Individual and class justification of cyberbullying and cyberbullying perpetration: A longitudinal analysis among adolescents

Journal of Applied Developmental Psychology 44 (2016) 81–89

Contents lists available at ScienceDirect

Journal of Applied Developmental Psychology

Individual and class justification of cyberbullying and cyberbullying perpetration: A longitudinal analysis among adolescents Manuel Gámez-Guadix a,⁎, Gianluca Gini b a b

Department of Biological and Health Psychology, Autonomous University of Madrid, Spain Department of Developmental and Social Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy

a r t i c l e

i n f o

Article history: Received 12 June 2015 Received in revised form 25 March 2016 Accepted 7 April 2016 Available online xxxx Keywords: Cyberbullying Justification Impulsivity Class norms Adolescence

a b s t r a c t The main aim of the study was to investigate the role of individual and class justification of cyberbullying in predicting adolescents' cyberbullying perpetration over 6 months. The effects and moderating role of impulsivity, age, and gender in the hypothesized relationship between justification and cyberbullying were also tested. A sample of 750 Spanish adolescents (453 girls; mean age = 14.76; SD = 0.96) completed self-report measures at two time points during the same school year. Results from hierarchical linear modeling showed that individual-level cyberbullying justification at Time 1 significantly predicted higher levels of cyberbullying perpetration at Time 2 but only at low levels of impulsivity. Class-level justification significantly explained betweenclasses variability in cyberbullying perpetration at Time 2. Interestingly, this effect is moderated by age, indicating that the role of class justification was significant only for younger adolescents. Intervention efforts to prevent cyberbullying should center around the peer group at the class level and start during early adolescence. © 2016 Elsevier Inc. All rights reserved.

1. Introduction Cyberbullying, a serious societal problem that affects many youth, is defined as repetitive, aggressive, intentional behavior carried out by an individual or group using electronic means (e.g., the Internet, mobile phones) against victims who cannot easily defend themselves (Hinduja & Patchin, 2008; Smith et al., 2008). This behavior has been linked to negative outcomes for victims' psychosocial adjustment, including depression (Gámez-Guadix, Gini, & Calvete, 2015; GámezGuadix, Orue, Smith, & Calvete, 2013), drug and alcohol abuse (Vieno, Gini, & Santinello, 2011), and suicide ideation and attempts (Gini & Espelage, 2014; van Geel, Vedder, & Tanilon, 2014). A key to preventing cyberbullying is identifying its predictive risk factors. However, understanding of cyberbullying and the contributing processes is incomplete. From a broad perspective, the general aggression model (GAM; Anderson & Bushman, 2002; Kowalski, Giumetti, Schroeder, & Lattanner, 2014) and, more specifically, the social cognitive framework can help explain this form of peer aggression. This framework suggests that individuals' cognitions regarding behavior (e.g., justification of cyberbullying) play a central role in aggressive actions and the stability of that behavior over time (Crick & Dodge, 1994; Huesmann & Eron, 1984). Social cognitive theories suggest that, in addition to individual characteristics, social processes and contextual ⁎ Corresponding author at: Department of Biological and Health Psychology, Autonomous University of Madrid, 28049 Madrid, Spain. E-mail addresses: [email protected] (M. Gámez-Guadix), [email protected] (G. Gini).

http://dx.doi.org/10.1016/j.appdev.2016.04.001 0193-3973/© 2016 Elsevier Inc. All rights reserved.

variables, such as group justification of aggression, can influence behavior in peer relationships (e.g. Caravita, Sijtsema, Rambaran, & Gini, 2013, Faris & Ennett, 2012, Salmivalli, 2010). Furthermore, although the GAM relies on cognitions to explain cyberbullying, it provides a comprehensive framework that integrates other situational and personal factors, such as impulsivity, sex, and age (Kowalski et al., 2014). Based on the social cognitive framework, this study is aimed at advancing knowledge of the variables affecting cyberbullying. To predict perpetration, the role of individual and class cyberbullying justifications and their interplay with other important personal factors (e.g., impulsivity) are analyzed from a longitudinal and multilevel perspective. In the following section, we explain the theoretical and empirical bases for these aims.

2. Individual justification and cyberbullying Social cognitive theories of aggressive behavior have been widely used to explain traditional forms of aggression, including peer bullying (Swearer, Wang, Berry, & Myers, 2014). An important tenet of these theories is that people store in their memory certain knowledge structures based on their life experiences (“schemas” or “scripts,” Huesmann, 1988; “database,” Crick & Dodge, 1994). These structures affect future behavior and regulate actions by establishing allowable or prohibited behaviors. In the case of aggressive behavior, previous research has uncovered schemas related to justification of the use of violence that have significant associations with actual aggressive behaviors (Bosworth, Espelage, & Simon, 1999; Calvete, 2008; Huesmann & Guerra, 1997).

82

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

Justifications of aggression have also been positively associated with cyberbullying perpetration (Calvete, Orue, Estévez, Villardón, & Padilla, 2010; Heirman & Walrave, 2012; Williams & Guerra, 2007). Among the few longitudinal studies conducted to date, Barlett and colleagues (Barlett & Gentile, 2012; Barlett et al., 2014; Barlett, Gentile, & Chew, 2014) found that attitudes justifying cyberbullying were associated with cyberbullying perpetration two months later. However, these studies focused on young adults (college students), and little is known about the longitudinal relationship between justification and cyberbullying in adolescents. A recent meta-analysis (Kowalski et al., 2014) found a medium association (r = .37) between cyberbullying and justifications of aggressive behavior. However, this finding was limited by the scarcity of longitudinal studies that have examined this relationship in adolescents, and researchers have called for further investigations that “explore the issue of moral justifications in online aggressive relationships” (Gini, Pozzoli, & Bussey, 2014, p. 64). Most previous studies on cyberbullying have used cross-sectional designs. However, longitudinal studies allow more stringent analyses of the temporal relationships of variables and minimize the risk of common method biases (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Therefore, to contribute to the growing body of knowledge on the risk factors for cyberbullying, we analyze the short-term longitudinal association between adolescents' actual cyberbullying behavior and their tendency to justify aggressive behaviors in online peer relationships. Consistent with the positive association between justification of aggressive behavior and traditional bullying and cyberbullying reported in previous studies (Gini et al., 2014; Kowalski et al., 2014), we hypothesize that higher levels of cyberbullying justifications at Time 1 (T1) predict cyberbullying behavior six months later. In addition, previous studies suggest that the relationship between individual justification and cyberbullying varies according to individual characteristics, such as sex and age. For example, evidence indicates that cyberbullying perpetration and justification are both higher among boys than girls (Athanasiades & Deliyanni-Kouimtzis, 2010; Boulton, Lloyd, Down, & Marx, 2012; Calvete et al., 2010). It has also been found that cyberbullying appears to peak around eighth grade and then decline with age (Tokunaga, 2010) and that bullying justification also decreases with age (Salmivalli & Voeten, 2004). Based on these findings, we expect that the relationship between individual justification and cyberbullying perpetration to be strongest among boys and younger adolescents.

