Classroom bullying norms and peer status: Effects on victim-oriented and bully-oriented defending

Abstract

Defending a victimized peer is a socially risky behavior that may require high peer status and may depend on how popular or disliked bullies are in the classroom (i.e., within-classroom correlations between bullying and status). Past research has investigated defending as a unidimensional construct, though it can involve confronting the bully (bully-oriented defending) or supporting the victim (victim-oriented defending). This study used multilevel modeling to examine the effects of individual peer status, gender, and bullying as well as two indicators of classroom norms—the bullying-popularity norm and the bullying-rejection norm—on both types of defending. Our sample included 1,460 Dutch adolescents (50% girls; M _age 11 years) from 59 classrooms in 50 schools. Likability and popularity were positively associated with both types of defending. Being female and lower in bullying was associated with victim-oriented defending, whereas being male and higher in bullying was associated with bully-oriented defending. In classrooms where bullies were more rejected, both types of defending were more prevalent, and the positive associations of likability and popularity with victim-oriented defending were stronger. The positive effect of the bullying-rejection norm on victim-oriented defending was stronger for girls. Moreover, the effect of popularity on bully-oriented defending was stronger in classrooms where bullies were less popular.

Keywords

Bullying victim-oriented defending bully-oriented defending classroom norms peer status perceived popularity likability

School bullying, defined as intentional aggression enacted repeatedly against a less powerful peer, affects millions of children worldwide and can compromise their psychological adjustment (e.g., Reijntjes et al., 2010). The prevalence of bullying in a classroom depends partly on how bystanders behave: Peer interventions in favor of the victim can stop the bullying in 57% of bullying incidents (Hawkins et al., 2001) and bullying is higher in classes where defending is less common (Salmivalli et al., 2011). Understanding factors that promote defending is, therefore, essential for anti-bullying intervention efforts.

Defending victimized peers involves social risks, as defenders may experience decreased likability (Meter & Card, 2015) or increased victimization (Huitsing et al., 2014). Accordingly, defending depends on the social context (Peets et al., 2015; Yun & Graham, 2018). First, high likability (or social preference) and high perceived popularity (i.e., visibility and prominence among peers) are key predictors of defending (e.g., van der Ploeg et al., 2017). Second, classroom norms of bullying may affect adolescents’ decision to stand up for victims (Peets et al., 2015). Defending may be more difficult to carry out in classrooms where bullies are highly popular or less disliked, as the consequences in terms of status loss and increased victimization may be worse in those contexts. The degree to which bullying is rewarded with popularity, as indexed by the within-classroom correlation between these two constructs, has been referred to as the bullying norm salience (e.g., Dijkstra & Gest, 2015). In this article, we refer to class-level associations between bullying and popularity and between bullying and rejection as the bullying-popularity norm and the bullying-rejection norm, respectively.

Defending encompasses two distinct components (Reijntjes et al., 2016): confronting the bully (bully-oriented defending) and supporting the victim (victim-oriented defending). It is unknown whether classroom bullying norms differentially relate to the two defending types. This study examined the main effects of classroom bullying-popularity norms and bullying-rejection norms on each type of defending in a sample of early adolescents. It also investigated whether these classroom norms moderate the associations between likability, popularity, and gender, and each type of defending.

The Role of the Social Context in Defending

Several traits and emotions have been linked to defending in childhood and adolescence: High affective empathy (e.g., Barchia & Bussey, 2011), agreeableness (Pronk et al., 2015), and moral emotions, such as guilt and shame (Pronk et al., 2016). Girls also tend to defend more than boys (e.g., Pöyhönen et al., 2010), presumably because, unlike boys, they have been socialized to exhibit prosocial behavior (Brody, 1999). However, such personal features may not suffice to trigger intervention if the bystander does not enjoy high enough status among peers.

Two types of peer status have been found to be positively associated with defending in bullying situations: Likability (e.g., Caravita et al., 2009, 2010; Pöyhönen et al., 2010) and perceived popularity (e.g., Duffy et al., 2017; Pöyhönen et al., 2010; van der Ploeg et al., 2017). It should be noted that gender was found to moderate status-defending links (Duffy et al., 2017); popularity positively predicted defending only for girls, possibly because girls tend to perceive bullying as more hurtful than boys (Galen & Underwood, 1997) and, therefore, might need to feel popular enough to dare challenge bullies by helping their victims (Duffy et al., 2017). Further, peer status may moderate links between personal qualities and defending: Early adolescents high in affective empathy defended more than others only if they were popular and liked (Pöyhönen et al., 2010). This suggests that likability and/or popularity could be prerequisites to engage in defending. As potential defenders may be concerned about status loss and increased odds of being victimized, having a strong social position could make them feel secure enough to defy bullies without running the risk of becoming a target. Feeling accepted by others may also raise their motivation to act prosocially toward them.

Concerns about possible decreases in status and future victimization may inhibit defending even more when those who bully are socially powerful. Bullies’ high status could discourage defending because the achievement of popularity becomes prioritized in early adolescence (LaFontana & Cillessen, 2010), and affiliating with popular peers can increase one’s popularity (Dijkstra et al., 2010; Marks et al., 2012). If befriending popular peers can enhance one’s status, challenging them by defending their targets—who tend to be socially stigmatized (Juvonen & Galvan, 2008)—may result in being perceived as “uncool” and lead to a decline in status. Therefore, social contexts where bullies are popular should deter defending in two ways. First, such contexts should discourage adolescents who aim to increase their status; they should tend to side with the popular bullies and thus refrain from defending the victims of these bullies. Second, given that individuals are often more concerned with “maintaining face” than “gaining face” (Ho, 1976, p. 131), these contexts should also discourage adolescents who wish to avoid losing status. Moreover, an anti-bullying program aimed at increasing defending was found to be least effective with highly popular bullies (Garandeau et al., 2014), which could suggest that peers intervene less when the bullying is perpetrated by popular classmates. Consistent with this reasoning, lower defending was found in classrooms where bullies were more popular, and popular students were less likely to defend in those classrooms (Peets et al., 2015).

