Bullying Prevention in Adolescence: Solutions and New Challenges from the Past Decade.

Bullying among youth at school continues to be a global challenge. Being exposed to bullying may be especially hurtful in adolescence, a vulnerable period during which both peer group belonging and status become key concerns. In the current review, we first summarize the effectiveness of the solutions that were offered a decade ago in the form of anti-bullying programs. We proceed by highlighting some intriguing challenges concomitant to, or emerging from these solutions, focusing especially on their relevance during adolescence. These challenges are related to (1) the relatively weak, and highly variable effects of anti-bullying programs, (2) the complex associations among bullying, victimization, and social status, (3) the questions raised regarding the beneficial (or possibly iatrogenic) effects of peer defending, and (4) the healthy context paradox, that is, the phenomenon of remaining or emerging victims being worse off in contexts where the average levels of victimization decrease. We end by providing some suggestions for the next decade of research in the area of bullying prevention among adolescents.

Bullying is a pervasive global problem that has attracted researchers’ attention for five decades. It is typically defined as repeated, intentional hurting of a person who is weaker or less powerful than the perpetrator(s) (e.g., Olweus, 1978; Salmivalli & Peets, 2018). Bullying can be direct, such as physical or verbal attacks, indirect (also referred to as relational bullying), such as social exclusion and rumor‐spreading, or it can happen online.

Although systematic bullying of selected peers already exists at a young age (e.g., Perren & Alsaker, 2006), adolescence is a period when the importance of peers—in terms of both belonging and prominence in the peer group—is pronounced (LaFontana & Cillessen, 2010; Newman & Newman, 2001). It is also a period of elevated risk for the onset of anxiety and depression, and peer victimization is predictive of both (Stapinski, Araya, Heron, Montgomery, & Stallard, 2015). Being bullied affects the development of personal identity as well (van Hoof, Raaijmakers, van Beek, Hale, & Aleva, 2008)—a key developmental task of adolescence. Victimization can be even directed at one’s (ethnic, sexual, etc.) identity: bias‐based bullying targeted at youth with marginalized identities is common (Galán, Stokes, Szoko, Abebe, & Culyba, 2021). Peer‐victimized adolescents have often experienced victimization for a long time (Troop‐Gordon, 2017), are less likely than younger children to tell adults about it (Blomqvist, Saarento, & Salmivalli, 2020), and have little trust in adults’ capability to help solve the problem (Elledge et al., 2013). Aggressors tend to become more popular (Cillessen & Borch, 2006; Cillessen & Mayeux, 2004) and peer group norms are increasingly approving of bullying (Salmivalli & Voeten, 2004) in (pre)adolescence. Peer‐victimized adolescents’ plight certainly deserves attention.

The present review focuses on bullying and victimization among (pre)adolescents in the school context, with an eye to prevention and intervention. Rather than systematically reviewing research from the past decade and ending with implications for interventions, we start the current paper with a summary of the effectiveness of the solutions that were offered a decade ago (i.e., implementing school‐based anti‐bullying programs) and then selectively focus on a few timely topics that represent either enduring challenges in the field of bullying prevention, or new challenges raised by the solutions that have been developed and implemented. Although research on cyberbullying has grown exponentially, we chose to exclude it from the current review, due to the massive amount of research and reviews recently published on the topic (including a “mapping review of systematic reviews” (Kwan et al., 2020). Also, aggression and victimization in the romantic/dating context (Wincentak, Connolly, & Card, 2017) are beyond the scope of the present article. Finally, the decade ended with the covid‐19 pandemic, which affected many aspects of adolescents’ lives, including schooling and social contacts overall. This bears consequences for peer relations, including bullying and victimization. Studies have only started to emerge around this topic, and it is not the time to review them, yet.

The decade started with (at least modest) optimism regarding bullying prevention, which was soon mitigated by findings showing that not all youth benefited from school‐based prevention and intervention. First, adolescents, in particular, did not respond to interventions in the hoped‐for manner. Second, interventions continue to face challenges in reducing bullying in part related to the associations among bullying perpetration, victimization, and peer status that are not fully understood. Third, some interventions were based on elements involving unforeseen complexities; in our review, we ask whether peer defending, an important component of many anti‐bullying programs, might have some iatrogenic effects—either for the defenders themselves or for the victimized students they defend. Fourth, we review evidence regarding the healthy context paradox: even in the context of successful interventions, youth who remain victimized (or become new victims) might be even worse off than victimized students in contexts where the overall level of victimization remains high.

Our review concerns bullying prevention, rather than the prevention of aggression in general. When we refer to studies that looked at aggression more generally, we explicitly say this.

PREVENTION AND INTERVENTION

With respect to bullying prevention, the past decade started with good news. The largest‐so‐far meta‐analysis on the effects of school‐based anti‐bullying programs was published (Ttofi & Farrington, 2011; based on Farrington & Ttofi, 2009) and concluded that such programs are, on average, effective. The programs led to significant average reductions in the proportion of students who were bullied, as well as those bullying others. Other reviews and meta‐analyses echoed this finding (Fraguas et al., 2021; Jimenes‐Barbero, Ruiz‐Hernández, Llor‐Zaragoza, Pérez‐García, & Llor‐Esteban, 2016; Lee, Kim, & Kim, 2015; Ng, Chua, & Shorey, 2020). Some of them included—in addition to the prevalence of perpetrators and victimized youth—outcomes such as attitudes about bullying, school climate, and mental health problems (Fraguas et al., 2021; Jimenez‐Barbero et al., 2016) or bystander intervention (Polanin, Espelage, & Pigott, 2012). The positive effects were welcome, as previous reviews (Ferguson et al. 2007; Merrell, Gueldner, Ross, & Isava, 2008) had made more pessimistic conclusions regarding the effectiveness of anti‐bullying programs.

The good news were, however, accompanied with bad news: the average effects were disappointingly modest and there was large variation in effects across studies and programs. Both conclusions are still valid today. The updated meta‐analysis by Gaffney, Ttofi, and Farrington (2019), now including 100 program evaluations—almost twice as much as in the original work by Farrington and Ttofi—showed that the average reduction in the prevalence of bullying perpetrators and victims was 19–20% and 15–16%, respectively. These prevalence changes are slightly smaller than the ones reported 10 years earlier (20–23% and 17–20% for bullying and victimization, respectively). For adolescents, the conclusions might be even more pessimistic.

Effects of School‐Based Prevention and Intervention among Adolescents

An important debate from the past decade concerned the effects of anti‐bullying programs on adolescents versus younger children. Based on theoretical considerations and empirical findings, Smith (2010) argued that adolescents are more difficult to influence with school‐based interventions. More evidence was emerging; for instance, the KiVa anti‐bullying program was evaluated in all grade levels in Finnish elementary and middle schools (age range of 7–15 years), not only once but twice: in a massive randomized controlled trial (RCT) involving almost 30 000 students (Kärnä et al., 2013; Kärnä, Voeten, Little, Poskiparta, Kaljonen, et al., 2011) and during the first year of nationwide implementation using a cohort‐longitudinal design (Kärnä, Voeten, Little, Poskiparta, Alanen, et al., 2011). The findings consistently showed the weakest effects among 13‐ to 15‐year‐old middle school students.

In light of the above, it was somewhat surprising that the first Ttofi and Farrington meta‐analysis (2011) found larger effects among older (>11 years) students, suggesting that programs work better among adolescents than among younger children. The age effect found in another meta‐analysis, including only RCTs (Jiménez‐Barbero et al., 2016) was opposite to that of Ttofi and Farrington (2011): the interventions had a significantly greater impact on bullying perpetration among children younger than 10  years.

Yeager, Fong, Lee, and Espelage (2015) pointed out that previous analyses comparing program effects across age groups were all between‐study (and thus, between‐program) tests of moderation. A more correct strategy would be to analyze within‐study moderation of efficacy by age (comparing the effects of the same programs in various age groups) to estimate age trends in responsiveness to interventions. Their multilevel meta‐analysis using this approach showed that bullying behavior was effectively reduced in youth younger than 14 years, after which the effect sizes dropped sharply.