consistent with the social-ecological model widely used to study adolescent development in various life domains (Bronfenbrenner, 1979). To date, only a few researchers have conducted cross-sectional analyses of the role of class-level justifications in traditional bullying (Pozzoli, Gini, & Vieno, 2012a; Salmivalli & Voeten, 2004). These studies show that the between-class variability of bullying behavior can be partly explained by higher beliefs that justify such behavior at the class level. However, so far, little is known about class-level influences on cyberbullying. Classrooms are important socialization contexts that can influence how adolescents construct their digital worlds (Subrahmanyam & Greenfield, 2008; Subrahmanyam, Reich, Waechter, & Espinoza, 2008). This influence could be stronger for younger students than for older adolescents. For example, in Spain, younger students (in compulsory secondary education) are likely to spend more time and share more classes with the same classmates than older adolescents (in high school), who often change classrooms and classmates throughout the day. In addition, although cyberbullying does not necessarily occur on school premises, there is evidence that school factors, such as a negative school climate and low school safety, have a negative effect on it (Kowalski et al., 2014). Therefore, the second aim of this study is to test whether class justification, that is, the degree to which justifications of cyberbullying are present within a classroom, predict between-class differences in cyberbullying perpetration over time. Determining the role of group justification in cyberbullying is especially important to better design prevention efforts at the classroom level. Consistent with previous studies assessing the influence of class norms on traditional bullying (e.g. Pozzoli et al., 2012a, Salmivalli & Voeten, 2004), we anticipate a greater likelihood of cyberbullying behavior at Time 2 (T2) in classrooms with higher levels of class justification at T1. As do individual justifications, class-level justifications might also interact with important demographic variables, such as age and gender, to predict cyberbullying. For example, previous studies have found that girls tend to be more resistant to group and peer influence than boys, including in situations involving antisocial behavior (Steinberg & Monahan, 2007). Regarding age, the relevance of adherence to group norms has been found to be especially high in early adolescence but to gradually decrease during late adolescence (Rubin, Bukowski, & Parker, 1998; Steinberg & Monahan, 2007). Therefore, we hypothesize that the relationship between class justification and cyberbullying perpetration is stronger for boys than girls and for younger adolescents than older adolescents.

3. Class justifications and cyberbullying

4. Justification, impulsivity, and cybervictimization

Social cognitive theories do not explain aggressive behavior only by individual psychological processes. Social processes and contextual variables can also influence behavior in peer relationships, and some authors stress the importance of the normative context in which peer aggression takes place (e.g. Caravita et al., 2013, Faris & Ennett, 2012, Salmivalli, 2010). Social influence processes among classmates can be especially significant because classrooms are among the most important normative contexts for children and adolescents. Classrooms are characterized by a moral climate and social norms which, even when they do not reflect group members' private attitudes, implicitly or explicitly confer varying levels of approval on negative conduct, thus affecting the behavior of group members (Espelage & Swearer, 2003; Gini et al., 2014; Juvonen & Galván, 2008). A crucial aspect of the normative context for cyberbullying is the level of class justification of cyberaggression (Elledge et al., 2013). Class justification refers to the extent to which classmates develop shared injunctive beliefs regarding the appropriateness of cyberaggression. Class justification can develop through individual cognitive and affective processes, such as imitation, social comparison, competition, group conformity, and norms (e.g. Bandura, 1977, Brown, Clasen, & Eicher, 1986, Sieving, Perry, & Williams, 2000). This theory is

Although the GAM relies on cognitions to explain cyberbullying, this model also provides a comprehensive framework integrating other situational and personal factors (Anderson & Bushman, 2002). Impulsivity and cybervictimization have been identified as two important factors leading to cyberbullying encounters (Kowalski et al., 2014). Regarding impulsivity, it has been reported that an individual who does not have sufficient cognitive or emotional resources to deal with a stressful or threatening situation might act impulsively and automatically by, for example, sending an insulting or threatening message (Kowalski et al., 2014). Thus, cyberbullying is more likely to occur when individuals act impulsively without fully considering the possible consequences for the victims (Bhat, 2008). Empirical evidence suggests a positive association of impulsive traits with frequent cyberbullying behavior (Gámez-Guadix, Villa, & Calvete, 2014; Kokkinos, Antoniadou, & Markos, 2014; Vazsonyi, Machackova, Sevcikova, Smahel, & Cerna, 2012). Therefore, we expect that high levels of impulsivity are a significant risk factor that predicts a greater likelihood of cyberbullying perpetration by adolescents at T2. In addition to this main effect, impulsivity can interact with cyberbullying justification, leading to cyberbullying perpetration. A recent study showed that justifications of negative acts through

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

disengagement mechanisms and impulsivity interact to explain aggressive behavior in Italian adolescents (Gini, Pozzoli, & Bussey, 2015). Specifically, regardless of impulsivity levels, youth who reported high levels of justifications of aggressive behavior were more likely to engage in (both overt and relational) reactive aggression toward peers. Aggressive behavior by adolescents less likely to justify negative conduct was positively associated with their level of impulsivity and lack of behavioral control. In keeping with these findings, we hypothesize an interaction between justifications of cyberbullying and impulsivity, in that justifications of cyberbullying behavior could be less influential on the outcome behavior for youth high in impulsivity. Conversely, adolescents with low impulsivity are expected to rely more on their cognitions (i.e., justifications) when they act aggressively online. Therefore, we test the hypothesis that individual justification of cyberbullying more strongly predicts cyberbullying perpetration by youth with low impulsivity than their peers with high impulsivity. Regarding victimization, the GAM and previous empirical evidence indicate that experiences of victimization can activate dysfunctional appraisal and decision-making processes, which might lead to involvement in future cyberbullying perpetration (Kowalski et al., 2014). Research has shown that being a victim of cyberbullying significantly increases the likelihood of becoming a perpetrator of it (e.g., Barlett & Gentile, 2012), “perhaps triggering a chain of back-and-forth cyberbullying/cybervictimization episodes” (Kowalski et al., 2014, p. 52). Studies have also shown that victims are more likely than nonvictimized peers to think that cyberbullying perpetration is acceptable to a degree and to justify it (e.g., O'Brennan, Bradshaw, & Sawyer, 2009). Therefore, we include cybervictimization as a control variable for cyberbullying perpetration. 5. The present study Based on the social cognitive framework of aggressive behavior, we investigate whether individual and class justifications of cyberbullying contribute to explaining the levels of cyberbullying perpetration in a sample of adolescents over 6 months. We control for levels of T2 individual justification, T1 and T2 cybervictimization, and class-levels of cyberaggression to avoid confounding effects caused by the overlap of these variables with T1 individual and class justifications. We also examine the interaction of T1 individual justifications and impulsivity based on evidence (e.g., Gini et al., 2015) that individual justification is related to aggression only at low levels of impulsivity. Finally, following previous studies suggesting that the relationship between justification and cyberbullying perpetration varies as a function of age and gender, we examine whether the effects of individual and class justifications depend on age and gender. 6. Method 6.1. Participants The initial sample consisted of 969 adolescents between 13 and 18 years old recruited from 43 classrooms (average class size = 25.72, SD = 11.09) in 12 randomly selected public and private secondary schools in Bizkaia, Spain. Of these, 750 participants (453 girls, 297 boys; mean age = 14.76, SD = 0.96) completed the 2 study waves during the same school year (attrition rate = 22.6%). A series of t-tests was conducted to identify any differences in the study variables at T1 between adolescents who did and did not complete the study. None of these differences was significant. In the sample, 54.4% of participants were in the 3rd year of compulsory secondary education (the equivalent of 9th grade in the United States; 20 classrooms), 21% in the 4th year of compulsory secondary education (10th grade in the United States; 10 classrooms), and 24.6% in the 1st year of high school (11th grade in the United States; 12 classrooms).

