we only included two time points that were 5 months apart. Although this is a relatively short interval, we cannot establish what changes in the network may have occurred between the two time points. For instance, cliques that appear to be stable may have been broken up for a while. Although the method we used to examine peer influence is common (Popp et al.
2008), we cannot be certain that similarities between clique members’ levels of bully victimization are actually due to influence processes, rather than selection processes, for boys. Only stochastic actor-based modeling allowed to clearly disentangle selection and influence effects, because in this approach unobserved changes between discrete observations were simulated (see Steglich et al.
2010) (see Steglich et al.
2010). Future research should thus include more time points and shorter intervals.