Similarly, in classrooms where bullies are highly disliked, bystanders should find it easier to stand up for victims. If bullies are rejected by the group, thwarting their behavior does not run counter to class norms; it should be perceived positively by others and thus should not result in a decline in status. Potential defenders may also fear becoming a target less, as they may expect that being bullied by a highly disliked classmate will be easier to deal with. Thus, both bullies’ level of popularity and level of rejection can be expected to influence defending of their victims.

Two Types of Defending

All findings discussed above are based on studies in which defending in bullying situations was operationalized as a single construct. However, defending can encompass various behaviors: Victim-oriented defending includes offering support to victims by being friendly to them and informing school personnel about the situation. Bully-oriented defending includes showing one’s anger to the perpetrators and trying to stop them. These two types of defending were first shown to be distinct in a study where early adolescents were presented with imaginary bullying scenarios and asked to rate how likely they were to employ different intervention strategies (Pronk et al., 2013), and subsequently in a study on actual defending behavior (Reijntjes et al., 2016). The latter study revealed that 25% of the children in the sample engaged in victim-oriented defending but refrained from bully-oriented defending, whereas 13% showed the reverse profile. Only 10% were high in both types of defending. The two groups engaging in a single type of defending differed in many respects: Victim-oriented defenders were more likely to be girls, low in bullying, highly liked, but not highly popular. Bully-oriented defenders were more likely to be male, high in bullying and in perceived popularity, but not high in likability. The finding that some students both defend and bully seems counterintuitive. Studies of adolescents’ social networks provide a possible explanation (Huitsing et al., 2014): bullies may target out-group members while being defended by members of their own group. Overall, these findings emphasize the necessity to consider the heterogeneity of defending when investigating its predictors.

It is unknown whether the main effect of classroom bullying norms on defending and their moderating effect on the relation between individual status and defending apply to both types of defending to the same degree. We envisioned two possibilities. First, we may consider that bully-oriented defending, which involves direct confrontation of the bully, is more socially hazardous than providing emotional support to the victim. When confronted directly, the bully is aware of being challenged and may retaliate against the defenders by targeting them and damaging their social reputation. Giving emotional support to the victim, however, may be done privately, out of sight of the perpetrators. Consequently, it should be safer with regard to risks of future status loss and victimization. Following this line of reasoning, bully-oriented defending should be more sensitive to classroom bullying norms than victim-oriented defending. That is, the main effects of the bullying-popularity norm and the bullying-rejection norm should be stronger for bully-oriented defending. High-status adolescents would be more likely to use their social position to defend others in classrooms where bullies are less popular or more disliked (Peets et al., 2015). These moderating effects of classroom norms on the status-defending links should also be stronger for bully-oriented defending.

We also considered an alternative possibility. Because bully-oriented defenders were found to engage in bullying themselves (Reijntjes et al., 2016), it is possible that this type of defending is used by ringleader bullies or their followers. In this case, their defending would be aimed at standing up for their bullying friends, exposed, for instance, to the reactive aggression of their victims, since bullies with the same victims tend to defend each other over time (Huitsing et al., 2014). If bully-oriented defending is partly done by adolescents who also bully, it should be less socially risky and unlikely to be prevented by how popular or disliked bullies are in the class. In this case, the main effects of classroom bullying norms on defending and their moderating effects on status-defending relations should be stronger for victim-oriented defending.

Gender differences in these effects also deserve further investigation. First, we expected that girls would be more likely than boys to use victim-oriented defending, and boys would be more likely than girls to use bully-oriented defending (Reijntjes et al., 2016). Boys and girls differ in their relationship styles: girls are more apt to self-disclose to their friends, to respond prosocially to conflict situations, and to be socialized to show nurturance (Brody, 1999; Rose & Rudolph, 2006). Therefore, defending by supporting the victim fits girls’ relationship style better, whereas confrontation of bullies is more in line with boys’ relationship style. As victim-oriented defending is more normative for girls and bully-oriented defending more normative for boys, we also expected that gender might moderate the strength of the associations between status and each type of defending. As behaviors are more rewarded with status when exhibited by the gender for which they are more normative (Chang, 2004), high likability and high popularity should be more strongly associated with victim-oriented defending for girls than for boys and with bully-oriented defending for boys than for girls. As it would be logical for classroom norms to have a smaller effect on adolescents exhibiting a type of defending that is less stereotypical for their gender, we expected that the effects of classroom norms on victim-oriented defending would be stronger for girls and that their effects on bully-oriented defending would be stronger for boys.

The Present Study

Our first objective was to examine the effects of known individual-level predictors of defending on both victim-oriented and bully-oriented defending. In addition to likability, perceived popularity, and gender, we tested the effects of bullying. We hypothesized that higher likability, lower bullying, and being a girl would predict victim-oriented defending, whereas higher perceived popularity, higher bullying, and being a boy would predict bully-oriented defending. We hypothesized that the links between individual peer status and defending would be moderated by gender; specifically, we expected the positive associations between status and victim-oriented defending to be stronger for girls, and the positive associations between status and bully-oriented defending to be stronger for boys. Further, we explored the moderating effect of gender on the link between bullying and each type of defending.