83

Most participants were Spanish (91.2%), while the remaining were South American (6.1%), Eastern European (0.6%), African (0.7%), Asian (0.3%), or another ethnicity (1.1%). Socioeconomic levels were determined by applying the criteria recommended by the Spanish Society of Epidemiology and Family and Community Medicine, including parental occupation and income (Domingo-Salvany, Regidor, Alonso, & Alvarez-Dardet, 2000). Using these criteria, 11.7% of participants had a low, 25.8% medium-low, 22.1% medium, 23% medium-high, and 17.4% high socioeconomic level. 6.2. Measures 6.2.1. Cyberbullying questionnaire The Cyberbullying Questionnaire (CBQ; Calvete et al., 2010; GámezGuadix et al., 2014) is composed of two scales on cyberbullying perpetration and cybervictimization. The perpetration scale has 14 items that measure the frequency of aggressive online behaviors (e.g., “sending threatening or insulting messages to other people,” “posting or sending humiliating images of classmates”). The victimization scale has 9 items that measure the frequency with which adolescents experience cyberbullying behaviors (e.g., “receiving insulting or threatening messages via the Internet or a cell phone”). The response scale measures how many times each behavior had ever occurred using the following format: 0 (never), 1 (1 or 2 times), 2 (3 or 4 times), or 3 (5 or more times). The item scores for each scale were added to obtain total cyberbullying and cybervictimization scores. The psychometric properties of the CBQ among Spanish-speaking adolescents have been analyzed previously and supported factorial and convergent validity and good level of internal consistency (Gámez-Guadix et al., 2014). The internal consistency (Cronbach's alpha) for this sample was .74 at T1 and T2 for cyberbullying perpetration and .61 at T1 and .68 at T2 for cybervictimization. 6.2.2. Justification of cyberbullying scale (Gámez-Guadix et al., 2014). The cyberbullying justification scale is made up of 5 items assessing whether adolescents justify cyberbullying behaviors for various reasons. The items cover a range of ideas justifying aggression to take revenge (“threatening or insulting another person who previously did those things to you via a cell phone or the Internet”); to gain restitution (“recording or sending videos or pictures to post on the Internet of someone else who deserved it,” “posting humiliating images or links on the Internet of a classmate who deserved it”); to make fun of someone (“Sending messages, e-mails, or pictures to someone else to mock or laugh at him or her”); and to deliberately harm a person (“writing or sending jokes, rumors, gossip, or comments to ridicule a classmate”). Participants were asked to rate the extent to which they thought these behaviors are justified on a 5-point scale ranging from 1 (never justified) to 5 (always justified). This scale has been shown to have factorial and convergent validity and good internal consistency (Gámez-Guadix et al., 2014). Given the importance of this measure to the present study, we analyzed the factorial validity of this scale in the current sample by performing confirmatory factor analysis with EQS 6.1 (Bentler, 2005). The model was tested using the robust maximum likelihood estimation method, which includes the Satorra–Bentler-scaled χ2 index (S–B χ2) and other corrected statistics. The scale items were hypothesized to be explained by a 1factor structure. This measurement model showed good fit indices: χ2 (4, N = 750) = 13.63, p b 0.01; comparative fit index (CFI) = .93; normed fit index (NFI) = .91; root mean square error of approximation (RMSEA) = .05; and standardized root mean square residua (SRMR) = .04 (CFI and NFI values of .90 or higher reflect an acceptable fit; RMSEA and SRMR values less than .06 indicate a good fit; Hu & Bentler, 1999). All item loadings were equal to or higher than .44 (p b .001). The internal consistency in the sample is α = .74. Individual cyberbullyingjustification scores were calculated for each participant by averaging

84

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

their responses across all items. The higher the score is, the more likely participants are to justify cyberbullying behavior.

6.2.3. Impulsivity The impulsive–irresponsible subscale of the Spanish version of the Youth Psychopathic Inventory (van Baardewijk et al., 2010) was used. This subscale consists of 6 items that measure impulsivity on a 4-point scale ranging from 0 (does not apply at all) to 3 (applies very well). A sample item is “I consider myself a pretty impulsive person.” This scale has shown good construct and predictive validity and reliability among Spanish adolescents (Hilterman, 2010). Impulsivity is a stable personality trait (e.g., Niv, Tuvblad, Raine, Wang, & Baker, 2012), so it was measured only at T1. The internal consistency in the sample was α = .71.

6.3. Procedure We informed students that we were conducting a study on various behaviors among adolescents, including the use of new technologies, and invited them to participate. The responses were recorded anonymously to promote honesty, participation was voluntary, and all adolescents gave consent to participate. Parents were notified and given the option of refusing to allow their child to participate in the two waves of the study. No parents refused to allow their child to participate. Participants completed the questionnaires in their classrooms at 2 measurement points 6 months apart. To pair questionnaires at T1 and T2, a code known only to the participant was used. The questionnaires took about 30–40 min to complete. The Ethics Committee of the University of [eliminated for blind review] approved this study (Protocol Number: [eliminated for blind review]).

7. Results 7.1. Descriptive statistics and correlations Table 1 presents the Pearson correlations and descriptive statistics (mean and SD) for the study variables. As shown, the correlations between the variables are statistically significant. Potential gender differences were analyzed with a series of t-tests (see Table 1). The effect sizes are expressed as Cohen's d based on the pooled standard deviation of the two groups. Girls have higher scores than boys on T1 cybervictimization (boys: M = 0.89, SD = 1.50; girls: M = 1.27, SD = 1.67; t (748) = −3.16, p b .01; d = 0.24), and boys have higher scores than girls on T2 cyberbullying justification (boys: M = 1.37, SD = 0.53; girls: M = 1.28, SD = 0.45; t (748) = −3.21, p b .05; d = 0.18). The effect size of the differences was small. Other differences between boys and girls were not statistically significant.