Besides testing the effects of school‐based multi‐component programs, some studies looked specifically into indicated interventions targeting students who had been directly involved in bullying (i.e., teachers or other school adults addressing the bullying case by discussions with students). The first study comparing the effects of such discussions across school levels (Garandeau, Poskiparta, & Salmivalli, 2014) found them equally effective in elementary and middle schools and across different forms of bullying. In this study, however, success (whether bullying had stopped) was reported to school adults in a follow‐up discussion 2 weeks after the intervention. Based on a large data set collected with anonymous annual questionnaires over6 years in more than 1,200 schools implementing the KiVa program in Finland, Johander, Turunen, Garandeau, and Salmivalli (2020) found that the success rate of intervention discussions—a component of the program—was higher in elementary school grades (students with 10–12 years of age) than in middle schools (students with 13–15 years of age). This effect was consistently present, whether success was reported by the teachers doing the intervention or the students who had been victimized. Another study (E. Johander, T. Turunen, J. Trach, C. Garandeau & C. Salmivalli, unpublished data) found that being in a higher grade did in and of itself make it less likely that bullying stopped after an intervention, regardless of how long it had lasted. This time, the finding was confirmed by both victim and perpetrator reports.

Program Components Contributing to Effects: “What Works?”

In the past 10 years, researchers started to pay attention to disentangling the “effective ingredients” of anti‐bullying programs from less effective (or even iatrogenic) ones. Ttofi and Farrington (2009) took an important first step in this direction. They coded the whole‐school anti‐bullying programs involved in their meta‐analysis with respect to whether or not they included 20 components, and investigated the association between program effects and the presence of each component. These efforts were followed by other meta‐analyses focusing on effective components (Huang et al., 2019; Jimenez‐Narbero et al., 2016; Ng et al., 2020).

The most recent work in this area, also including the largest number of studies, is the updated meta‐analysis by Gaffney, Ttofi, and Farrington (2021). The findings indicated that two components were effective in reducing the prevalence of bullying as well as victimization: information provided for parents, and informal peer involvement. In addition, several components were effective in reducing bullying perpetration only: whole‐school approach, anti‐bullying policies, classroom rules, and work with victims. There was no association between the effectiveness of a program and the number of intervention components included in it, contrary to what was found in the 2011 analysis.

With respect to the role of parental involvement, Gaffney et al. (2021) concluded that providing information for parents is more effective than involving parents in meetings and discussions. However, another recent meta‐analysis by Huang et al. (2019) synthesized evidence regarding the effects of bullying prevention programs involving a parental component (e.g., information meetings, workshops for parents, or communication sent home). They found that these programs were overall effective in reducing bullying and victimization, and their effects were not moderated neither by the degree of parental involvement nor by school level. It is worth noting that the programs included many other components besides parental involvement, and the authors did not compare the effects with those of programs not involving parents (as in Gaffney et al., 2021).

“Work with peers”, which had originally been found counterproductive (Ttofi & Farrington, 2011), was in the more recent analysis by Gaffney et al. (2021) divided into three components which were coded as present or absent in the programs. These were “informal peer involvement” (small group or whole‐class discussions and activities related to bullying experiences and attitudes, thus targeting bystanders indirectly rather than directly), “formal peer involvement” (such as peer‐led anti‐bullying activities, peer‐mentoring schemes, or training students to provide active support to participants experiencing bullying), and “encouraging bystanders”. The findings now suggested that informal peer involvement was the only of the three components that contributed significantly to reductions in both bullying perpetration and victimization. It is not completely clear, however, when a component was coded as “informal peer involvement” versus “encouraging bystanders”. The names of the categories may be misleading, as many anti‐bullying programs aim at mobilizing bystanders to behave constructively when witnessing bullying (not reinforcing bullying but rather supporting the peers who are victimized) exactly by whole‐class discussions and activities around this topic, that is, through informal peer involvement.

It is not surprising that the findings of the 2011 and 2021 meta‐analyses were somewhat different; the more recent one included a theoretically better justified, more precise coding and a substantially larger number of studies. The differences in results, as well as discrepancies in the findings from other meta‐analytic studies (e.g., Huang et al., 2019; Jimenez‐Narbero et al., 2016; Ng et al., 2020) show, however, that the search for the most effective components is still at an early stage and clear policy implications may not yet be warranted.

Overall, although analyses on effective components are an important step toward understanding “what works”, it involves several pitfalls. First, analyses are correlational, and thus inferring causality is a stretch. Experimental work testing the effects of different components is needed, but also very costly. Second, researchers are coding components that are included in program manuals, rather than components that were actually implemented by schools. Some components may not have been implemented at all, or perhaps were being implemented in rare cases (e.g., “disciplinary methods” targeted at very few identified bullies; anti‐bullying lessons delivered in a limited number of classrooms) or implemented differently than recommended (Johander et al., 2020). Third, some components may only work, or work better, in the presence of other component(s), and the examination of unique contributions of components might hide such interactive effects. Fourth, and most relevant for the present article, findings regarding the components associated with larger effect sizes may radically change when the effects of different components are examined in various subgroups, for instance, young children versus adolescent samples.

Altogether different strategies might be needed when preventing and intervening in bullying among adolescents, as compared with younger children. For instance, bullying is likely to be increasingly driven by the need to gain or demonstrate peer status in adolescence; yet, many prevention programs focus on factors, such as social‐emotional learning or theory of mind skills, which may not play a major role in bullying during this developmental period (Yeager et al., 2015). In addition, the effects of parental and peer involvement in bullying prevention might well be age specific. It is conceivable that while parental involvement is a key in early and middle childhood, involvement of peers turns out more effective among adolescents, due to the role of peer bystanders in bullying (Pouwels, Lansu, & Cillessen, 2016), adolescents’ heightened sensitivity to peer feedback (Albert, Chein, & Steinberg, 2013), as well as their tendency to mimic the aggressive behavior of their popular peers (Juvonen & Ho, 2008). So far, effective components have not been studied separately in different age groups, and the decisions regarding relevant components to include have not been developmentally informed.

What (Might) Work Among Adolescents

Thus far, most of the available evidence seems to support the view that adolescents are less responsive to school‐based interventions (whether preventive programs as a whole or their specific, targeted components) than younger students. There are both developmental and contextual factors that might explain this trend.

Developmental changes include puberty‐related hormonal changes and maturation of the brain. Increased testosterone levels, for instance, have been associated with a higher activation of the neural systems regulating reward and social motivation (Murray‐Close, 2012). Brain areas processing emotional experiences change more rapidly than those mediating cognitive regulation (Albert et al., 2013). All these changes may contribute to adolescents’ self‐focus, risk‐taking, and sensitivity to peer feedback, as well as to their status needs and socially dominant behaviors (Albert et al., 2013; Murray‐Close, 2012).

With respect to contextual changes, adolescents transfer from small elementary schools to larger middle schools. The transition involves re‐shuffling of the peer landscape, which may create increasing concerns about peer relationships and status. Although empirical evidence of the effect of middle school transition on bullying is mixed (Farmer, Hamm, Leung, Lambert, & Gravelle, 2011; Pellegrini, 2002; Pellegrini et al., 2010), and difficult to disentangle from developmental changes, the transition is often assumed to influence the prevalence as well as prevention of bullying. Besides changes in classroom compositions, friendships, and peer group dynamics, the transition brings along a more complex organizational structure and changing role of teachers (who may feel less responsible for anti‐bullying work in middle, as compared with elementary school)—such changes might make the implementation of anti‐bullying programs especially demanding.