Our second objective was to investigate the main effects of two classroom norms—the bullying-popularity norm and the bullying-rejection norm—on each type of defending as well as their interactive effects with likability, popularity, and gender. In line with the findings on general defending, we hypothesized that both types of defending would be facilitated in classrooms where bullies were less popular (i.e., negative main effects of the bullying-popularity norm) and in classrooms where bullies were more disliked (i.e., positive main effects of the bullying-rejection norm). We also hypothesized that better-liked adolescents would be more likely to support the victim and more popular adolescents would be more likely to confront the bully in classrooms where bullies had lower status. We did not formulate a hypothesis about the differential effects of classroom norms on the two types of defending. We considered the possibility that bully-oriented defending, compared with victim-oriented defending, would be more socially risky and, therefore, more sensitive to classroom norms or that it would be used by bullies to defend other bullies and, therefore, less socially hazardous and less dependent on classroom norms. Finally, we hypothesized that the effects of classroom norms on victim-oriented defending would be stronger for girls and that their effects on bully-oriented defending would be stronger for boys.

Method

Sample and Procedure

Cross-sectional data were collected in the spring of three consecutive years (2010–2012). The total sample included 1,460 adolescents (50% girls) from 59 classrooms in 50 schools throughout the Netherlands. Their mean age was 11.06 years (range = 9.0–14.6). Most (96.2%) were native Dutch. The participation rate was 98.2%. The students were in grades 4–6, which are the last three years of elementary school in the Dutch school system. After obtaining consent from the schools and classroom teachers, we sent the parents of the students an information letter about the research, procedures, and data storage. They returned a signed form or contacted the school if they did not want their child to participate. On the days of data collection, students were given the option of declining or opting out of participation at any time, which none of them did.

The peer nomination procedures were individually administered by a research assistant in a 30-min interview session. Interviews took place in a quiet room in the participants’ own school. Participants were told that all information they gave would remain confidential. In addition, they were urged not to talk about the survey with their peers. Participants were shown a list of their classmates to use as a reference in making their nominations. They could nominate an unlimited number of classmates of both sexes and could choose to nominate nobody.

Measures

Two types of defending

Both victim-oriented defending and bully-oriented defending were assessed using a single peer-nomination item from the Bullying Role Nomination Procedure (BRNP; Olthof et al., 2011). For victim-oriented defending, participants were presented with the following description: “There are children who want to help another child when he or she is being victimized. These children tell the victim to ignore the bullies; they console the victim afterwards; they are friendly to the victim during recess; they go and see an adult to talk about the bullying; or they go and tell the teacher about the bullying.” They were then asked “Do you know any children in your classroom who try to help the victim in any of these ways? Can you give me their names?”

Bully-oriented defending was described as follows: There are also children who get angry when another child is being bullied. These children do not hesitate and immediately go after the bully and try to stop them with the aim of helping the victim. Participants were then asked “Do you know any children in your classroom who immediately go after the bully in order to stop him or her? Can you give me their names?” Proportion scores for each type of defending were obtained by dividing the number of nominations received by the number of nominators.

Peer status

Participants nominated the classmates they liked the most and the least as well as the classmates they considered the most popular. Likability was operationalized as the proportion of liked-most nominations, rejection as the proportion of liked-least nominations, and perceived popularity as the proportion of most-popular nominations.

Bullying

Perpetration of bullying was also assessed using the BRNP (Olthof et al., 2011). Participants were given a definition of bullying, which included its defining features: intention to harm, repetition, and power differential. Five types of bullying were assessed using five peer nomination items: physical, verbal, material, direct relational (e.g., ignoring), and indirect relational (e.g., saying nasty things about someone to damage their reputation). Participants were provided with an explanation of each type of bullying and asked, “Do you know which classmates carry out that particular form of bullying?” and “If so, could you give us their names?” Proportion scores were computed for each type of bullying and averaged across the 5 items (α = .89).

Classroom-level variables

Classroom norms were the bullying-popularity norm, operationalized as the within-classroom correlation between bullying and perceived popularity, and the bullying-rejection norm, operationalized as the within-classroom correlation between bullying and rejection. In addition, we controlled for class size (i.e., number of students per classroom), as most of our variables were proportion scores based on peer nominations and these can be affected by classroom size (Velásquez et al., 2013).

Demographic variables

All analyses included the effects of gender (1 = boy, 0 = girl), as we expected gender to be differentially associated with victim-oriented defending and bully-oriented defending. As the age range of our sample was relatively large, age (in years) was included as a covariate to ensure that the main effects of interest were not confounded by age.

Analysis Plan

We conducted multilevel analyses to account for the clustered nature of our data (i.e., students nested within classrooms) and to examine the effects of age, gender, likability, popularity, and bullying (Level 1), as well as classroom bullying norms and classroom size (Level 2), on victim-oriented and bully-oriented defending. For each type of defending, we tested four multilevel models, using Mplus 7.3 (Muthén & Muthén, 1998–2012). First, we examined the effects of individual-level predictors—age, gender, likability, perceived popularity, and bullying—on each type of defending (Model 1). Second, we tested whether the effects of each type of status and the effect of bullying were moderated by gender (Model 2). These two models included individual-level predictors only and were meant to facilitate comparisons with previous findings on the correlates of the two types of defending (Reijntjes et al., 2016). Third, we added three classroom-level predictors: bullying-popularity norm, bullying-rejection norm, and classroom size (Model 3). For significant main effects of classroom norms on defending, Wald tests were conducted to compare the size of these effects across the two types of defending. In the next step, random slopes were defined for popularity, likability, and gender by creating latent variables representing the slope of the relationship between each of these predictors and each of the two outcomes. We tested whether the variance of these random slopes was significant. Finally, to examine whether each classroom norm moderated the association between the two indices of status, gender, and each type of defending, we tested whether each classroom norm significantly predicted each of these random slopes (Model 4). These effects are described as cross-level interactions between the two classroom norms and the three individual-level predictors.