7.2. Multilevel analysis To test our hypothesis and estimate individual- and class-level influences on cyberbullying perpetration, we performed hierarchical linear modeling (HLM) using HLM 7 software (Raudenbush, Bryk, & Congdon, 2010). HLM includes the multilevel structure of the data (students nested in classes and classes nested in schools) in individual- and group-level influences on a variable of interest without losing variability or violating assumptions of independence (Raudenbush & Bryk, 2002). First, we analyzed possible classroom- and school-level effects by calculating the 3-level null model in HLM 7. For the null model, there are no predictors at any level. The intercept of perpetration at Level 1 was modeled for the classroom grouping effect at Level 2 and the school grouping effect at Level 3. The results showed that the variance in the Level 2 intercept was statistically significant, supporting the existence of a level 2 classroom effect, χ2 (33) = 240.64, p b .001. However, the Level 3 intercept was not statistically significant, suggesting the absence of significant school effects, χ2 (11) = 16.98, p = .11. This result indicates that it is necessary to model the classroom level with a multilevel design. Second, we calculated the intraclass correlations coefficient (ICC) to distinguish within- and between-class variance in the dependent variable. The results show that classes explain approximately 8% (ICC = .079) of the variance in students' cyberbullying perpetration. Cyberbullying, the dependent variable, is a count variable representing the number of aggressive behaviors adolescents perpetrated, so we performed the following analyses to allowing for a Poisson distribution in the outcome variable (Huang & Cornell, 2012). In all models, cyberbullying perpetration at T1 was included as a predictor of cyberbullying perpetration at T2. This approach allowed assessing the extent to which T1 predictors (individual justification, impulsivity, age, and gender) explain cyberbullying at T2, adjusting for T1 levels (Little, 2013). T2 justification, and T1 and T2 victimization were also included in the models as control variables. All missing values (2.3% of the total) in the items were missing at random and were estimated using the expectation maximization algorithm. We first examined the influence of T1 cyberbullying justification on T2 cyberbullying perpetration and whether this association differs as a function of age, gender, and impulsivity at T1. We estimated a model including the main effects of age, gender, T1 and T2 justification, T1 and T2 cybervictimization, and T1 impulsivity. The model also has 6 two- and three-way interaction terms of T1 justification with T1 impulsivity, gender, and age (justification × impulsivity; justification × gender; justification × age; justification × impulsivity × gender; justification × impulsivity × gender; justification × gender × age). The model initially estimated showed that only 1 interaction was statistically significant. For the sake of parsimony, nonsignificant interactions were removed from the model, which was reestimated (see the Appendix A for the equations of the final model). Table 2 presents the final model with the main results. As shown, higher scores for T1 perpetration and T1 impulsivity significantly predicted an increase in individual cyberbullying perpetration at T2. Cyberbullying justification and

Table 1 Pearson correlations and descriptive statistics (means and standard deviation) for the variables in this study. 1 1. Age 2. Cyberbullying perpetration T1 3. Cyberbullying perpetration T2 4. Justification T1 5. Justification T2 6. Cyberbullying victimization T1 7. Cyberbullying victimization T2 8. Impulsivity T1

−.03 .02 .02 −.05 .00 −.08 −.08

2

3

4

5

6

7

8

Full sample M (SD)

Boys M (SD)

Girls M (SD)

t

d

.08

.03 .51⁎⁎⁎

.07 .54⁎⁎⁎ .34⁎⁎⁎

−.01 .29⁎⁎⁎ .36⁎⁎⁎ .45⁎⁎⁎

.08 .45⁎⁎⁎ .30⁎⁎⁎ .32⁎⁎⁎ .15⁎⁎

.03 .39⁎⁎⁎ .40⁎⁎⁎ .27⁎⁎⁎ .16⁎⁎ .59⁎⁎⁎

.02 .35⁎⁎⁎ .29⁎⁎⁎ .28⁎⁎⁎ .28⁎⁎⁎ .24⁎⁎⁎ .25⁎⁎⁎

14.77 (0.97) 1.41 (2.26) 1.56 (2.45) 1.22 (0.38) 1.32 (.48) 1.12 (1.62) 1.20 (1.79) 1.07 (0.55)

14.77 (1.01) 1.53 (2.50) 1.52 (2.47) 1.20 (0.37) 1.37 (0.53) 0.89 (1.50) 1.06 (1.78) 1.08 (0.55)

14.77 (0.95) 1.34 (2.09) 1.60 (2.44) 1.24 (0.42) 1.28 (0.45) 1.27 (1.67) 1.26 (1.80) 1.06 (0.55)

−0.17 −1.05 0.41 −1.33 7.70⁎ −3.16⁎⁎

0 .01 .04 .10 .18 .24 .11 .03

.54⁎⁎⁎ .51⁎⁎⁎ .26⁎⁎⁎ .35⁎⁎⁎ .25⁎⁎⁎ .24⁎⁎⁎

.32⁎⁎⁎ .40⁎⁎⁎ .38⁎⁎⁎ .47⁎⁎⁎ .27⁎⁎⁎

.27⁎⁎⁎ .20⁎⁎ .21⁎⁎⁎ .25⁎⁎⁎

.16⁎⁎ .10 .22⁎⁎⁎

.51⁎⁎⁎ .23⁎⁎⁎

.24⁎⁎⁎

Note. Correlations for women are above the diagonal and correlations for men below the diagonal. ⁎ p b .05. ⁎⁎ p b .01. ⁎⁎⁎ p b .001.

−1.44 0.65

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

85

Table 2 Multilevel modeling with class justification as Level 2 predictor of CB perpetration at Time 2 (estimated final model). Fixed effects

Level-1 Gender Agea T2 justification of cyberbullyinga T2 cybervictimizationa T1 cyberbullying perpetration (baseline)a T1 impulsivitya T1 justification of cyberbullyinga T1 cybervictimizationa T1 justification × T1 impulsivitya Level-2 Class justification of cyberbullyingb Class perpetration of cyberbullyinga Class justification of cyberbullyingb × age Random effects Group mean Age slope

Model 1

Model 2

Coefficient

SE

T ratio

Event rate ratio

Coefficient

SE

T ratio

Event rate ratio

−0.07 0.04 0.09 0.12 0.13 0.05 0.02 0.05 −0.12

.05 0.01 0.01 0.03 0.02 0.01 0.02 0.03 0.03

−1.46 0.34 7.68⁎⁎⁎ 4.26⁎⁎⁎ 6.48⁎⁎⁎ 4.34⁎⁎⁎

0.93 1.04 1.09 1.13 1.13 1.05 1.03 1.05 0.88

−0.08 0.08 0.09 0.11 0.18 0.05 0.01 0.03 −0.13

0.04 0.10 0.01 0.03 0.02 0.01 0.02 0.02 0.03

−1.60 0.82 8.22⁎⁎⁎ 4.02⁎⁎⁎ 10.33⁎⁎⁎ 5.21⁎⁎⁎ 0.60 1.50 −4.93⁎⁎⁎

0.93 1.09 1.09 1.12 1.20 1.06 1.01 1.03 0.88

0.45 −0.002 −0.36 Variance 0.32 0.48

0.18 0.12 0.16 Df 40 41

2.54⁎ −0.02 −2.27⁎ χ2 328.58⁎⁎⁎ 122.27⁎⁎⁎

Variance 0.30

Df 44

1.11 1.82† −4.09⁎⁎⁎

χ2 341.85⁎⁎⁎

1.57 0.99 0.69

† p b .10. ⁎ p b .05. ⁎⁎⁎ p b .001. a Centered around its class mean. b Centered around its grand mean.