Adolescence is also characterized by strains in relationships with adults, especially those in authority positions. Attempts to control youth’s behavior may, therefore, lead to psychological reactance against adult injunctions to think or behave in a certain way, leading even to behaviors opposite to the adult suggestions. Teacher‐led lessons or suggestions to change one’s behavior may indeed be counterproductive among adolescents. Yeager et al. (2015) suggest using autonomy‐supportive approaches and language, rather than direct injunctions (e.g., “you might” instead of “you should”). They provide an example from smoking prevention: campaigns directly telling youth they should not smoke are less effective than campaigns portraying tobacco companies as the “authority” seeking to manipulate adolescents, and non‐smoking youth as rebellious for standing up to them. Another example, also from smoking prevention, is the ASSIST program, where influential students are identified and trained to act as peer supporters during informal interactions, encouraging their peers not to smoke. The intervention successfully reduced the likelihood of being a smoker among 12‐ to 13‐year‐old adolescents.

The idea of avoiding adult‐imposed rules or injunctions to behave in certain ways has also been applied in the context of preventing conflict, violence, and bullying. One approach is to identify highly influential students who will generate solutions and initiatives, make these solutions visible to others, and take a public stance against problem behaviors (Paluck, Shepherd, & Aronow, 2016). Bowes et al. (2019) utilized students as change‐makers in their schools in the ROOTS Indonesia adolescent bullying intervention, where highly influential students led activities and sessions around themes, such as violence prevention and constructive bystander behaviors. The intervention also included a teacher‐training component. The program was considered feasible, but the effects on bullying and victimization were mixed.

Another example of mobilizing peers in bullying prevention is the Italian NoTrap! program, which has produced promising effects on both traditional and cyberbullying among 14‐ to 15‐year‐old adolescents (Palladino, Nocentini, & Menesini, 2016). After an adult‐led phase, trained peer educators lead both online and offline activities around empathy, problem‐solving, and bullying, including positive bystander behaviors. The Meaningful Roles intervention (Ellis, Volk, Gonzalez, & Embry, 2015), currently under evaluation in the Netherlands, acknowledges adolescents’ goals for status and attempts to work with, rather than against, such goals in preventing bullying. Students are assigned to roles in which they can fulfill their status goals in prosocial ways, taking social responsibility—this is expected to reduce bullying behaviors, as it provides youth with alternative ways of gaining status.

Moderators of Prevention Effects

Research on moderators of anti‐bullying program effects is scarce in adolescent samples. With respect to gender, a recent meta‐analysis suggested that anti‐bullying programs overall work slightly better among boys than among girls (Kennedy, 2020). Among adolescents in particular, there is some indication of stronger anti‐bullying program effects among boys, as well as in classrooms with a larger proportion of boys (Kärnä et al., 2013), but also studies that found no gender moderation of effects between (pre‐)adolescent boys and girls (Gradinger, Yanagida, Strohmeier, & Spiel, 2015; Palladino et al., 2016). Research on other moderators consists of single studies (mostly done in middle school/preadolescent samples) that await replication. They have looked, for instance, at environmental sensitivity (Nocentini, Menesini, & Pluess, 2018), temperament (Nocentini, Palladino, & Menesini, 2019), and popularity among peers (Garandeau, Poskiparta, et al., 2014).

Building on the frameworks of differential susceptibility and vantage sensitivity tested whether individual differences in environmental sensitivity moderated the effects of a bullying prevention program. The study was conducted in the context of a large (> 2000 fourth and sixth graders) RCT of the KiVa program in Italy. The findings provided some evidence that highly sensitive children (especially boys) benefited more from the intervention, showing larger decreases in victimization. In a sample of sixth graders (12‐year‐old early adolescents) from the same trial, Nocentini et al. (2019) examined whether temperament moderated the effects of the program on bullying perpetration and victimization. Both effortful control and negative emotionality moderated the effects of the KiVa program on bullying perpetration: effects were only seen among youth with high effortful control and those with low and medium levels of negative emotionality. Positive emotionality, on the other hand, strengthened the intervention effects on victimization.

Garandeau, Lee, and Salmivalli (2014) categorized 10‐ to 12‐year‐old preadolescents into three groups (low, moderate, and high) on the basis of their perceived popularity among classmates. They found that popularity moderated the effect of the KiVa program on students’ peer‐reported bullying behavior, such that intervention effects were only found among low‐ and medium‐popular, but not among the highly popular students. This finding might be particularly relevant among adolescents, as bullying is believed to be increasingly driven by status needs in this developmental period. Whereas the Meaningful Roles intervention described above assumes that popular students (or the ones aiming for high popularity) will decrease bullying when they gain popularity by other means (Ellis et al., 2015), the finding by Garandeau, Lee, et al. (2014) suggests that popular students’ bullying behavior is especially hard to change. Rather than refraining from aggression, these adolescents may end up using both prosocial and aggressive strategies to maximize their status (Hawley, 2003; Laninga‐Wijnen et al., 2020).

Intervention trials targeted at adolescents should carefully consider the role of peer status, as well as goals, and investigate moderation of intervention effects by these factors. Also other, developmentally relevant factors such as psychological reactance have not yet been assessed in the context of anti‐bullying interventions. There are also potential contextual moderators of intervention effects, such as school climate (see Low & van Ryzin, 2014, for a study in elementary schools). From the developmental perspective, the most relevant contextual moderators of intervention effects in adolescence might have to do with peer group norms. Peer dynamics may be difficult to change in classrooms where bullying is rewarded by popularity (Dijkstra & Gest, 2015) or by encouraging gestures from bystanders (Pouwels, van Noorden, Lansu, & Cillessen, 2018).

THE ROLE OF STATUS IN BULLYING AND VICTIMIZATION

A challenge to anti‐bullying programs (particularly in adolescence) is that bullies may be rewarded with high status, which is increasingly important to adolescents (e.g., LaFontana & Cillessen, 2010). Given the importance of peer relationships and peer status (e.g., popularity, peer acceptance/rejection) in adolescence, a substantial amount of studies have examined the associations between bullying, victimization, and status among youth. One of the key contributions of this research has been a shift in the conceptualization of bullying perpetrators from maladjusted and socially isolated to socially adept and skilled. Through this lens, research in the last decade has increasingly recognized bullying as a strategic, goal‐directed behavior (e.g., Hawley, 2015; Volk, Dane, & Marini, 2014), and has led to important advances in the field. In this section, we review three key questions related to bullying and status: 1. how bullying and status are related to one another over time, 2. how youth’s status cognitions and goals are related to bullying, and 3. whether high status is always a protective factor against victimization. We primarily consider two forms of status: popularity (i.e., social visibility, prestige, and/or dominance) and peer acceptance/rejection (i.e., the extent to which adolescents are (dis)liked by peers).

Longitudinal Associations between Bullying and Status

Numerous studies have demonstrated a positive concurrent association between bullying perpetration and popularity (e.g., de Bruyn et al., 2010), and between bullying and peer rejection (e.g., Pouwels et al., 2016). Based on these cross‐sectional findings, a common perspective is that perpetrators are generally popular, but disliked. At the potential cost of being disliked, adolescents with high status (particularly popularity) may use aggression over time to maintain their status, such that they demonstrate their dominance to their peers (e.g., van den Berg, Burk, & Cillessen, 2019). Likewise, bullying is thought to be one tool that youth may use to gain popularity (e.g., by acquiring material and/or social resources; Volk, Camilleri, Dane, & Marini, 2012). However, research from the past decade has resulted in a more complex picture of the direction of this association over time.