Among individual-level variables, likability and perceived popularity were centered at the classroom mean, as is recommended when cross-level interactions involving continuous variables are of interest (Enders & Tofighi, 2007). All classroom-level variables, as well as age, gender, and bullying, were grand-mean centered. We used maximum likelihood estimators. The intra-class correlations indicated that 10.0% of the variance in victim-oriented defending and 15.3% of the variance in bully-oriented defending was due to between-classroom differences.

Table 1 provides descriptive statistics and correlations for all study variables. Tables 2 and 3 present the unstandardized estimates for the multilevel analyses. Standardized estimates are provided in the text. We conducted post hoc probing of all interactions—among Level 1 predictors and cross-level interactions (Preacher et al., 2006).

Table 1.

Means (and Standard Deviations), Range, and Correlations for the Main Study Variables (Uncentered).

	M (SD)	Range	1.	2.	3.	4.	5.
Individual-level variables (N = 1,460)
1. Age (years)	11.06 (0.95)	8.86 to 14.55	—
2. Victim-oriented defending	0.09 (0.11)	0.00 to 0.68	−.06*	—
3. Bully-oriented defending	0.05 (0.07)	0.00 to 0.50	.03	.43***	—
4. Likability	0.17 (0.11)	0.00 to 0.65	−.02	.40***	.18***	—
5. Perceived popularity	0.12 (0.19)	0.00 to 1.00	.18***	.09**	.30***	.21***	—
6. Bullying	0.05 (0.09)	0.00 to 0.67	.00	−.14***	.24***	−.09**	.46***
Classroom-level variables (N = 59)
1. Victim-oriented defending	0.09 (0.04)	0.01 to 0.20	—
2. Bully-oriented defending	0.05 (0.03)	0.00 to 0.14	.76***	—
3. Bullying-popularity norm	0.48 (0.33)	−0.71 to 0.97	−.10	−.10	—
4. Bullying-rejection norm	0.62 (0.31)	−0.56 to 0.96	.39**	.18	.13	—
5. Size	24.75 (4.11)	15 to 33	−.46***	−.48***	.12	.11	—

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

Table 2.

Unstandardized Estimates, Confidence Intervals, and p Values for Models Predicting Victim-Oriented Defending.

	Model 1			Model 2			Model 3			Model 4
	Est.	95% CI	p Value	Est.	95% CI	p Value	Est.	95% CI	p Value	Est.	95% CI	p Value
Intercept	.093	.081, .103	<.001	.091	.079, .102	<.001	.091	.082, .100	<.001	.090	.081, .098	<.001
Individual-level predictors
Age	.000	−.007, .007	.985	.000	−.006, .007	.890	.001	−.006, .007	.873	−.001	−.007, .005	.798
Gender	−.057	−.067, −.048	<.001	−.055	−.064, −.045	<.001	−.054	−.063, −.044	<.001	−.057	−.070, −.043	<.001
Likability	.324	.274, .373	<.001	.316	.267, .365	<.001	.314	.265, .363	<.001	.311	.249, .373	<.001
Popularity	.052	.023, .082	<.001	.064	.035, .094	<.001	.066	.036, .095	<.001	.051	.014, .088	.007
Bullying	−.178	−.246, −.110	<.001	−.241	−.323, −.160	<.001	−.250	−.331, −.170	<.001	−.203	−.287, −.120	<.001
Gender × Likability				−.147	−.246, −.047	.004	−.141	−.240, −.042	.005	−.173	−.271, −.076	<.001
Gender × Popularity				−.074	−.133, −.016	.012	−.074	−.132, −.016	.013	−.101	−.159, −.043	.001
Gender × Bullying				.203	.049, .356	.010	.205	.052, .358	.009	.302	.149, .455	<.001
Classroom-level predictors
Norm bullying-popularity							−.015	−.043, .012	.271	−.015	−.042, .012	.265
Norm bullying-rejection							.064	.035, .093	<.001	.061	.033, .089	<.001
Classroom size							−.006	−.008, −.004	<.001	−.005	−.007, −.003	<.001
Cross-level interactions
Norm Bullying-Pop × Like										−.133	−.331, .064	.186
Norm Bullying-Pop × Pop										−.027	−.141, .087	.641
Norm Bullying-Pop × Gender										.008	−.035, .051	.713
Norm Bullying-Reject × Like										.363	.160, .567	<.001
Norm Bullying-Reject × Pop										.161	.041, .280	.009
Norm Bullying-Reject × Gender										−.052	−.096, −.007	.023
Residual var. (within)	.008	.007, .008	<.001	.007	.007, .008	<.001	.007	.007, .008	<.001	.006	.006, .007	<.001
Residual var. (between): intercept	.002	.001, .002	<.001	.002	.001, .002	<.001	.001	.000, .001	<.001	.001	.000, .001	<.001
Residual var. (between): slope-like										.025	.004, .046	.019
Residual var. (between): slope-pop										.007	.002, .013	.008
Residual var. (between): slope-gender										.002	.001, .003	.002
Fit statistics
Loglikelihood	1,409.227			1,422.217			1,439.722			1,516.041

Note. N = 1,434. The likability and popularity variables are classroom-mean centered; all other variables are grand-mean centered. Gender was coded as 1 for boys, and 0 for girls. Var = variance; pop = popularity; like = likability; reject = rejection.

Table 3.

Unstandardized Estimates, Confidence Intervals, and p Values for Models Predicting Bully-Oriented Defending.