cybervictimization at T2 was also significantly associated with cyberbullying perpetration at the same time point. T1 justification and T1 cybervictimization were not related to an increase in T2 cyberperpetration. However, the interaction term between T1 impulsivity and T1 justification was significant. This interaction was further analyzed by calculating simple slopes at 1 standard deviation above and 1 standard deviation below the mean, using the computational tools developed by Preacher, Curran, and Bauer (2006). The simple-slope coefficients for low and high impulsivity were: blow = .16, t = 3.72, p b .001; bhigh = −.06, t = − 1.73, ns. Thus, justification was significantly associated with cyberbullying perpetration only at lower levels of impulsivity. Fig. 1 depicts the interaction effect. As seen, cyberbullying perpetration at T2 was low at low levels of impulsivity and low levels of justification of cyberaggression. We next examined possible class-level influences on cyberbullying perpetration at T2. We computed class cyberbullying justification by aggregating students' responses by classroom following the procedure proposed by Krull and MacKinnon (2001); see also Pozzoli, Gini, & Vieno, 2012b. Following the recommendations of Raudenbush and Bryk (2002), we included class justification as a predictor of T2 cyberbullying perpetration and examined cross-level interactions by including class justification as a predictor of random slopes for Level 1 age

and gender. Classroom-level cyberaggression was added to the model as a control variable. The results of this model showed that the interaction between class justification and gender was not significant. For parsimony, this nonsignificant interaction was removed from the model, which was then reestimated. This final model showed a significantly better fit than the model which includes only Level 1 variables, Δχ2(4; N = 750) = 13.27, p b .01. This final model is reported in Table 2. The results show that class justification explains a significant portion of the between-class variance in T2 cyberbullying perpetration. This main effect is qualified by the significant interaction between age and class justification. To follow up the results of the significant age–class justification interaction, a simple-slope analysis was conducted using the coefficients from the model (Preacher et al., 2006). This analysis, depicted in Fig. 2, shows that class justification has a significant association with cyberbullying perpetration among younger adolescents (1 SD below the mean) (b = 0.79, t = 3.66, p b 0.001) but not older adolescents (1 SD above the mean) (b = 0.09, t = 0.44, p = 0.66). As seen in Fig. 2, older adolescents showed moderate levels of cyberbullying perpetration at T2 regardless of their class justification. In contrast, younger adolescents showed low levels of cyberbullying perpetration when class justification for cyberaggression was low and high levels of cyberbullying perpetration when class justification for cyberbullying was high.

Fig. 1. Relationship between T1 justification and T2 cyberbullying as a function of impulsivity.

Fig. 2. Relationship between T1 class justification (Level 2) and T2 cyberbullying as a function of participants' age.

86

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

We further analyzed this developmental effect by determining in what grades the relation between class justification and cyberbullying perpetration is significant. We estimated an additional model including grade as a predictor variable at Level 1 and the interaction between grade and class justification. The results show that the association between class justification and cyberbullying perpetration is most significant and strongest for 3rd graders in secondary education (i.e., U.S. 9th graders) (b = 0.76, t = 4.16, p b .001). It is also significant for 4th graders (i.e., U.S. 10th graders) (b = 0.29, t = 2.16, p b .05) but not for adolescents in the 1st year of high school (i.e., U.S. 11th graders) (b = − 0.16, t = − 0.65, p = .51). Finally, we further explored if the ICC for justification varies by age, which, if stronger at younger ages, may account for the cross-level interaction. The results showed that ICC tends to be higher and significant for secondary students (.06; p b .05) and lower for high-school students (0.001, ns), which suggests that class group tends to be more important for younger adolescents. 8. Discussion Justification of aggressive behavior as acceptable to a degree for various reasons has been reported to be a significant risk factor for behaviors, including traditional bullying and cyberbullying (Kowalski et al., 2014). In this study, we expanded this line of research by testing the role of individual- and class-level justification of cyberbullying in adolescents' cyberbullying perpetration using a longitudinal design. Cyberbullying perpetration at T2 was significantly explained by cyberbullying justification at the same time point. In addition, cyberbullying justification at T1 showed a significant association with cyberbullying perpetration only at lower levels of impulsivity. That is, participants reported higher levels of cyberbullying at T2 when they were high in impulsivity, independently of their level of justifications. However, cyberbullying justifications significantly predicted higher levels of cyberbullying perpetration when impulsivity was low. Being low in both impulsivity and justifications emerged, therefore, as a protective condition for cyberbullying perpetration, whereas both high justification and high impulsivity could be considered risk factors. This result is consistent with a recent study (Gini et al., 2015) finding that impulsivity and the specific mechanisms of moral disengagement interacted to explain reactive aggression among Italian adolescents. Similarly, the findings of the present study indicated that holding broader normative beliefs justifying cyberbullying affects subsequent cyberbullying behavior only among adolescents with lower impulsivity. This result expands previous findings to show that the cognitive justification of cyberbullying is a psychological process that facilitates acting immorally (i.e., being aggressive toward others). Although, separately, impulsivity and justification were risk factors for perpetrating cyberbullying, we can conclude that the effects of justification and impulsivity on cyberbullying are not additive. It remains unclear why a combination of high impulsivity and high justifications did not further increase the risk for cyberbullying perpetration above the levels predicted by high impulsivity and high justifications alone. Future studies should examine this issue. At the class level, our findings support the main hypothesis that class justification was significantly related to cyberbullying perpetration. Aggressive acts in online peer interactions were more likely in class groups whose members tend to justify cyberbullying behavior in certain circumstances for purposes, such as revenge. Interestingly, this effect was qualified by an interaction with age, which revealed that class justification had a significant association with cyberbullying perpetration only among younger adolescents. Specifically, this association is strongest in the 3rd grade of secondary education (the equivalent of 9th grade in the United States), weaker but still significant in the 4th grade of secondary education (10th grade in the United States), and nonsignificant in the 1st year of high school (11th grade in the United States). In addition, results suggested greater consensus among classmates at younger ages as to whether cyberaggression is justifiable, as

