Bullying in school and cyberspace: Associations with depressive symptoms in Swiss and Australian adolescents

As children develop, the peer context acquires increasing importance for health and well-being [7]. Peer problems during childhood and adolescence can often result in disruptions to healthy functioning both for those who engage in disruptive behaviours as well as those who are victimised.

It is well established that being a victim of bullying has negative short- and long-term consequences. Furthermore, it is reported that negative peer relations such as lack of acceptance in the peer group and peer victimisation are associated with loneliness, social dissatisfaction and social withdrawal [8] and emotional and behavioural symptoms [9]. Importantly, evidence from several longitudinal studies has demonstrated that peer victimisation and exclusion may also increase children's depressive symptoms [10‐13]. These findings indicate that peer rejection and victimisation may play a causal role in the development of depressive symptoms. Consistently, the causal influence of peer victimisation on symptoms of depression was supported by the results of a recent twin study [14].

A meta-analytic review of cross-sectional associations between peer victimisation and psychosocial maladjustment provided clear evidence that peer victimisation is most strongly related to symptoms of depression and least strongly to anxiety [15]. Peer victimisation is also associated with low self-esteem, health problems, suicidality, and poor school adjustment [16‐20].

Young people who bully others also often experience negative consequences related to their behaviour, some of which are not immediately apparent [21]. For example, primary and middle school students who bully others often seem unscathed, as their social standing and self-concept are similar to that of observers and markedly better than those who are bullied. Early on, these young people are seen as positive leaders with a good sense of humour, high self-esteem qualities and positive early friendship qualities and popularity [22, 23].

Nevertheless, as children grow older bullying behaviours become increasingly maladaptive. Whereas young children solve disputes by fighting, adolescents and adults prefer negotiation to solve a conflict [24]. Children who bully others often do not learn to interact and communicate in socially appropriate ways and therefore have difficulty in interacting adequately with their older peers. This often results in persistent maladaptive behavioural patterns [25], as well as representing an elevated risk for serious injury [26], alcohol dependency [27], and delinquency [28]. These findings suggest that children and adolescents who bully others, frequently also show other forms of antisocial behaviour and that some of those students show a pattern of life-course persistent antisocial behaviour [29].

Furthermore, adolescents who bully others are found to have more psychological and physical problems than their peers [30], and have an increased risk for depression and suicidal ideation [31]. Bullying research traditionally differentiates between children or adolescents who are only victims, only bullies or both [28]. Regarding potential outcomes of bullying, it has been shown that those who both bully others and are victimised (i.e. bully-victims) report the highest levels of externalizing and internalizing symptoms [31, 32].

In sum, bullying perpetration and victimisation may have highly negative consequences for children's and adolescents' mental health and well-being. In general, bullying others is most strongly associated with externalizing problems, while being a victim of bullying is strongly associated with internalizing symptoms.

The existing (albeit limited) literature on cyber-bullying suggests that the consequences of cyber-bullying may be similar to traditional bullying. Cyber-bullying, like traditional bullying, correlates significantly with physical and psychological problems [33]. A large scale Australian-based bullying study also demonstrated that cyber-victimisation is associated with higher levels of stress symptoms [4]. Moreover, adolescent victims of cyber-bullying not only reported higher depressive symptoms but also that they engage in other types of problematic behaviour, such as increased alcohol consumption, a tendency to smoke and poor school grades [34]. Cross-sectional studies showed that aggressors are at increased risk for school problems, assaultive behaviours, and substance use [35]. These findings suggest that cyber-victimisation, like traditional victimisation, increases the risk of internalizing (and externalizing) problems.

However, as traditional and cyber-bullying forms are strongly associated and frequently co-occur within the same individuals [1, 36‐39] it is important to investigate both forms of bullying simultaneously. Few studies have systematically analysed the impact of cyber versus traditional bullying on adolescents' adjustment and mental health.