Although there are reasons to expect bidirectional associations between bullying and status, the extant literature is quite inconsistent. In a large Finnish sample of early adolescents, no significant longitudinal associations were found between bullying and popularity (Sentse, Kretschmer, & Salmivalli, 2015; Sentse, Veenstra, Kiuru, & Salmivalli, 2015). In contrast, a recent study found a positive, bidirectional association between bullying and popularity over the span of 1 year in a sample of Italian early adolescents (Pozzoli & Gini, 2021). Other studies have found evidence that popularity is more likely to predict aggression over time, rather than aggression predicting popularity (e.g., Lu, Li, Niu, Jin, & French, 2018; Malamut, van den Berg, Lansu, & Cillessen, 2020; Ojanen & Findley‐Van Nostrand, 2014). Furthermore, Pouwels et al. (2016) found that youth who belonged to a stable popular group in late childhood/early adolescence were more likely to be bullies later in adolescence. A different study using joint trajectory analysis of bullying, popularity, and peer acceptance concluded that stable high levels of bullying often overlapped with stable high levels of popularity, but that bullying was not a prerequisite to achieving popularity (Reijntjes et al., 2013). They found that most early adolescents who belonged in the high‐bullying group also belonged to the high‐popularity group, whereas less than half of those in the high‐popularity group were simultaneously high in bullying. de Vries, Kaufman, Veenstra, Laninga‐Wijnen, and Huitsing (2021) found that bullies had higher popularity than other groups, but belonging to a bully group did not predict changes in popularity over time. Thus, popularity generally appears to be a robust predictor of aggression, whereas less consistent evidence has been found for aggression predicting popularity. Similar patterns have been found for both physical (e.g., hitting others) and indirect/relational (e.g., excluding others) forms of aggression (e.g., Lu et al., 2018; Malamut, Berg, Lansu, & Cillessen, 2020; Ojanen & Findley‐Van Nostrand, 2014).

Our understanding of the longitudinal links between bullying and status is further complicated by the fact that peers may not have a uniform reaction to bullying. Indeed, person‐centered approaches have found that some subgroups of bullies are rejected by peers, whereas other subgroups of bullies have average social acceptance (e.g., Reijntjes et al., 2013), or even have high levels of acceptance and number of friends (e.g., De Vries et al., 2021). These findings are consistent with a cross‐sectional study that found that bullies (and victims) were not particularly disliked by their classmates—instead, bullies’ and victims’ reputations for being disliked were primarily driven by mutual antipathies between bullies and victims (Hafen, Laursen, Nurmi, & Salmela‐Aro, 2013). In one longitudinal study, limited prospective associations were found between bullying and rejection: for adolescent boys, peer rejection predicted higher levels of bullying 5 months later, but bullying did not predict becoming more rejected by peers (Sentse, Kretschmer, et al., 2015; Sentse, Veenstra, et al., 2015). Similarly, Pozzoli and Gini (2021) found a weak, negative association between social preference (i.e., being well‐liked) and subsequent bullying, but did not find that bullying was a significant predictor of social preference over time. Taken together, these studies suggest that bullies may not incur as many social costs amongst peers (in terms of likeability or affection) as previously suggested by most cross‐sectional research.

Therefore, somewhat surprisingly, the extent to which bullying and status reinforce each other over time remains unclear, despite years of research on this topic. One challenge to disentangling the temporal associations of bullying and status is the high stability of status (particularly popularity). Thus, it may be more difficult to identify the impact of bullying on popularity over time; especially as popularity has many behavioral and personality correlates other than aggression (Vaillancourt & Hymel, 2006). Another challenge is that bullies represent a heterogenous group (Peeters, Cillessen, & Scholte, 2010)—some bullies may be high in status, whereas others are low in status. For example, bullies who are also victimized by peers (i.e., bully‐victims) are generally lower in popularity and social preference than “pure” bullies (e.g., Guy, Lee, & Wolke, 2019). Yet, longitudinal studies on the associations between bullying and status often do not take this into account.

Moreover, a growing body of research in the past decade has highlighted that the extent to which bullying and aggression are normative or rewarded with status varies across contexts (e.g., schools, classrooms; Dijkstra & Gest, 2015; Garandeau, Laninga‐Wijnen, & Salmivalli, 2011; Laninga‐Wijnen et al., 2017). Following the social misfit model (Wright, Giammarino, & Parad, 1986), adolescents experience social sanctions if they behave in ways that are inconsistent with the norms or values of their peer group. Indeed, peer norms appear to impact how accepted or rejected bullies are in the peer group (Dijkstra & Gest, 2015). Still, these studies have been primarily cross‐sectional, and the few longitudinal studies (e.g., Berger & Caravita, 2016; Sentse, Kretschmer, et al., 2015; Sentse, Veenstra, et al., 2015) have not found a consistent pattern of results in adolescence. Further research is still needed to understand how the peer norms influence the bidirectional, longitudinal associations between bullying and status.

Due to these challenges, it is difficult to evaluate the extent to which interventions can impact the associations between bullying and status. Importantly, and contrary to common conceptions, there is little to no evidence that bullying is costly in terms of peer acceptance or social preference over time. This poses a challenge for intervention efforts, as it may be difficult to persuade adolescents to cease bullying if they do not perceive there to be any social costs to their behavior. In adolescence, being reprimanded by teachers or adults may not be seen as a salient “cost” to the same extent as low peer status. Furthermore, more tailored intervention strategies may be needed for classrooms or schools with differing bullying norms.

Bullying and Status Goals

Another challenge to understanding the link between bullying and status is that adolescents vary in the extent to which they value popularity and peer acceptance. Although research on the impact of adolescents’ status goals on aggression and bullying originated a long time ago (e.g., Hawley, 1999; Pellegrini, 2002), the shift in focus to bullying as a strategic, goal‐directed behavior in the last decade corresponded with an increased interest in the underlying goals, motivations, and status cognitions of youth who bully.

The idea that bullying may be used strategically to gain status presumes that the individual engaging in bullying wants to gain status, which suggests that status goals should be a unique predictor of aggression. Cross‐sectional studies have found that status goals (e.g., popularity goals, agentic goals, social demonstration goals, and social dominance goals) are positively related to bullying (e.g., van den Broek, Deutz, Schoneveld, Burk, & Cillessen, 2016) and aggression more generally (e.g., Cillessen, Mayeux, Ha, de Bruyn, & LaFontana, 2014). In a sample of Canadian high schoolers, high popularity motivations in the beginning of the school year were associated with elevated levels of indirect/relational aggression at the end of the school year (Dumas, Davis, & Ellis, 2019). Moreover, adolescents high in bullying are more likely to find it more important to be popular than well‐liked (Garandeau & Lansu, 2019). Although status goals related to popularity or dominance may be one explanation for why adolescents bully, not all youth will be equally equipped to act on those goals. Several studies suggest that aggression and popularity are more likely to be longitudinally associated when youth have high motivation to be popular or socially dominant (e.g., Dawes & Xie, 2014). For example, in a sample of Finnish early adolescents, both physical and indirect/relational aggression only predicted elevated popularity over time when youth had higher agentic goals (Ojanen & Findley‐Van Nostrand, 2014). Still, not all studies have found a significant interaction between status and status goals in predicting aggression (e.g., Dumas et al., 2019).

In addition to individual characteristics, there are also contextual factors to consider that may impact the association between status goals and bullying. In a study of Chinese early adolescents, Pan et al. (2020) found that social dominance goals were only associated with increasing bullying perpetration across 1 year in classrooms where popularity was asymmetrically distributed (i.e., high status hierarchy). This further suggests that adolescents strategically use aggression: adolescents with high social dominance goals were more likely to bully in contexts where there are likely to be social benefits to bullying.

Victimization and Status

Insofar as bullying in adolescence is strategic, goal‐directed behavior, youth should also be strategic about who they target with aggression. In keeping with this perspective, research has traditionally conceptualized victims of aggression as low status, socially marginalized youth—in other words, “easy” targets. Through this lens, bullies would presumably not be taking a large risk of losing affection or facing retaliation (e.g., Veenstra, Lindenberg, Munniksma, & Dijkstra, 2010). However, even though targeting low‐status peers may be seen as “low risk”, it also may not be seen as “high reward”. From a strategic standpoint, there is more to potentially gain by strategically targeting a social competitor (i.e., instrumental targeting) than a peer who is already low in the social hierarchy (e.g., normative targeting) (e.g., Andrews, Hanish, & Santos, 2017; Faris, 2012).