showed by the higher ICC for justifications in secondary student groups. A possible explanation of this result is that, similar to traditional bullying (Juvonen, Graham, & Schuster, 2003; Salmivalli, 2010), cyberbullying might be partly motivated by the desire to enhance one's peer status, which has been found to be especially important in early adolescence (LaFontana & Cillessen, 2010; Wegge, Vandebosch, Eggermont, & Pabian, 2014). In addition, that students in compulsory secondary education in Spain share more classrooms and time with the same classmates than high school students could explain why class group is more influential for younger adolescents. This result is consistent with previous findings that group influences, whether positive or negative, are stronger in early adolescence (e.g., in adherence to class norms) (Juvonen & Galván, 2008; Pozzoli et al., 2012b), while resistance to peer influence increases gradually between ages 14 and 18 (Steinberg & Monahan, 2007). Neither individual nor group justification showed a significant interaction with gender in predicting cyberbullying. These findings indicate that individual and class justifications of cyberbullying perpetration might have comparable influence on boys and girls. This is consistent with the results of a recent meta-analysis by Gini, Pozzoli, and Hymel (2014) finding an identical association between morally disengaged justification and different forms of aggressive behavior by groups of two genders. Nevertheless, future studies should continue to investigate this issue because the evidence on potential gender differences in the association of cyberbullying justification and perpetration is scarce, and the results regarding the role of gender in cyberbullying are mixed (Kowalski et al., 2014). Finally, 8% of the total variation in perpetration was due to differences between classrooms. This percentage is somewhat smaller but still comparable to that found in other studies on traditional bullying (e.g., 10% in Kärnä, Salmivalli, Poskiparta, & Voeten, 2008; 14% in Pozzoli et al., 2012b). Although cyberbullying has important distinctive features (e.g., it can occur in any place outside the classroom, has increased perceptions of anonymity, reaches a larger audience, and is primarily indirect rather than face-to-face, which can increase moral disengagement; see Smith, 2012), these findings stress the need to consider the potential peer-group influences on cyberbullying and to regard cyberbullying, like traditional bullying, as a group process (Salmivalli, 2010). 8.1. Limitations and implications This study has several limitations that should be considered when interpreting the findings. First, the results are based on self-reported accounts of adolescents' experiences with cyberbullying, which could have introduced bias. The covert nature of cyberbullying, which can be difficult for adults or even peers to detect, makes the individual adolescent the most important source of information (Gradinger, Strohmeier, & Spiel, 2010). However, future studies should try to include reports from others (e.g., peers and teachers) and use additional assessment techniques (e.g., peer nominations) to validate the current findings. Second, this study measured short-term relationships between variables. A longer longitudinal study with more than two waves could further explore the interplay among variables over time. Including younger participants could also help reveal the origins of these dysfunctional relationships. Finally, this study was conducted in Spain, so researchers should attempt to replicate the findings in other cultural contexts. Despite the potential cultural differences between Spanish- and Englishspeaking samples, the estimated rates of cyberbullying in Spain are similar to those in other European countries (Ortega et al., 2012), and research to date suggests that this problem is the same across countries (Menesini et al., 2012). This study has important strengths, including the short-term longitudinal design and multilevel approach, and advances understanding of the role of cyberbullying justification in the etiology of cyberbullying behavior. The findings indicate the need to adopt an ecological

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

perspective to understand cyberbullying situations, in which individual variables and group-level influences should be taken into account to explain perpetration. Furthermore, our findings have important implications for the design of interventions and prevention programs. At the individual level, it is important to promptly identify and modify dysfunctional beliefs that can maintain or increase perpetration (e.g., “It is OK to send electronic messages to someone to mock or laugh at him or her”). Moreover, our study confirms that impulsivity is an important predictor of cyberbullying (Gámez-Guadix et al., 2014; Kokkinos et al., 2014), so programs should be designed to promote impulse control strategies as responses to possible cyberbullying encounters or provocations. The results emphasize that prevention programs should act not only on the individual level but also on the peer group at the class level. By extension, intervention programs should involve not only those who are cyberbullies or victims but also all group members, including those not involved in bullying. Working with group attitudes and beliefs that justify cyberbullying based on different reasons could make a major contribution to prevention. Prevention strategies should be started early because group justification seems to have a stronger effect during early adolescence. Strategies focused on increasing group pro-victim attitudes and awareness of cyberbullying and developing positive group norms (Pozzoli et al., 2012b) could also make important contributions to preventing cyberbullying.

87

β6 j ¼ γ60 β7 j ¼ γ70 β8 j ¼ γ80 β9 j ¼ γ90

A.2. Model 2 Level-1 model

cyberbullyingperpetration ðγ00 Þ ¼ β0 j þ β1 j  ðgenderÞ þ β2 j  ðageÞ þ β3 j  ðT2 justificationÞ þ β4 j  ðT2 cybervictimizationÞ þ β5 j  ðT1 cyberbullyingperpetrationÞ þ β6 j  ðT1 justificationÞ þ β7 j  ðimpulsivityÞ þ β 8 j  ðcybervictimizationÞ þ β9 j  ðjustification  impulsivityÞ

Acknowledgment Funding for this study was provided by Ministerio de Economía y Competitividad (MEC; Spanish Government) grant PSI2012-31550. MEC had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication. Appendix A. Equations for estimated final models A.1. Model 1 Level-1 model

Level-2 model

β0 j ¼ γ00 þ γ 01  ðmean justificationÞ þ γ 02  ðmean perpetrationÞ u0 j β1 j ¼ γ10 β2 j ¼ γ20 þ γ 21  ðmean justificationÞ þ u2 j β3 j ¼ γ30 β4 j ¼ γ40

cyberbullyingperpetrationðγ 00 Þ ¼ β0 j þ β1 j  ðgenderÞ þ β2 j  ðageÞ þ β3 j  ðT2 justificationÞ þ β4 j  ðT2 cybervictimizationÞ þ β5 j  ðT1 cyberbullyingperpetrationÞ þ β6 j  ðT1 justificationÞ þ β7 j  ðimpulsivityÞ þ β8 j  ðcybervictimizationÞ þ β9 j  ðjustification  impulsivityÞ

Level-2 model β0 j ¼ γ00 þ u0 j β1 j ¼ γ10 β2 j ¼ γ20 β3 j ¼ γ30 β4 j ¼ γ40 β5 j ¼ γ50

β5 j ¼ γ50 β6 j ¼ γ60 β7 j ¼ γ70 β8 j ¼ γ80 β9 j ¼ γ90

References Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27–51. http://dx.doi.org/10.1146/annurev.psych.53.100901.135231. Athanasiades, C., & Deliyanni-Kouimtzis, V. (2010). The experience of bullying among secondary school students. Psychology in the Schools, 47(4), 328–341. Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change. Psychological Review, 84(2), 191. Barlett, C. P., & Gentile, D. A. (2012). Attacking others online: The formation of cyberbullying in late adolescence. Psychology of Popular Media Culture, 1, 123–135. http://dx.doi.org/10.1037/a0028113. Barlett, C. P., Gentile, D. A., & Chew, C. (2014). Predicting cyberbullying from anonymity. Psychology of Popular Media Culture.. http://dx.doi.org/10.1037/ppm0000055. Barlett, C. P., Gentile, D. A., Anderson, C. A., Suzuki, K., Sakamoto, A., Yamaoka, A., & Katsura, R. (2014). Cross-cultural differences in cyberbullying behavior a shortterm longitudinal study. Journal of Cross-Cultural Psychology, 45, 300–313. http://dx. doi.org/10.1177/0022022113504622. Bentler, P. M. (2005). EQS 6.1 for Windows. Structural equations program manual. Encino, CA: Multivariate Software.