	Model 1			Model 2			Model 3			Model 4
	Est.	95% CI	p Value	Est.	95% CI	p Value	Est.	95% CI	p Value	Est.	95% CI	p Value
Intercept	.048	.041, .056	<.001	.047	.040, .055	<.001	.048	.041, .054	<.001	.048	.041, .054	<.001
Individual-level predictors
Age	.003	−.001, .008	.170	.003	−.002, .008	.225	.003	−.001, .008	.189	.002	−.003, .006	.438
Gender	.012	.006, .018	<.001	.014	.008, .021	<.001	.014	.008, .021	<.001	.013	.004, .022	.006
Likability	.088	.055, .122	<.001	.087	.054, .121	<.001	.086	.052, .119	<.001	.077	.031, .124	.001
Popularity	.068	.048, .088	<.001	.068	.048, .089	<.001	.070	.050, .090	<.001	.067	.035, .098	<.001
Bullying	.076	.031, .122	.001	.031	−.024, .086	.273	.024	−.031, .079	.386	.057	.000, .114	.048
Gender × Likability				.103	.036, .171	.003	.103	.036, .171	.003	.082	.016, .147	.014
Gender × Popularity				.006	−.034, .045	.780	.006	−.033, .171	.762	−.006	−.046, .034	.767
Gender × Bullying				.151	.047, .255	.005	.152	.048, .256	.004	.182	.077, .287	.001
Classroom-level predictors
Norm bullying-popularity							−.011	−.031, .010	.305	−.010	−.031, .010	.313
Norm bullying-rejection							.026	.005, .048	.015	.024	.003, .045	.025
Classroom size							−.003	−.005, −.002	<.001	−.004	−.005, −.002	<.001
Cross-level interactions
Norm Bullying-Pop × Like										−.048	−.197, .100	.524
Norm Bullying-Pop × Pop										−.125	−.225, −.025	.014
Norm Bullying-Pop × Gender										−.016	−.045, .013	.271
Norm Bullying-Reject × Like										.081	−.072, .235	.299
Norm Bullying-Reject × Pop										.051	−.053, .156	.336
Norm Bullying-Reject × Gender										.021	−.009, .051	.161
Residual var. (within)	.003	.003, .004	<.001	.003	.003, .004	<.001	.003	.003, .004	<.001	.003	.002, .003	<.001
Residual var. (between): intercept	.001	.000, .001	<.001	.001	.000, .001	<.001	.000	.000, .001	<.001	.000	.000, .001	<.001
Residual var. (between): slope-like										.018	.007, .029	.002
Residual var. (between): slope-pop										.009	.004, .014	<.001
Residual var. (between): slope-gender										.001	.000, .001	.001
Fit statistics										.048
Loglikelihood	1,969.780			1,979.398			1,989.473			2,078.121

Results

Victim-Oriented Defending

Effects of individual-level predictors

Model 1 (in Table 2) tested only the main effects of individual-level predictors—age, gender, likability, popularity, and bullying—on victim-oriented defending. Around 27% of the within-classroom variance in victim-oriented defending was explained by these predictors. There was no significant effect of age. Higher popularity, higher likability, lower bullying, and being a girl were associated with more victim-oriented defending. However, the effects of popularity, likability, and bullying were moderated by gender, as shown by the significant interactions in Model 2. The association between popularity and victim-oriented defending was positive for girls, γ = .101, SE = .023, p <.001, but was not significant for boys, γ = .027, SE = .020, p = .174; the association between likability and victim-oriented defending was stronger for girls, γ = .390, SE = .035, p <.001, than for boys, γ = .243, SE = .037, p <.001, and the negative association between bullying and victim-oriented defending was stronger for girls, γ = −.343, SE = .061, p <.001, than for boys, γ = −.140, SE = .014, p = <.001. The unexplained within-classroom variance in victim-oriented defending was reduced by only 1.4% by the addition of these gender interactions.

Effects of classroom-level predictors

Model 3 shows the effects of three classroom-level predictors—bullying-popularity norm, bullying-rejection norm, and classroom size. They explained around 30% of the between-classroom variance in victim-oriented defending. Contrary to our expectations, the data did not show significant evidence that the bullying-popularity norm was related to victim-oriented defending, γ = −.122, SE = .110, p = .271. Consistent with our hypothesis, there was a positive main effect of the bullying-rejection norm, γ = .486, SE = .106, p = <.001: Adolescents engaged in more victim-oriented defending in classrooms where bullies were more disliked. Victim-oriented defending was also higher in smaller classrooms, γ = −.594, SE = .095, p = <.001.

We built a model (not shown in Table 2) to estimate the variance of three latent slopes for likability, popularity, and gender to examine the extent to which the association of each of these variables with victim-oriented defending varied between classrooms. We also estimated the covariances between the slopes and the intercept. All variances were significant: Var = .042, SE = .014, p = .002 for likability, Var = .010, SE = .003, p = .003 for popularity, and Var = .002, SE = .001, p = .001 for gender.

Model 4 shows that the effects of likability, popularity, and gender on victim-oriented defending varied across classrooms depending on the classroom bullying-rejection norm. These significant interactions were decomposed using simple slopes at high (+ 1 SD) and low (−1 SD) levels of the bullying-rejection norm variable, which corresponded to classroom correlations between bullying and rejection of .93 and .30, respectively. The probing was calculated using the computational tools provided by Preacher et al. (2006); the covariates included in the analyses did not play a role in the probing of the interactions. As shown in Figure 1, the positive association between likability and victim-oriented defending was stronger in classrooms where bullies were more rejected, γ = .425, SE = .046, p < .001, than in classrooms where bullies were less rejected, γ = .197, SE = .046, p < .001. The effect of the classroom bullying-rejection norm on victim-oriented defending was stronger for students high in likability (+1 SD), γ = .097, SE = .021, p < .001, than for students low in likability (−1 SD), γ = .025, SE = .013, p = .045. The bullying-rejection norm explained around 38% of the variation in the relation between likability and victim-oriented defending that was observed over classrooms. Regarding the moderating effects of the bullying-rejection norm on the association between perceived popularity and victim-oriented defending (see Figure 2), popular students were found to defend more in classrooms where bullies were more rejected, γ = .102, SE = .027, p < .001; there was no statistically significant evidence of a link between perceived popularity and victim-oriented defending in classrooms where bullies were less rejected, γ = .003, SE = .027, p = .991. Similar to our findings on likability, the bullying-rejection norm mattered for the level of victim-oriented defending of students high in popularity (+1 SD), γ = .091, SE = .017, p < .001, but there was no statistically significant evidence that it mattered for the level of victim-oriented defending of students low in popularity (−1 SD), γ = .031, SE = .019, p = .108. The bullying-rejection norm explained around 23% of the between-classroom variation in the association between popularity and victim-oriented defending.