Although many victims of bullying do have low status (e.g., de Bruyn et al., 2010), growing evidence has highlighted that youth with high status can also be victims of bullying and aggression (see Dawes & Malamut, 2020 for a review). This idea originated in the 1990s (e.g., see Adler & Adler, 1995; Merten, 1997; Prinstein & Cillessen, 2003), but research in the last decade has increasingly called attention to this phenomenon. For example, Sainio, Veenstra, Huitsing, and Salmivalli (2012) found that being seen as highly popular by other‐sex peers increased the likelihood of being bullied by other‐sex peers. High‐status victims were also found in a recent study using latent profile analysis to identify different subtypes of victimized youth (Malamut, Dawes, et al., 2021). Although there is evidence of popular youth being targeted with multiple forms of victimization (Dawes & Malamut, 2020), they are particularly likely to be targeted with indirect/relational forms of victimization (e.g., Badaly, Kelly, Schwartz, & Dabney‐Lieras, 2013; Closson, Hart, & Hogg, 2017; Dyches & Mayeux, 2012; Malamut, Dawes, & Xie, 2018; Malamut, Luo, & Schwartz, 2020).

Consistent with ideas of instrumental targeting, popular bullies are more likely to target high‐status peers (Malamut, van den Berg, Lansu, & Cillessen, 2020; Peets & Hodges, 2014), and targeting peers with higher status (e.g., social centrality) can lead to increased social network prestige for the aggressor (Andrews et al., 2017). Moreover, bistrategic youth who engage in both aggressive and prosocial behavior (Hawley, 2003) were more likely to target popular peers with social aggression (Wurster & Xie, 2014). There is also some indication that popular youth are likely to be victimized within the contexts of friendships (e.g., Closson & Watanabe, 2018); perhaps, because friends (and friends‐of‐friends) are likely direct social competitors (see Faris, Felmlee, & McMillan, 2020).

This growing body of research challenges long‐standing assumptions that high status is always a desirable characteristic associated with positive outcomes, as popular youth appear to be more at risk for negative outcomes than previously thought (e.g., Dawes & Malamut, 2020; Schwartz & Gorman, 2011), including psychosocial maladjustment (e.g., lower social satisfaction and social self‐concept, poorer best friendship quality: Ferguson & Ryan, 2019). High‐status victims were found to have larger increases in internalizing symptoms compared to low‐status victims, possibly because they have “more to lose” (Faris & Felmlee, 2014). Moreover, the overlap between high status and victimization may be one factor contributing to aggression in the peer group. Highly popular adolescents who reported high levels of indirect victimization were more indirectly aggressive 1 year later (Malamut, Luo, et al., 2020).

Thus, it is critical to understand the consequences of victimization amongst popular youth; however, there are some methodological challenges to identify high‐status victims. Other informants (e.g., peers, teachers) may not always recognize or report popular youth as being victimized, because these youth do not fit the idea of a typical victim (e.g., Bjereld, Daneback, & Mishna, 2021; Dawes, Norwalk, Chen, Hamm, & Farmer, 2019). Distinguishing between specific forms of victimization, examining curvilinear associations between status and victimization, and measuring victimization via dyadic nominations all may assist with identifying victims with high status (Dawes & Malamut, 2020). It is essential to account for the victimization experiences of popular adolescents to have a comprehensive understanding of victimization dynamics, and their experiences may also have important implications for intervention efforts. For example, there is already evidence that interventions are less effective for popular bullies (Garandeau, Poskiparta, et al., 2014)—popular adolescents’ experiences with, or their perceptions of, being the target of aggression may partially explain the limited effectiveness of interventions on popular bullies.

CONSEQUENCES OF DEFENDING

Bullying is not solely a matter between bullies and victims—it often is a group process in which multiple peers are involved (Salmivalli, 2010). Some peers may assist the bully whereas others merely observe the situation. Prior work shows that peers are present in 80% of bullying episodes on average (e.g., Craig & Pepler, 1997; Jones et al., 2015), and even though most students disapprove of bullying (Pouwels, Lansu, & Cillessen, 2017), often only a minority of bystanders (10–25%) reaches out to help victims (Trach et al., 2010; Quirk & Campbell, 2015). The silence of passive bystanders can hurt even more than the bullying itself (Jones et al., 2015).

The observation that most peers remain passive in bullying situations has led various anti‐bullying programs to adopt the encouragement of peer defending as a central component in combatting bullying (Gaffney et al., 2021). Defending can be defined as a type of prosocial behavior that is shown in response to bullying situations. Studies have recently started to distinguish two main types of defending: direct, bully‐oriented defending (such as publicly confronting bullies) and indirect, victim‐oriented defending (comforting victims or asking help from adults; often occurring more privately; Lambe & Craig, 2020; Reijntjes et al., 2016; Yun & Juvonen, 2020).

Adolescents are assumed to undertake several sequential steps before they decide upon defending a victimized peer (bystander intervention model; Latané & Darley, 1968). These steps include: 1) noticing the event, 2) interpreting it as an emergency situation, that is, someone is suffering and in need of help, 3) feeling personally responsible for defending, 4) knowing how to defend, and 5) endorsing this defending behavior. Interventions tap into these steps in various ways. For instance, they might aim to raise empathy for victims (Garandeau et al., 2021), or increase individuals’ feelings of responsibility for defending (Peets, Pöyhönen, Juvonen, & Salmivalli, 2015). Moreover, regarding the fourth and fifth steps, the various—direct and indirect—ways in which individuals can defend are brainstormed and discussed with youth, and practiced in online environments or role‐play sessions (e.g., Salmivalli, 2014).

Theoretically, it can be argued that encouraging defending is beneficial for victims: psychological stress‐buffering theories posit that receiving help in stressful circumstances promotes adaptive appraisal and coping, which should buffer against psychosocial problems (Cohen & Wills, 1985). Moreover, social protection theory suggests that bullies are less likely to target children who are protected by peers, due to fear of retaliation (Hodges & Perry, 1999). Thus, peer defending should reduce bullying and promote victims’ psychosocial adjustment. Despite the theorized benefits of defending, researchers have started to ask whether—under some circumstances—defending can be risky for defenders or even for victims.

Is Defending Risky for Defenders?

It has been theorized that intervening in favor of victims could have detrimental social and psychological consequences for defenders (Meter & Card, 2015; Pozzoli & Gini, 2012). In terms of social consequences, it has been argued that defending can result in a loss of social status, or increase the risk of becoming a victim oneself. Bullies are powerful peers who often do not operate on their own. Preventing aggressors from reaching their goals (Pozzoli & Gini, 2012) and challenging the status and power of perpetrators (Salmivalli, Voeten, & Poskiparta, 2011) can be risky and result in retaliation—particularly if defending proves to be unsuccessful (retaliation hypothesis; Spadafora et al., 2018). In qualitative studies, one of the main reasons why adolescents are hesitant to defend a victimized peer is that they are afraid they will lose status or become the target of bullying themselves (Strindberg, Horton, & Thornberg, 2020). This fear is not wholly unfounded: some studies have suggested that defending can result in a decrease in social preference (Meter & Card, 2015; Pozzoli & Gini, 2012). A social network study on an elementary school sample also indicated that defenders ran the risk of becoming victimized by the bullies of the peers that they defended (Huitsing, Snijders, Van Duijn, & Veenstra, 2014). It is likely that this risk of defending is particularly present during adolescence. First, compared to childhood, adolescent bullies are more powerful and have higher status (Dawes & Xie, 2014)—which makes it riskier to stand up against them. Moreover, if defending is encouraged by adults in interventions, defending may be seen as less genuinely driven by a feeling of moral injustice (Healy, 2020) and rather “soft”, teacher‐obedient behavior, and hence may be sanctioned with lower status among peers.