In a recent study with 761 adolescents from Austria the combined victim group (cyber and traditional victimisation) showed the highest level of internalizing problems [6]. In this study, combined bully-victims showed the most maladjusted pattern. Similarly, a Swedish study found that cyber-victimisation contributed over and above traditional victimisation to adolescents' social anxiety [40]. Cyber-victimisation is also associated with a range of negative emotions [41]. Qualitative data suggest that in comparison with traditional bullying forms, cyber-bullying evoked stronger negative feelings, fear and a clear sense of helplessness [42]. Therefore, being a victim of cyber-bullying might be even more strongly associated with depressive symptoms than traditional victimisation.

This paper describes the relationship between traditional and cyber forms of bullying/victimisation and psychological outcomes. Several hypotheses were generated: (1) there is an overlap between traditional bullying/victimisation and cyber-bullying/victimisation; (2) traditional victims and bully-victims experience higher levels of depressive symptoms than those who bully others and non-involved students; and (3) cyber-victimisation represents an independent risk factor - over and above traditional victimisation - for higher levels of symptoms of depression.

In addition to the three main hypotheses, we examined the influence of culture on the relationship between perpetration/victimisation and outcome. Eslea and colleagues showed in a large dataset from seven different countries that victims of traditional bullying were significantly more disadvantaged on all measures (e.g., mental health, friendships) in all samples, whereas bullies did not differ consistently in all samples. The authors concluded that traditional bullying is a universal phenomenon with many negative correlates for victims and few (if any) for bullies [43]. The consequences associated with cyber-victimisation are not as well established as associations with traditional bullying/victimisation. Moreover, no cross-national comparison has been conducted regarding cyber-bullying so far. Given this, we investigated if the outcomes associated with traditional and cyber forms of bullying were similar for young people in Switzerland and Australia, i.e. we tested whether the results were replicated in both countries (Switzerland versus Australia).

To increase comparability between the two countries' data and due to different requirements for obtaining consent and subsequent low consent rates in government schools, only results from secondary non-government co-educational schools are reported below.

Switzerland. Nineteen school classes (Grades 7 to 9 in the city of St. Gallen) participated in the study [44]. Schools and participating classrooms were selected to represent all city districts (Schulkreise) and to represent all three school types at the secondary level in Switzerland: Realschule with basic classes (low achievement level school, N = 7 classes), Sekundarschule with broader classes (average achievement level school, N = 6 classes) and Kantonschule with advanced classes (high achievement level school, N = 6 classes).

Following Swiss legislation, permission from the respective school councils to conduct the study was first obtained. Second, teachers from the selected schools volunteered. The survey procedure and the goal of the study were explained to the students who then had the opportunity to refrain from participation without negative consequences (informed oral consent). Students who did not want to participate were offered another activity during the respective school hour. Participating school classes received a voucher for books and media worth 50 Swiss Franks. Teachers and students received general feedback about the occurrence of bully/victim behaviours in their classes and an information flyer that provided contact information for students who may require help following completion of the survey.

In the following we differentiate between bullying (= perpetration) and victimisation (being a victim of bullying).

Both samples. Due to the nature of cyber-bullying, repetition as a defining feature of this bullying behaviour may be hard to assess [5]. Therefore, no established cut-offs for being a cyber-bully or cyber-victim exist. In addition, dichotomising these scores would have led to an unnecessary loss of information with regard to various degrees of perpetration/victimisation. Thus, cyber-victimisation and cyber-bullying were analysed as linear variables. Whilst the response categories varied between the studies, this was mostly at the upper end of the scale where there were relatively few responses.

Australia. Students completed a 14-item depression subscale of the Depression Anxiety Stress Scales (DASS) [45].

Switzerland. Students completed an 8-item scale addressing depressive symptoms. The scale has been validated in a longitudinal study [46, 47].

Data analyses accounted for the skew of the dependent variables through the use of tobit regressions, the data were log transformed to meet the requirement of normality of the non-censored scores as recommended by Osgood [48]. Our analyses also accounted for non-independence of the data resulting from the clustered sampling, which can lead to inflated Type I error rates, through the inclusion of a random intercept in the models. Clustering in the Australian data was by school (where secondary students within a year level move between classes for different subjects) and by class in the Swiss sample.