88

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89

Bhat, C. S. (2008). Cyber bullying: Overview and strategies for school counsellors, guidance officers, and all school personnel. Australian Journal of Guidance and Counselling, 18, 53–66. http://dx.doi.org/10.1375/ajgc.18.1.53. Bosworth, K., Espelage, D. L., & Simon, T. R. (1999). Factors associated with bullying behavior in middle school students. The Journal of Early Adolescence, 19, 341–362. Boulton, M., Lloyd, J., Down, J., & Marx, H. (2012). Predicting undergraduates' selfreported engagement in traditional and cyberbullying from attitudes. Cyberpsychology, Behavior and Social Networking, 15(3), 141–147. Bronfenbrenner, U. (1979). Contexts of child rearing: Problems and prospects. American Psychologist, 34, 844–850. http://dx.doi.org/10.1037//0003-066X.34.10.844. Brown, B. B., Clasen, D. R., & Eicher, S. A. (1986). Perceptions of peer pressure, peer conformity dispositions, and self-reported behavior among adolescents. Developmental Psychology, 22, 521–530. http://dx.doi.org/10.1037/0012-1649.22.4.521. Calvete, E. (2008). Justification of violence and grandiosity schemas as predictors of antisocial behavior in adolescents. Journal of Abnormal Child Psychology, 36, 1083–1095. Calvete, E., Orue, I., Estévez, A., Villardón, L., & Padilla, P. (2010). Cyberbullying in adolescents: Modalities and aggressors' profile. Computers in Human Behavior, 26, 1128–1135. http://dx.doi.org/10.1016/j.chb.2010.03.017. Caravita, S. C. S., Sijtsema, J. J., Rambaran, A., & Gini, G. (2013). Moral disengagement development: Peer influences on cognitive self-justification mechanisms of social transgression. Journal of Youth and Adolescence, 43, 193–207. http://dx.doi.org/10.1007/ s10964-013-9953-1. Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social informationprocessing mechanisms in children's social adjustment. Psychological Bulletin, 115, 74. Domingo-Salvany, A., Regidor, E., Alonso, J., & Alvarez-Dardet, C. (2000). Proposal for a social class measure. Working group of the Spanish Society of Epidemiology and the Spanish Society of Family and Community Medicine. Atencion Primaria, 25, 350–363. Hilterman (2010). El Youth Psychopathy Traits Inventory (YPI): validez convergente con el PCL:YV y validez predictiva con la reincidencia autoreportada por jóvenes en Cataluña. (Presented at VII Congreso Español de Criminología). Elledge, L. C., Williford, A., Boulton, A. J., DePaolis, K. J., Little, T. D., & Salmivalli, C. (2013). Individual and contextual predictors of cyberbullying: The influence of children's provictim attitudes and teachers' ability to intervene. Journal of Youth and Adolescence, 42, 698–710. Espelage, D. L., & Swearer, S. M. (2003). Research on school bullying and victimization: What have we learned and where do we go from here? School Psychology Review, 32, 365–383. Faris, R., & Ennett, S. (2012). Adolescent aggression: The role of peer group status motives, peer aggression, and group characteristics. Social Networks, 34, 371–378. http://dx. doi.org/10.1016/j.socnet.2010.06.003. Gámez-Guadix, M., Gini, G., & Calvete, E. (2015). Stability of cyberbullying victimization among adolescents: Prevalence and association with bully–victim status and psychosocial adjustment. Computers in Human Behavior, 53, 140–148. http://dx.doi.org/10. 1016/j.chb. 2015.07.007. Gámez-Guadix, M., Orue, I., Smith, P. K., & Calvete, E. (2013). Longitudinal and reciprocal relations of cyberbullying with depression, substance use, and problematic internet use among adolescents. Journal of Adolescent Health, 53, 446–452. http://dx.doi.org/ 10.1016/j.jadohealth.2013.03.030. Gámez-Guadix, M., Villa, F., & Calvete, E. (2014). Psychometric properties of the Cyberbullying Questionnaire (CBQ) among Mexican adolescents. Violence & Victims, 29, 232–247. http://dx.doi.org/10.1891/0886-6708.VV-D-12-00163R1. Gini, G., & Espelage, D. L. (2014). Peer victimization, cyberbullying, and suicide risk in children and adolescents. JAMA: The Journal of the American Medical Association, 312, 545–546. Gini, G., Pozzoli, T., & Bussey, K. (2014). Collective moral disengagement: Initial validation of a scale for adolescents. The European Journal of Developmental Psychology, 11, 386–395. http://dx.doi.org/10.1080/17405629.2013.851024. Gini, G., Pozzoli, T., & Bussey, K. (2015). Moral disengagement moderates the link between psychopathic traits and aggressive behavior among early adolescents. Merrill-Palmer Quarterly, 61, 51–67. Gini, G., Pozzoli, T., & Hymel, S. (2014). Moral disengagement among children and youth: A meta-analytic review of links to aggressive behavior. Aggressive Behavior, 40, 56–68. http://dx.doi.org/10.1002/ab.21502. Gradinger, P., Strohmeier, D., & Spiel, Ch. (2010). Definition and measurement of cyberbullying. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 4(2) (Available at http://cyberpsychology.eu/view.php?cisloclanku=2010112301&article=). Heirman, W., & Walrave, M. (2012). Predicting adolescent perpetration in cyberbullying: An application of the theory of planned behavior. Psicothema, 24, 614–620. Hinduja, S., & Patchin, J. W. (2008). Cyberbullying: An exploratory analysis of factors related to offending and victimization. Deviant Behavior, 29, 129–156. http://dx.doi. org/10.1080/01639620701457816. Hu, L., & Bentler, P. M. (1999). Cut off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. http://dx.doi.org/10.1080/10705519909540118. Huang, F. L., & Cornell, D. G. (2012). Pick your Poisson: A tutorial on analyzing counts of student victimization data. Journal of School Violence, 11, 187–206. Huesmann, L. R. (1988). An information processing model for the development of aggression. Aggressive Behavior, 14, 13–24. Huesmann, L., & Eron, L. (1984). Cognitive processes and the persistence of aggressive behavior. Aggressive Behavior, 10, 243–251. Huesmann, L. R., & Guerra, N. G. (1997). Children's normative beliefs about aggression and aggressive behavior. Journal of Personality and Social Psychology, 72, 408–419. http://dx.doi.org/10.1037//0022-3514.72.2.408. Juvonen, J., & Galván, A. (2008). Peer influence in involuntary social groups: Lessons from research on bullying. In M. Prinstein, & K. Dodge (Eds.), Peer influence processes among youth (pp. 225–244). New York: Guilford Press.