Besides these social consequences, defending may have psychological costs, perhaps because defenders actively intervene in bullying—a stressful and potentially traumatic situation (Lambe et al., 2017). Prior work on other stressful experiences, such as interparental conflict, has shown that youth who actively intervened experienced more internalizing and externalizing problems than youth who just passively observed it (Jouriles et al., 2014). Similarly, youth who actively try to stop a bullying episode could suffer more from psychosocial consequences than youth who just observe passively. Moreover, often, defenders are higher on affective empathy, which could also make them more susceptible to feel along with how victims feel (Olivia et al., 2014). Accordingly, several studies detected a positive concurrent association between defending and psychosomatic and internalizing problems among defenders (e.g., Callaghan et al., 2019; Jenkins & Fredrick, 2017; Lambe et al., 2017; Malamut, Trach, Garandeau, & Salmivalli, 2021). These studies point to potential adverse consequences of defending which poses a dilemma: should peer defending be encouraged if it is risky for defenders themselves?

In response to this emerging debate, an increasing number of studies started examining whether defending indeed is risky for defenders longitudinally, as there were also reasons to assume that defending could work out positively for them on the longer term. Specifically, defending could enhance someone’s popularity, because defenders show that they are powerful and brave enough to take a stance against bullying, which may foster respect and admiration among other peers (Reijntjes et al., 2016). Defending may also increase social preference among peers because they undertake prosocial actions that are helpful for others (Pronk et al., 2020). Indeed, qualitative work indicated that students believe defending may also result in higher social status (social preference and popularity) over time (Spadafora et al., 2018). Further, being able to help somebody in times of harassment may improve defenders’ self‐views and well‐being.

Most longitudinal work indicates that defending can promote social status. Two studies found that defending positively predicted social preference, but not popularity (Pronk et al., 2020; Pozzoli & Gini, 2021). Another study showed that defending predicted popularity 1 year later (Van der Ploeg, Kretschmer, Salmivalli, & Veenstra, 2017). An experimental study found that adolescents regarded defenders as most favorable compared to bullies, bystanders, or victims: actual and hypothetical defenders received more “liking” nominations; and in a computerized reaction task, defenders were evaluated more positively implicitly (Pouwels et al., 2017). Regarding psychological consequences, a longitudinal investigation (Malamut, Trach, et al., 2021) found that, in general, defending was unpredictive of future internalizing problems. Importantly, initial victimization status of defenders appeared to play a role in the extent to which defending related to internalizing symptoms: defending was associated with elevated depressive symptoms for low‐status, highly victimized youth, whereas defending related to lower levels of depressive symptoms for high‐status youth. Two intervention studies found that high school female students who were trained in a brief, bystander bullying intervention reported a greater decrease in internalizing symptoms over a 3‐month time span compared to female students in a control group (Dumas et al., 2019; Midgett & Doumas, 2019). However, these studies did not assess whether this decrease in internalizing symptoms could be explained by an increase in actual defending behavior as a result of this intervention.

To conclude, the preliminary longitudinal evidence suggests that defending is generally not a risky behavior for defenders, which means that most youth presumably can defend safely. However, more research is needed to further understand whether all youth can defend safely under all circumstances. For instance, victimized youth were more likely to develop internalizing symptoms when defending others (Malamut, Trach, et al., 2021), thus for some youth, defending may pose an additional risk. Moreover, classroom norms may affect the extent to which defenders are regarded favorably by their peers or not (Pouwels, van Noorden, & Caravita, 2019). Consequently, more research is needed to identify moderators of the effects of defending on adjustment and determine why and when defending may be risky for defenders. Even though qualitative studies indicate that some students fear the negative social consequences of defending, an important area for research is to which extent defenders actually care about a potential drop in status—students vary in the extent to which they strive for popularity or being liked (Dawes & Xie, 2017), and may predominantly defend because they feel empathy for the victim or because they consider bullying as morally wrong (Pouwels et al., 2019).

Is Defending Risky for Victims?

Theoretically, it seems plausible that being defended would be beneficial for victims. Bullies are less likely to target youth who are protected by other peers presumably for fear of retaliation (Hodges & Perry, 1999) and defending may signal to victims that others care about them and do not condone the bullying, which may be comforting for victims and help them cope with their plight (McDougall & Vaillancourt, 2015; Scardera et al., 2020). Thus, being defended by peers should logically reduce bullying and promote victims’ psychosocial adjustment over time.

Despite these clear theoretical arguments, empirical findings on the role of being defended in diminishing bullying and improving victims’ psychosocial functioning vary considerably across prior studies, and most studies have been conducted on elementary school samples—leaving it unknown whether defending can be helpful in adolescence. With regard to decreases in bullying, one longitudinal study found higher classroom levels of defending to relate to lower levels of bullying perpetration (Saarento, Garandeau, & Salmivalli, 2015; Saarento, Boulton, & Salmivalli, 2015. However, not every defending attempt effectively ends bullying episodes: a naturalistic observation study showed that this was true in about two‐third of the cases (Hawkins, Pepler, & Craig, 2001). Moreover, an intervention study on children and preadolescents found victims without a support group to report decreased victimization at the end of the school year, whereas victims with a support group did not report such a change. A support group was a teacher‐assigned group of six to eight students, including bullies and their assistants, defenders or friends of the victim, and prosocial classmates, designed to create mutual concern for the well‐being of victims and to encourage bullies to alter their behavior. Strikingly, a third of victims with a support group even reported that their victimization levels had increased—despite having more defenders (van der Ploeg et al., 2016). A longitudinal study indicated that defended and non‐defended victims (at the start of the school year) had comparable levels of victimization at the end of the school year (Laninga‐Wijnen, van den Berg, Garandeau, Mulder, & de Castro, in press).

Studies examining the role of being defended on victims’ psychological functioning also report mixed findings. Two cross‐sectional studies on (pre‐)adolescents demonstrated that defended victims had lower anxiety and depression (Ma et al., 2019; Sainio, Veenstra, Huitsing, & Salmivalli, 2011) and higher self‐esteem (Sainio et al., 2011) than non‐defended victims. Another study, however, detected no effect of being defended in victims’ distress (Jones et al., 2015) —even though this study also indicated that victims were particularly hurt if bystanders did nothing to intervene. Longitudinal work on the role of being defended in victims’ adjustment is scarce. Three longitudinal studies have examined the moderating effect of a related construct, namely, friendship support—such as having friends one can trust and rely on—in the association between victimization and future psychological adjustment. One study did not detect any significant role of friendship support in changes in victims’ psychological functioning (Burke, Sticca, & Perren, 2017). Results from the two other longitudinal studies were counter‐intuitive; receiving emotional support from friends increased victims’ depressive symptoms (Desjardins & Leadbeater, 2011) and enhanced maladaptive coping and distress among victimized girls (but not among victimized boys; Thompson & Leadbeater, 2013). To our knowledge, the impact of being defended on victims’ adjustment has been examined longitudinally only in one study. This study demonstrated that defended victims were higher on their feelings of connectedness to the classroom, but did not differ from non‐defended victims in terms of depressive symptoms or self‐esteem at the end of the school year (Laninga‐Wijnen et al., in press). Lastly, a recent meta‐analysis indicated that interventions that included informal peer involvement (such as group discussions) and encouragement of defending were effective in reducing victimization; however, interventions that did not include encouragement of peer defending were more effective than interventions that did include it (Gaffney et al., 2021).

These findings may indicate that tackling bullying by encouraging defending can be difficult, and that the benefits of defending for victims may be more limited than once assumed. As with every intervention, defending may go together with both positive and negative consequences. A recent commentary (Healy, 2020) theorizes that defended victims may be even worse off than non‐defended victims because defending could disempower victims by making them dependent on their helpers and encouraging the belief that they cannot solve problems themselves. Defending could also provoke additional bullying attempts if this defending is enacted in inappropriate or aggressive ways—this could elicit retaliation of bullies, or stigmatize victims by making them stand out from other peers as the ones needing special treatment in contrast to those who can stand up for themselves.