For the statistical analyses, a significance level of p < 0.05 was used.

Table 1 shows means and standard deviations of all study variables by sample and gender.

Both types of bullying and victimisation were significantly associated with each other (see Table 2 and Table 3). These relationships remained statistically significant (all p < .01) when examined by country, with stronger associations observed in the Australian sample. When comparing the traditional bully-victim categories, 41% of (traditional) bullies, 59% of bully-victims, 30% of victims and 16% of non-involved students reported perpetrating cyber-bullying behaviours at least once or twice. Thirty-nine percent of (traditional) victims, 50% of bully-victims, 22% of bullies and 17% of non-involved students were exposed to cyber-bullying behaviours at least once or twice. The association between bullying behaviour and mental health revealed some interesting results with depressive symptoms being most strongly correlated with traditional victimisation (Spearman's rho = .26 Australian sample, rho = .24 Swiss sample) and cyber-victimisation (rho = .22 Australian sample, rho = .12 Swiss sample).

Next, two tobit regression analyses were conducted to analyse differences between those who use traditional methods to bully, those who are victimised, the combined group (hereafter bully-victims for brevity) and non-involved students in terms of their tendency to cyber-bully others and be cyber-victimised (as log-transformed linear dependent variables). Age and gender and country were entered as control variables. As we were interested in whether country moderates the associations, location (i.e., Switzerland or Australia) was entered as an interaction effect in a first model.

To analyse differences between traditional bullies, victims, and bully-victims in relation to depressive symptoms, the same modelling procedure as described above was used. In the first analysis, only traditional bully/victim categorization was used (including a test of the interaction with country) with age and gender entered as control variables. In the second analysis, cyber-bullying and cyber-victimisation (as well as their interactions with country) were entered as additional independent variables.

The first hypothesis, which proposed a relationship between traditional and cyber forms of bullying and victimisation, was supported with statistically significant relationships between traditional and cyber forms of bullying perpetration and victimisation in the expected direction. Importantly, significant correlations were found between cyber-victimisation and gender (female), age, traditional bullying perpetration and victimisation. Furthermore, as participants aged, their self-reported bullying perpetration (traditional and cyber) increased, a relationship that remained significant only in the Australian sample when country-specific report was examined. Overall, all associations were stronger in the Australian sample.

These results add to the theoretical [5] and other empirical evidence [1, 4, 36‐39] demonstrating the relationship between traditional and cyber forms of bullying perpetration and victimisation. In accordance with other studies, our findings suggest that traditional and cyber-bullying form part of the same cluster of socially inappropriate behaviours and argue for a behavioural versus technical approach to intervention programs.

It was also hypothesized that those victimised using traditional methods (victims and bully-victims) would endorse more symptoms of depression than those who only reported bullying perpetration. Support for this hypothesis was found demonstrating that students who reported being victimised and bullying others as well as those only victimised were more likely to report depressive symptoms than were those who reported bullying perpetration only. This result was not moderated by country, indicating that the associations were comparable in both countries.

Finally, it was hypothesized that cyber-victimisation would represent an additional risk factor - independent of traditional victimisation - for the development of symptoms of depression. Strong support was found for the independent association that cyber-victimisation has with symptoms of depression over and above traditional bullying victimisation i.e. cyber-victimisation accounts for a significant amount of the variation in depressive symptoms even after controlling for possible effects of traditional victimisation. Importantly, this association was not moderated by country, which suggests that the relationship is not culturally dependent.

However, several differences between countries were found. For example, while Swiss students were more likely to report bullying others, the Australian students who bully others were more likely to report also using cyber-strategies. Despite these differences, it was demonstrated that cyber-victimisation was a significant predictor of depressive symptoms - a result that was culturally independent. This result suggests an additional negative mental health status associated with being exposed to bullying via technology, over and above that of being victimised by traditional means. Although fewer students reported being cyber-bullied via technology than traditional methods in both countries, clearly the inclusion of technology represents a risk factor for significantly higher rates of internalizing disorders for those victimised using both cyber and traditional methods.