Figure 1.

Moderating Effects of Classroom Bullying-Rejection Norm on the Association Between Classroom Mean-Centered Likability and Victim-Oriented Defending.

Figure 2.

Moderating Effects of Classroom Bullying-Rejection Norm on the Association Between Classroom Mean-Centered Popularity and Victim-Oriented Defending.

Finally, the significant interaction between gender and the bullying-rejection norm suggested that adolescents of both genders defended more in classrooms where bullies were more disliked, but this association was stronger for girls, γ = .087, SE = .020, p < .001, than for boys, γ = .035, SE = .016, p = .031. The bullying-rejection norm explained around 16% of the between-classroom variation in the effect of gender on victim-oriented defending. No significant moderating effect of the classroom bullying-popularity norm on the associations of likability, popularity, and gender with victim-oriented defending was found.

Bully-Oriented Defending

Effects of individual-level predictors

Around 11% of the within-classroom variance in bully-oriented defending was explained by the Level 1 predictors in Model 1. Results from this model indicated that adolescents higher in popularity and adolescents higher in likability engaged in more bully-oriented defending. There was no significant effect of age. In contrast to the findings obtained for victim-oriented defending, our analyses showed that being a boy and engaging in more bullying was associated with more bully-oriented defending. As shown in Model 2, the effects of likability and bullying were moderated by gender: The association between likability and bully-oriented defending was positive for boys, γ = .139, SE = .025, p < .001, and nonsignificant for girls, γ = .036, SE = .023, p = .127, as was the association between bullying and bully-oriented defending (γ = .107, SE = .026, p < .001 for boys and γ = −.045, SE = .048, p = .354 for girls). That is, boys engaged in more bully-oriented defending when they themselves had higher levels of bullying and when they were more liked by peers; there was no statistically significant evidence that this was the case for girls. The unexplained within-classroom variance in bully-oriented defending was reduced by 1.3% by the addition of these gender interactions.

Effects of classroom-level predictors

Model 3 tested the effects of the three classroom-level variables on bully-oriented defending. Together, they explained approximately 34% of the between-classroom variance in bully-oriented defending. Similar to our findings for victim-oriented defending, there was no statistically significant effect of the bullying-popularity norm on bully-oriented defending, γ = −.129, SE = .125, p = .305, but there was a positive main effect of the bullying-rejection norm, γ = .308, SE = .123, p = .015: Adolescents engaged in more bully-oriented defending in classrooms where bullies were more disliked. A Wald test revealed that the positive effect of the bullying-rejection norm on defending was larger for victim-oriented defending than for bully-oriented defending, χ²(1, 1434) = 11.23, p < .001. There was a negative effect of classroom size on bully-oriented defending, γ = −.525, SE = .107, p = <.001.

As for victim-oriented defending, we ran an additional model (not shown in Table 3) to estimate the variance of three latent slopes for likability, popularity, and gender as well as the covariances between the slopes and the intercept. All variances were significant: Var = .018, SE = .006, p =.002 for likability; Var = .010, SE = .003, p < .001 for popularity; and Var = .001, SE = .000, p = .001 for gender.

Model 4 tested the potential moderating effects of the bullying-popularity norm and the bullying-rejection norm on the associations between the main predictors of interest and bully-oriented defending. Only the interaction between the bullying-popularity norm and perceived popularity was significant (see Figure 3). As for victim-oriented defending, the interaction was probed using the tools provided by Preacher et al. (2006); covariates were ignored in the probing. Cutoffs of +1 SD and −1 SD were used to represent classrooms with a high bullying-popularity norm (r = .82) and a low bullying-popularity norm (r = .16). The association between popularity and bully-oriented defending was positive in classrooms where bullies were less popular, γ = .108, SE = .023, p < .001, but not statistically significant in classrooms where bullies were more popular, γ = .026, SE = .023, p = .256. Students high in popularity (+1 SD) engaged in more bully-oriented defending in classrooms where bullies were less popular compared with other classrooms, γ = .108, SE = .023, p < .001; for students low in popularity (−1 SD), this classroom norm was not significantly associated with their levels of bully-oriented defending, γ = .013, SE = .011, p = .209. The bullying popularity-norm explained around 10% of the between-classroom variation in the association between popularity and bully-oriented defending.

Figure 3.

Moderating Effects of Classroom Bullying-Popularity Norm on the Association Between Classroom Mean-Centered Popularity and Bully-Oriented Defending.

Discussion

There is growing evidence that defending victims of bullying is influenced by classroom norms (e.g., Yun & Graham, 2018). With an eye to better understanding the role of the social context in defending, our aim was to elucidate the effect of bullies’ status in the classroom while taking into account the heterogeneity of defending behaviors. We examined support for the victim (victim-oriented defending) and confrontation of the bully (bully-oriented defending) separately. In addition to investigating the differential associations of peer status, gender, and bullying with these two types of defending, we tested whether the classroom bullying-popularity norm—how popular bullies are—and the bullying-rejection norm—how disliked bullies are—predicted each type of defending and moderated the effects of individual peer status and gender on each type of defending, among early adolescents.