Even though the literature does not provide strong evidence for the statement that defending would work out adversely, it is clear that more research is needed to inform interventions on how to increase the effectiveness of defending attempts. This requires knowledge on the factors that may explain why defending can work out adversely versus beneficially (i.e., mediators), as well as on victim‐, defender‐, and contextual characteristics that may either exacerbate or mitigate the effectiveness of defending attempts (i.e., moderators).

Underlying reasons on why being defended either helps or hurts victims may be both internal and external to the victim. One internal reason for adverse effects of defending could be that being defended promotes maladaptive cognitions in victims about the solution for the bullying issue. For instance, defending may undermine victims’ feeling of autonomy; it can enhance victims’ beliefs that they are powerless to stand up against bullies themselves and that they need others to make the bullying stop (i.e., learned helplessness; Healy, 2020). Moreover, indirect forms of defending such as providing comfort or listening to the victims’ story may stimulate co‐rumination processes in which victims dwell on negative affect and repeatedly think about their negative experiences together with their defenders (Schacter & Juvonen, 2020). These maladaptive cognitions and rumination processes put victims at risk for further psychological maladjustment. On the other hand, an internal reason explaining why defending promotes victims’ adjustment is because defending promotes adaptive appraisal and adequate coping styles within victims, which, in turn, helps them to restore their self‐esteem or diminishes symptoms of depression (Cohen & Will, 1985).

External reasons why defending may work out adversely is that it can set victims even more apart from others: these victims are not only considered as weak because they are victimized, but also because they apparently are the ones in need of help. Because of this deviation from others, classmates may start blaming victims for their plight or see these victims as “social misfits”, which may backfire in victims’ psychological functioning (Healy, 2020). An external reason for why defending may work out beneficially for victims is that it clearly signals that the bullying is not condoned by other peers, which could discourage bullies in their behaviors.

The extent to which these internal or external processes occur after being defended may depend on the type of defending that is enacted, as well as on victim‐, defender‐, and contextual characteristics. For instance, regarding the type of defending, victims were found to have better psychosocial adjustment in classrooms where it was more common to defend victims in direct ways (i.e., by confronting the bully; Yun & Juvonen, 2020). It could be that these direct ways of defending signal to the victim that others blame the bully for the situation, which could diminish victims’ self‐blame (Yun & Juvonen, 2020). Furthermore, regarding victim characteristics, gender may play a role: victimized girls who received support from their victimized friends were found to have more internalizing problems compared to victimized boys who received support from their victimized friends (Schacter & Juvonen, 2020). It could be that girls tend to discuss emotions frequently, which may result in co‐rumination processes. Regarding defender characteristics, defending may be more effective if defenders possess sufficient popularity to deflect the bully’s dominance. Accordingly, a former study in classrooms where defending is endorsed by popular peers rather than by unpopular peers, all students—including victims—have higher well‐being at school and regard the classroom as more positive (Laninga‐Wijnen et al., 2021). At last, contextual factors such as the context of the bullying may play into how helpful defending is viewed to be by the victim. For instance, victims’ perceptions of the effectiveness of defending may depend upon whether there is an audience to the bully, or on the severity of the bullying. Broader classroom factors, such as how teachers react to bullying or classroom norms, may also play a role, because they determine the value of defending behaviors among youth. If teachers are strongly disapproving of the bullying and if classmates consistently consider bullying as an inappropriate behavior (an anti‐bullying norm), defenders may be more likely to be regarded favorably by their classmates and bullies may be more likely to comply to their anti‐bullying messages.

To conclude, more research is needed to understand which factors play a role in the effectiveness of defending in diminishing victimization and promoting victims’ adjustment. Teaching students to defend other peers is of vital importance: victims should not be left alone in their plight and previous studies have shown that having passive bystanders hurts victims even more than the bullying itself (Jones et al., 2015). Moreover, standing up for others in general (i.e., when somebody is being discriminated or threatened) is a general citizen skill which should be transferred to youth. Defending does not only have the potential to benefit victims, but also defenders, particularly in adolescence: a growing body of research indicates that adolescents are driven by a strong desire to contribute to society and to others’ well‐being, and that they reap emotional benefits from helping others (see Fuligni, 2019 for review). Thus, it is essential to gain more insights in the factors that contribute to the effectiveness of defending for victims, not only for victims themselves, but also for their defenders.

THE HEALTHY CONTEXT PARADOX

The past 20 years have seen increased recognition of the importance of the social environment in bullying and significant advances in the identification of contextual factors that promote bullying behavior (e.g., Saarento, Boulton, et al., 2015; Saarento, Garandeau, et al., 2015). This has led prevention efforts to increasingly focus on improving the classroom or school context as a whole. In the past decade, however, new findings have emerged to suggest that desirable or sought‐after features of the social context, such as low levels of victimization, could in fact have adverse effects on the adjustment of victimized youth (Garandeau & Salmivalli, 2019). The idea that “healthier” (i.e., more prosocial or less aggressive) classrooms may paradoxically increase maladjustment risks for vulnerable students was suggested in earlier research (e.g., Bellmore, Witkow, Graham, & Juvonen, 2004; Sentse, Scholte, Salmivalli, & Voeten, 2007), but the last 10 years have witnessed a renewed interest in this issue and a growing number of studies from different countries on a phenomenon that has come to be known as the healthy context paradox (Salmivalli, 2018; Huitsing et al., 2019).

Victimized adolescents have been found to have lower self‐esteem (Huitsing, Veenstra, Sainio, & Salmivalli, 2012) and higher levels of depressive symptoms (Yun & Juvonen, 2020), somatic complaints (Gini, Holt, Pozzoli, & Marino, 2020), and externalizing problems (Liu et al., 2021), in classrooms with lower levels of victimization. Verbally victimized fifth graders were also found to have more negative self‐views in low‐aggression than in high‐aggression classrooms (Morrow, Hubbard, & Sharp, 2019). These concurrent findings have been further supported by longitudinal research internationally. In a Finnish sample, youth who remained victimized across 1 year felt more depressed, more socially anxious, and were less liked at the end of the year in classrooms where the proportion of victims had decreased compared to stable victims in other classrooms (Garandeau, Lee, & Salmivalli, 2018). In a Chinese sample, victimized students had higher depression, lower self‐concept, and fewer opportunities for friends after 1 year in classrooms where victimization was less prevalent (Pan et al., 2021). Moreover, in an ethnically diverse sample of American adolescents, victimized youth showed stronger increases in characterological self‐blame (i.e., beliefs that they deserve their plight) one semester later in schools with lower levels of victimization (Schacter & Juvonen, 2015).

Empirical support for the healthy context paradox is not limited to evidence of a moderating effect of the average level of victimization in the classroom or school. Mean classroom levels of defending behavior were also found to be associated with victims’ maladjustment in adolescence. Research has shown that highly victimized youth had lower feelings of belonging and perceived less cooperation and cohesion in classrooms where defending was more common (Laninga‐Wijnen, Van den Berg, Mainhard, & Cillessen, 2021). Moreover, victimized youth may be worse off in schools implementing anti‐bullying interventions. In a study by Huitsing et al. (2019), chronic victims, as well as newly victimized students, had lower self‐esteem and higher depressive symptoms after 1 year in schools implementing an effective anti‐bullying program than their counterparts in control schools. It should be noted, however, that the intervention helped decrease bullying and was, therefore, beneficial for most students who were victimized at baseline.