The implications of these findings are important (e.g., for intervention programs) and demonstrate the scope of negative impact associated with cyber-victimisation. It is suggested that certain features of cyber-bullying (e.g., anonymity of perpetrator, accessibility of victim) present additional and difficult challenges for young people who are victimised [49]. It is often assumed that these challenges could contribute to a worsened mental health state for those victimised and the results of this study provide evidence in support of this.

Furthermore, some of the cyber-bullying strategies employed (e.g., nasty comments on SNS profiles) [4] mean that the audience potentially aware of the harassment is significantly larger. For example, if mean and nasty comments are posted on a SNS profile (social networking sites) or if an embarrassing picture is posted and the victim is identified in the picture by name (i.e., being tagged), all people in their network, in addition to other networks, can potentially see that humiliating content. Therefore, strategies against cyber-bullying should also include educating students about privacy settings and safe internet/mobile practices. Given the difficulty in removing comments or pictures from the Internet and the permanence of information shared online, it is not surprising that cyber-victimisation represent an additional and independent risk factor for the development of depressive symptomatology. Further investigation is needed to clarify if specific elements of cyber-victimisation that are associated with poorer mental health outcomes for young people. For example, what is the impact of bullying via social networking sites given comments, pictures, and video can be viewed by a larger network (i.e., more students). Nonetheless, the results of this study raise important questions, as well as concerns, for those young people experiencing mental health issues in addition to bullying via traditional and cyber methods.

There were a number of strengths to this study. This was the first study to describe cultural similarities in relation to the impact of cyber-victimisation on depressive symptom endorsement. Despite some cultural differences (e.g., more Australian students reported using multiple strategies to bully (traditional and cyber) compared to Swiss students), the evidence demonstrating the additive effect of cyber-victimisation on mental health is an important result. Furthermore, the (culturally independent) predictive nature of cyber-victimisation on depressive symptoms provides an important insight into the influence of technology on young people.

Overall, there were some limitations with this study. For example, some items that assessed bullying and victimisation were worded differently between the two data collection countries. Moreover, there were certain differences in the wording of response categories and number of items in both samples. Regarding cyber-bullying/victimisation, we found a significant difference between Swiss and Australian students regarding their use of cyberstrategies to bully others (Australians reporting higher levels of cyber-bullying/victimisation). This finding has important methodological implications. Swiss students reported on two rather global items on cyber-bullying, whereas Australian students reported on five different behavioural descriptors of cyber-bullying. This might have lead to an underreporting of cyber-bullying in Swiss students. Studies in traditional bullying research have shown that global items result in lower prevalence rates of bullying than specific behavioural items [50].

Regarding depressive symptoms, it is important to know that although Australian students reported on more items than Swiss students, the same number of symptoms were assessed (i.e. the Australian students reported on two items for each symptom, Swiss students on 1-2 items). Nevertheless, we found a significant country effect on depressive symptoms. We assume that these country differences are mainly due to methodological differences. It is unlikely that the differences are culturally-based given the similarities between Switzerland and Australia in relation to the prevalence of depressive symptomatology [51, 52].

There were some sample limitations (Swiss sample comprised students whose teachers volunteered while the Australian sample is comprised of students at religious-affiliated schools only), however, we do not anticipate that the associations examined would differ markedly from those in the general student population. Although there were some differences in sample demographics (e.g. age), these did not have an impact on the relationship between cyber-victimisation and self-reported depressive symptoms. Moreover, samples were highly similar regarding their access to technology. Other limitations concern the nature of the data collected. First, all measures were self-reports. Second, as with all cross-sectional studies the causal direction of the relationships cannot be determined, and thus our focus has been on associations between the variables involved.