In line with earlier findings (Reijntjes et al., 2016), higher likability, lower bullying, and being a girl were associated with victim-oriented defending, and higher popularity, higher bullying, and being a boy were associated with bully-oriented defending. However, higher popularity was also linked to victim-oriented defending and higher likability to bully-oriented defending. Moreover, positive links between each type of status and victim-oriented defending, and the negative link between bullying and victim-oriented defending, were stronger for girls. The effects of likability and bullying on bully-oriented defending were significant and positive for boys only. Although the effects were small, these gender interactions support the idea that defending is better rewarded by peers when done in a way that is more normative for the gender of the defender (Chang, 2004).

We expected that in classrooms where bullies were more popular and in classrooms where they were less disliked, levels of both types of defending would be lower. Our findings for the bullying-rejection norm were in line with this expectation: Both types of defending were more prevalent in classes where bullies were more disliked. Our analyses did not detect these effects for the bullying-popularity norm. At this point, we cannot offer a plausible explanation as to why bullies’ popularity might matter less for defending than bullies’ rejection. Further research on the impact of these social norms on defending is necessary.

Concerning the differential effects of classroom norms on the two types of defending, there was some indication that victim-oriented defending might be more sensitive to classroom norms—at least the bullying-rejection norm—than bully-oriented defending. This runs counter to the idea that bully-oriented defending is more socially risky and, therefore, more dependent on the status of bullies in the classroom. Some bully-oriented defenders may be affiliates of the bullies and their defending may aim at standing up for their bullying friends only. Future studies examining defender-victim dyads are necessary before any conclusion can be drawn.

As expected, the bullying-rejection norm moderated the relation between individual status and victim-oriented defending: Being more popular and being better liked were associated with more victim-oriented defending in classes where bullies were more rejected. When perpetrators do not benefit from the social approval of their peers, it might prompt popular peers, and especially well-liked peers, to use their high social position to support the victims. Regarding bully-oriented defending, we found a moderating effect of the bullying-popularity norm in the expected direction. Being popular was associated with more bully-oriented defending only in classes where bullies were less popular. High-status adolescents may be more inclined to take advantage of their privileged position and defend peers in contexts where the balance of power is less in favor of the bullies.

Our findings suggest that gender may be important to consider when investigating different types of defending. Not only did associations among individual-level variables show that each type of defending may be more normative for one gender, but there was an indication that girls’ victim-oriented defending behavior was more sensitive to bullies’ status in the classroom than boys’ victim-oriented defending. This might be related to girls’ stronger fear of negative social evaluation (see Rose & Rudolph, 2006), which could make them more acutely aware of the status of perpetrators and more likely to adjust their own behavior accordingly.

Limitations

To the best of our knowledge, our study is the first to investigate the effects of classroom bullying status norms on different types of defending behavior and to consider both popularity and rejection in the operationalization of these norms. Nevertheless, it has several limitations. First, its cross-sectional design did not allow us to disentangle the direction of effects between individual status and defending. High status may be a consequence rather than a predictor of defending. This might also depend on the type of defending. For example, being popular might be a prerequisite for confronting bullies, but a reward for supporting victims. These longitudinal effects might in turn be moderated by classroom norms. Only longitudinal multilevel analyses can shed light on these associations.

A second limitation concerns the lack of dyadic data. To better test whether adolescents are less likely to defend victims when the perpetrators have high status, it would be necessary to collect and analyze data on who defends against whom. A strong correlation between bullying and popularity in the class does not imply that each bully is popular (or each popular classmate is a bully). Furthermore, we made the assumption that some of the adolescents engaging in bully-oriented defending may do so to act on behalf of their own bullying friends, who may be targeted by classmates outside of their peer group. Only dyadic data and social network analyses distinguishing between types of defending can establish who defends whom (Huitsing et al., 2014) and provide a test of the effects of classroom status norms on bully-oriented defending directed at victims.

Finally, the use of peer nominations for the operationalization of status and bullying behavior has disadvantages. We theorized that high peer status was a prerequisite for defending as the awareness of being liked or being popular should make adolescents feel secure enough to challenge bullies. However, self-reports of status would capture these feelings of being liked or popular more accurately than peer reports. Future research on these questions would benefit from examination of the role of self-perceived status. Peer reports may also be problematic for assessing bullying, because they rely on participants reporting on their classmates and might make them afraid of being labeled a “snitch.” Nonetheless, our study used individual interviews, which can prevent children from feeling reluctant to answer (e.g., a child feeling his classmates’ eyes on his back): children were found to nominate fewer classmates using an Internet version of the questionnaire (Pronk et al., 2013) than using the interview version. Therefore, it is unlikely that the use of interviews deterred participants from nominating bullies.

Conclusions

Our findings show that bullies’ status in the classroom may affect classmates’ decision to stand up for victims of bullying as well as the degree to which popular and well-liked adolescents engage in defending. How disliked bullies are in the classroom mattered more for defending behavior than bullies’ level of popularity. Defending by supporting victims was more strongly related to classroom norms than was confronting bullies. Finally, tests of the relative effectiveness of each type of defending in decreasing bullying, and in improving victims’ adjustment, are necessary to translate these findings into practical suggestions for anti-bullying interventions.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was partially supported by the INVEST Research Flagship Center Academy of Finland Flagship Program, decision number: 320162.

ORCID iD

Claire F. Garandeau

References

Barchia

Bussey

(2011). Predictors of student defenders or peer aggression victims: Empathy and social cognitive factors. International Journal of Behavioral Development, 35, 289–297.

Brody

(1999). Gender, emotion, and the family. Harvard University Press.

Caravita

S. C. S.

Di Blasio

Salmivalli

(2009). Unique and interactive effects of empathy and social status on involvement in bullying. Social Development, 18, 140–163.