These findings are obviously concerning, as they bring to light possible iatrogenic effects of successful anti‐bullying interventions for those who remain or become victimized despite the intervention. It is urgent to understand why improved social contexts may exacerbate the psychological and social difficulties that victims of bullying experience. So far, only one study has tested potential mechanisms accounting for the adverse effects of low‐victimization classrooms on the development of internalizing problems in victimized students (Pan et al., 2021). First, they demonstrated that such classrooms exacerbated victims’ depressive symptoms by reducing their opportunities for friendships. Indeed, victims tend to affiliate with other victims (Huitsing et al., 2014) and their non‐victimized peers are often reluctant to form friendships with them (Sentse, Dijkstra, Salmivalli, & Cillessen, 2013; Sijtsema, Rambaran, & Ojanen, 2013). Moreover, individuals who are dissimilar from the rest of the group—“social misfits” —are more likely to be rejected by their peers (Sentse et al., 2007; Wright et al., 1986). For these reasons, being in a social environment with fewer victims likely makes victimized students more socially isolated, with fewer friends, which, in turn, increases their depressive symptoms (Pedersen, Vitaro, Barker, & Borge, 2007). Second, the effect of lower levels of classroom victimization on future depressive symptoms was partly explained by its negative effect on victimized students’ self‐concept (Pan et al., 2021). When few peers in the classroom are subjected to bullying, youth who are bullied should be more apt to engage in upward social comparisons (Wills, 1981), which can be damaging to their self‐regard (Tesser, Millar, & Moore, 1988). Another likely mechanism for these effects is the increase in characterological self‐blame that contexts of lower victimization have been shown to promote (Schacter & Juvonen, 2015), since self‐blame tends to play an important role in depressive disorder (e.g., Zahn et al., 2015). With regard to the moderating effect of classroom victimization level on the link between victimization and externalizing problems, one cross‐sectional study has suggested that increased hostile attribution bias was at play (Liu et al., 2021). In classrooms with lower levels of victimization, frequent victims of bullying were more likely to attribute hostile intent to peers, which was associated with higher levels of externalizing problems.

A growing body of studies conducted with a diversity of samples is providing support for a healthy context paradox. However, it is important to keep in mind that these findings are not unanimous; for instance, one study with middle schoolers found that friendless victims felt less anxious, lonely, and unsafe after 1 year in schools with stronger peer prosocial norms (Schacter & Juvonen, 2018). Also, many of the above‐mentioned studies were conducted with pre‐ or early adolescents, and therefore it remains uncertain whether the healthy context paradox is still occurring later in adolescence. More longitudinal studies using samples from older age groups and examining different types of healthy contexts are needed for a better understanding of the phenomenon.

FUTURE DIRECTIONS

Bullying prevention programs have desirable effects, as compared with treatment as usual. However, the effects are modest, and they seem to be especially weak in adolescence. Furthermore, research carried out in the past decade has shown that some of the solutions proposed may rest on shaky assumptions—especially when targeted at adolescent populations—or involve components that pose new challenges, such as potential iatrogenic effects for some youth in some circumstances.

Several explanations have been proposed for why adolescents are less responsive to prevention and intervention efforts than younger children. However, few studies have directly tested whether factors such as strong status needs (Sijtsema, Veenstra, Lindenberg, & Salmivalli, 2009), or psychological reactance (Yeager et al., 2015) undermine the effects of school‐based prevention programs in adolescence. Even evidence regarding the much‐discussed effect of school transition on bullying is mixed (Farmer et al., 2011; Pellegrini, 2002; Pellegrini et al., 2010), and there are no studies investigating whether transition affects the success of interventions.

Analyses on effective components of anti‐bullying programs emerged 10 years ago and so far, they have not been done separately among younger versus older youth. Decisions regarding relevant components to include should be considered through a developmental lens. We acknowledge, however, that, for instance, autonomy supportive intervention components were not taken into account in meta‐analyses aiming to identify effective ingredients of programs, simply because there were not enough of them.

Due to the importance of peers, as well as the difficulty of accepting adult injunctions in adolescence, mobilizing peers in prevention work seems like a fruitful approach. But even then, it is necessary to first get the influential peers on board. In addition, it is critical to increase our knowledge about the kind of peer involvement that is most beneficial in interventions (see Gaffney et al., 2021). For instance, the consequences of defending for both defenders and victims need to be better understood. Even though there are strong reasons to assume that refraining from defending is hurtful for victims (Jones et al., 2015), more insight is needed in the relative effectiveness of different types of defending and the moderators of their effects (which victim‐, defender‐, and contextual factors may mitigate or even turn around the beneficial effects of peer defending for victimized youth).

There is some indication that popular youth might not respond to existing interventions by decreasing their bullying behavior. A recently developed Meaningful Roles intervention attempts to utilize adolescents’ need for status by providing prosocial, constructive roles for youth to gain status, rather than trying to stop them from doing something (bullying) that is rewarded with status. It is not clear, however, whether alternate ways to gain status would make youth refrain from aggression; in contrast, they may end up using both prosocial and aggressive strategies to maximize their status (Hawley, 2003; Laninga‐Wijnen et al., 2020). Overall, the longitudinal associations among high status, status goals, bullying behavior, and victimization are not as clear as one might think they are; this is an area in need of further inquiry.

Besides age, other moderators of the effects of anti‐bullying programs have been identified, such as students’ temperament, or their popularity among peers. These findings raise new challenges, especially the question of whether (and to what extent) tailoring interventions is realistic. Universal interventions can be adapted to better fit the classroom, school, and broader context, but even such adaptations should be informed by evidence. Tailoring targeted interventions for different individuals and situations is even more complex, and would need some kind of a decision model for adults. Kaufman et al. (2021) recently provided a preliminary process model for this; however, in their model the actual “tailoring” (deciding what should be done in a particular case) is still up to individual teachers. Future research will hopefully inform evidence‐based recommendations regarding intervention approaches that are most likely to be effective among specific subgroups of youth (such as popular bullies) or across different classrooms.

We argue that extensive focus on the average program effects no longer moves the field forward, even when based on stringent RCTs. We hope to see an increase in adolescence‐specific interventions models and studies investigating the moderators and mediators of their effects. The outcomes of program effects could be widened to look at different forms of bullying and victimization, including bullying in online contexts (see Salmivalli et al., 2011; Williford et al., 2013) and identity‐based bullying (Earnshaw et al., 2018). Very little is known about individual‐ and group‐level moderators of intervention effects in decreasing bullying and victimization, especially in adolescence. Also, despite the attention to the importance of studying mediation in intervention trials (e.g., Bradshaw, 2015; Eisner & Malti, 2013), studies doing this are almost non‐existent (for an exception, see Saarento, Boulton, et al., 2015; Saarento, Garandeau, et al., 2015).

Besides intended effects, intervention trials should examine both positive and negative side effects of prevention programs, including mental health outcomes (e.g., Williford, Noland, Little, Kärnä, & Salmivalli, 2012). Specifically, we call for more longitudinal tests of the mechanisms explaining the adverse effects of “healthy contexts” on victims’ adjustment. Researchers should examine different types of “healthy” contexts, such as average levels of classroom defending, or implementation of a particular anti‐bullying intervention. Future studies should also seek to identify mitigating factors for the healthy context paradox. In other words, are there individual or contextual characteristics that can make victims less susceptible to the adverse influence of healthier contexts? The strong practical implications of the healthy context paradox make this research particularly important. School professionals should not be discouraged from aiming for less aggressive and more prosocial classroom climates; however, a higher awareness and better understanding of these effects should make it easier to counteract them and better protect students who remain victimized.

It should be noted that preventing adolescent bullying and victimization does not only need to happen in adolescence; quite the contrary, it may be critical to start much earlier. If programs implemented at a younger age (e.g., elementary school) had long‐term effects, their implementation would be another way to reduce bullying and victimization in adolescence. Whether this is the case is not established, as most program evaluations to date are short‐term trials. However, there are reasons to believe that intervention work started early on lays the foundation for successful anti‐bullying work in adolescence.

ACKNOWLEDGEMENTS

Writing of the current review was supported by the Academy of Finland Flagship Program (decision number: 320162), ERC Advanced Grant for the first author (ERC AdG 2019 Challenge, 884434), and in part by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number F32HD100054 for the third author. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding organizations.