Juvonen, J., Graham, S., & Schuster, M. A. (2003). Bullying among young adolescents: The strong, the weak, and the troubled. Pediatrics, 112, 1231–1237. http://dx.doi.org/10. 1542/peds.112.6.1231. Kärnä, A., Salmivalli, C., Poskiparta, E., & Voeten, M. J. M. (2008, May). Do bystanders influence the frequency of bullying in a classroom. In the XIth EARA conference, Turin, Italy. Kokkinos, C. M., Antoniadou, N., & Markos, A. (2014). Cyber-bullying: An investigation of the psychological profile of university student participants. Journal of Applied Developmental Psychology, 35, 204–214. http://dx.doi.org/10.1016/j.appdev.2014.04. 001. Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M. R. (2014). Bullying in the digital age: A critical review and meta-analysis of cyberbullying research among youth. Psychological Bulletin, 140, 1073–1137. http://dx.doi.org/10.1037/ a0035618. Krull, J. L., & MacKinnon, D. P. (2001). Multilevel modeling of individual and group level mediated effects. Multivariate Behavioral Research, 36, 249–277. http://dx.doi.org/10. 1207/S15327906MBR3602_06. LaFontana, K. M., & Cillessen, A. H. (2010). Developmental changes in the priority of perceived status in childhood and adolescence. Social Development, 19, 130–147. http:// dx.doi.org/10.1111/j.1467-9507.2008.00522.x. Little, T. (2013). Longitudinal structural equation modelling. New York: Guilford. Menesini, E., Nocentini, A., Palladino, B. E., Frisén, A., Berne, S., Ortega-Ruiz, R., ... Naruskov, K. (2012). Cyberbullying definition among adolescents: A comparison across six European countries. Cyberpsychology, Behavior and Social Networking, 15, 455–463. Niv, S., Tuvblad, C., Raine, A., Wang, P., & Baker, L. A. (2012). Heritability and longitudinal stability of impulsivity in adolescence. Behavior Genetics, 42, 378–392. http://dx.doi. org/10.1007/s10519-011-9518-6. O'Brennan, L. M., Bradshaw, C. P., & Sawyer, A. L. (2009). Examining developmental differences in the social–emotional problems among frequent bullies, victims, and bully/ victims. Psychology in the Schools, 46, 100–115. http://dx.doi.org/10.1002/pits.20357. Ortega, R., Elipe, P., Mora-Merchán, J. A., Genta, M. L., Brighi, A., Guarini, A., ... Tippett, N. (2012). The emotional impact of bullying and cyberbullying on victims: A European cross-national study. Aggressive Behavior, 38, 342–356. Podsakoff, P. M., MacKenzie, S. B., Lee, J. -Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879. http://dx.doi.org/10.1037/00219010.88.5.879. Pozzoli, T., Gini, G., & Vieno, A. (2012a). The role of individual correlates and class norms in defending and passive bystanding behavior in bullying: A multilevel analysis. Child Development, 83, 1917–1931. http://dx.doi.org/10.1002/ab.21442. Pozzoli, T., Gini, G., & Vieno, A. (2012b). Individual and class moral disengagement in bullying among elementary school children. Aggressive Behavior, 38, 378–388. http://dx. doi.org/10.1002/ab.21442. Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006). Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437–448. http://dx.doi.org/10. 3102/10769986031004437. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Aplications and data analysisi methods. Newbury Park, CA: Sage. Raudenbush, S. W., Bryk, A. S., & Congdon, R. (2010). HLM 7.00 for Windows. Scientific Software International Inc, Lincolnwood. Rubin, K. H., Bukowski, W. M., & Parker, J. G. (1998). Peer interactions, relationships, and groups. Handbook of Child Psychology. III. 10, . http://dx.doi.org/10.1002/ 9780470147658.chpsy0310. Salmivalli, C. (2010). Bullying and the peer group: A review. Aggression and Violent Behavior, 15, 112–120. http://dx.doi.org/10.1016/j.avb.2009.08.007. Salmivalli, C., & Voeten, M. (2004). Connections between attitudes, group norms, and behaviour in bullying situations. International Journal of Behavioral Development, 28, 246–258. http://dx.doi.org/10.1080/01650250344000488. Sieving, R. E., Perry, C. L., & Williams, C. L. (2000). Change friendships? Examining paths of influence in young adolescents' alcohol use. Journal of Adolescent Health, 26, 27–35. Smith, P. K. (2012). Cyberbullying and cyber aggression. In A. B. N. S. R. Jimerson, M. J. Mayer, & M. J. Furlong (Eds.), Handbook of school violence and school safety: International research and practice (pp. 93–103) (2nd ed.). New York: Routledge. Smith, P. K., Mahdavi, J., Carvalho, M., Fisher, S., Russell, S., & Tippett, N. (2008). Cyberbullying: Its nature and impact in secondary school pupils. Journal of Child Psychology and Psychiatry, 49, 376–385. http://dx.doi.org/10.1111/j.1469-7610.2007. 01846.x. Steinberg, L., & Monahan, K. C. (2007). Age differences in resistance to peer influence. Developmental Psychology, 43, 1531. http://dx.doi.org/10.1037/0012-1649.43.6.1531. Subrahmanyam, K., & Greenfield, P. (2008). Communicating online: Adolescent relationships and the media. The Future of Children: Children and Media Technology, 18, 119–146. Subrahmanyam, K., Reich, S., Waechter, N., & Espinoza, G. (2008). Online and offline social networks: Use of social networking by emerging adults. Journal of Applied Developmental Psychology, 29, 420–433. http://dx.doi.org/10.1016/j.appdev.2008.07. 003. Swearer, S. M., Wang, C., Berry, B., & Myers, Z. R. (2014). Reducing bullying: application of social cognitive theory. Theory Into Practice, 53, 271–277. http://dx.doi.org/10.1080/ 00405841.2014.947221. Tokunaga, R. S. (2010). Following you home from school: A critical review and synthesis of research on cyberbullying victimization. Computers in Human Behavior, 26, 277–287. http://dx.doi.org/10.1016/j.chb.2009.11.014. van Baardewijk, Y., Andershed, H., Stegge, H., Nilsson, K. W., Scholte, E., & Vermeiren, R. (2010). Development and tests of short versions of the youth psychopathic traits inventory and the youth psychopathic traits inventory-child version. European Journal of Psychological Assessment, 26, 122–128. http://dx.doi.org/10.1027/1015-5759/a000017.

M. Gámez-Guadix, G. Gini / Journal of Applied Developmental Psychology 44 (2016) 81–89 van Geel, M., Vedder, P., & Tanilon, J. (2014). Relationship between peer victimization, cyberbullying, and suicide in children and adolescents: A meta-analysis. JAMA Pediatrics, 168, 435–442. http://dx.doi.org/10.1001/jamapediatrics.2013.4143. Vazsonyi, A. T., Machackova, H., Sevcikova, A., Smahel, D., & Cerna, A. (2012). Cyberbullying in context: Direct and indirect effects by low self-control across 25 European countries. European Journal of Developmental Psychology, 9, 210–227. http://dx.doi.org/10.1080/17405629.2011.644919. Vieno, A., Gini, G., & Santinello, M. (2011). Different forms of bullying and their association to smoking and drinking behavior in Italian adolescents. Journal of School Health, 81, 393–399. http://dx.doi.org/10.1080/17405629.2011.644919.

89

Wegge, D., Vandebosch, H., Eggermont, S., & Pabian, S. (2014). Popularity through online harm: The longitudinal associations between cyberbullying and sociometric status in early adolescence. Journal of Early Adolescence. http://dx.doi.org/10.1177/ 0272431614556351 (published online before print October 27, 2014). Williams, K. R., & Guerra, N. G. (2007). Prevalence and predictors of internet bullying. Journal of Adolescent Health, 41, S14–S21. http://dx.doi.org/10.1016/j.jadohealth. 2007.08.018.