Caravita

S. C. S.

Di Blasio

Salmivalli

(2010). Early adolescents’ participation in bullying: Is ToM involved? Journal of Early Adolescence, 30, 138–170.

Chang

(2004). The role of classroom norms in contextualizing the relations of children’s social behaviors to peer acceptance. Developmental Psychology, 40, 691–702.

Dijkstra

J. K.

Cillessen

A. H. N.

Lindenberg

Veenstra

(2010). Basking in reflected glory and its limits: Why adolescents hang out with popular peers. Journal of Research on Adolescence, 20, 942–958.

Dijkstra

J. K

Gest

S. D.

(2015). Peer norm salience for achievement, prosocial behavior and bullying: Implications for adolescent school experiences. Journal of Early Adolescence, 35, 1–18.

Duffy

A. L.

Penn

Nesdale

Zimmer-Gembeck

M. J.

(2017). Popularity: Does it magnify associations between popularity prioritization and the bullying and defending behavior of early adolescent boys and girls? Social Development, 26, 263–277.

Enders

C. K.

Tofighi

(2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12, 121–138.

10.

Galen

B. R.

Underwood

M. K.

(1997). A developmental investigation of social aggression among children. Developmental Psychology, 33, 589–600.

11.

Garandeau

C. F.

Lee

I. A.

Salmivalli

(2014). Differential effects of the KiVa anti-bullying program on popular and unpopular bullies. Journal of Applied Developmental Psychology, 35, 44–50.

12.

Hawkins

D. L.

Pepler

D. J.

Craig

W. M.

(2001). Naturalistic observations of peer interventions in bullying. Social Development, 10, 512–527.

13.

D. Y.

(1976). On the concept of face. American Journal of Sociology, 81, 867–884.

14.

Huitsing

Snijders

T. A. B.

van Duijn

M. A. J.

Veenstra

(2014). Victims, bullies, and their defenders: A longitudinal study of the coevolution of positive and negative networks. Development and Psychopathology, 26, 645–659.

15.

Juvonen

Galván

(2008). Peer influence in involuntary social groups: Lessons from research on bullying. In Prinstein

M. J.

Dodge

K. A.

(Eds.), Understanding peer influence in children and adolescents (pp. 225–244). Guilford.

16.

LaFontana

K. M.

Cillessen

A. H. N.

(2010). Developmental changes in the priority of perceived status in childhood and adolescence. Social Development, 19, 130–147.

17.

Marks

P. E. L.

Cillessen

A. H. N.

Crick

N. R.

(2012). Popularity contagion among adolescents. Social Development, 21, 501–521.

18.

Meter

D. J.

Card

N. A.

(2015). Effects of defending: The longitudinal relations among defending victimized peers, own victimization and peer-perceived liking. Social Development, 24, 734–747.

19.

Muthén

L. K.

Muthén

B. O.

(1998–2012). Mplus user’s guide (6th ed.). Muthén & Muthén.

20.

Olthof

Goossens

F. A.

Vermande

M. M.

Aleva

E. A.

van der Meulen

(2011). Bullying as strategic behavior: Relations with desired and acquired dominance in the peer group. Journal of School Psychology, 49, 339–359.

21.

Peets

Pöyhönen

Juvonen

Salmivalli

(2015). Classroom norms of bullying alter the degree to which children defend in response to their affective empathy and power. Developmental Psychology, 51, 913–920.

22.

Pöyhönen

Juvonen

Salmivalli

(2010). What does it take to stand up for the victim of bullying? The interplay between personal and social factors. Merrill-Palmer Quarterly, 56, 143–163.

23.

Preacher

K. J.

Curran

P. J.

Bauer

D. J.

(2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437–448.

24.

Pronk

Goossens

F. A.

Olthof

De Mey

Willemen

A. M.

(2013). Children’s intervention strategies in situations of victimization by bullying: Social cognitions of outsiders versus defenders. Journal of School Psychology, 51, 669–682.

25.

Pronk

Olthof

Goossens

F. A.

(2015). Differential personality correlates of early adolescents’ bullying-related outsider and defender behavior. Journal of Early Adolescence, 35, 1069–1091.

26.

Pronk

Olthof

Goossens

F. A.

(2016). Factors influencing interventions on behalf of victims of bullying: A counterfactual approach to the social cognitions of outsiders and defenders. Journal of Early Adolescence, 36, 267–291

27.

Reijntjes

Kamphuis

J. H.

Prinzie

Telch

M. J.

(2010). Peer victimization and internalizing problems in children: A meta-analysis of longitudinal studies. Child Abuse & Neglect, 34, 244–252.

28.

Reijntjes

Vermande

Olthof

Goossens

F. A.

Aleva

van der Meulen

(2016). Defending victimized peers: Opposing the bully, supporting the victim, or both? Aggressive Behavior, 42, 585–597.

29.

Rose

Rudolph

K. D.

(2006). A review of sex differences in peer relationship processes potential trade-offs for the emotional and behavioral of girls and boys. Psychological Bulletin, 132, 98–131.

30.

Salmivalli

Voeten

Poskiparta

(2011). Bystanders matter: Associations between reinforcing, defending, and the frequency of bullying behavior in classrooms. Journal of Clinical Child and Adolescent Psychology, 40, 668–676.

31.

van der Ploeg

Kretschmer

Salmivalli

Veenstra

(2017). Defending victims: What does it take to intervene in bullying and how is it rewarded by peers? Journal of School Psychology. 65, 1–10.

32.

Velásquez

A. M.

Bukowski

W. M.

Saldarriaga

L. M.

(2013). Adjusting for group size effects in peer nomination data. Social Development, 22, 845–863.

33.

Yun

H. Y.

Graham

(2018). Defending victims of bullying in early adolescence: A multilevel analysis. Journal of Youth and Adolescence, 47, 1926–1937.