Overall the present study gives important insights into how adolescents define cyberbullying. This is the first study conducted in different countries, systematically manipulating the three “traditional bullying criteria” (intentionality, repetition, and imbalance of power) and the two new “specific cyberbullying criteria” (public vs. private and anonymity) in order to test their relevance for adolescent's definition, taking into account different types of cyberbullying behaviors.
Using the scenarios developed, we were able to discriminate the relevance of different criteria. The MDS analyses across country and types of behavior suggested a clear first dimension characterized by imbalance of power and a clear second dimension characterized by intentionality and, at a lower level, by anonymity. This shows that when adolescents evaluate a scenario as cyberbullying they mainly consider the presence of the traditional bullying criteria with an exception: the criterion of repetition. This is arguably not so important in the virtual context, because the nature of Information and Communication Technology can lead to an immense number of victimizations without the contribution of the perpetrator.2,4,15
The strongest criterion needed to define cyberbullying is imbalance of power, in this study defined as consequences on the victim who was upset and did not know how to defend him/herself. The relevance of this criterion is confirmed across all the countries and across all the types of behavior. It is also more relevant than intentionality, and we might ask why. Research on bullying has highlighted the dynamic between the bully's power and the weakness (social, psychological, or physical) of the victim who cannot easily defend him/herself.8
Our definition of imbalance of power focused on the consequences for the victims. In face-to-face contexts, bullies have been described as more popular, smarter, and stronger.8
But the imbalance of power is not simply based on the social status of bullies, it is also based on the microprocess of action and reaction. If the bully attacks and the victim is upset and does not know how to defend him or herself, then this creates the imbalance within the dyad and, by definition, a bullying attack. Our definition of power imbalance did not specify why the victim cannot defend him/herself or why he/she is weak as compared with the perpetrator, but it gives a clear information about the reaction of the victim and about his/her status in the relationship. This definition introduces a more interactional description of imbalance of power criterion which needs further investigation.
The second dimension that emerged from the MDS is intentionality. This is part of the definition of general aggressive behaviors.8
Almost all the definitions of bullying and cyberbullying include this attribute,15
and several studies have confirmed that the perpetrator must have the intention to harm in order for it to be defined as cyberbullying, otherwise the behavior is perceived as a joke.2, 12–14
The present findings clearly support this view.
Finally, another criterion seems to define the second dimension together with intentionality: the anonymity. When the imbalance of power is not present, we have a higher probability to perceive it as cyberbullying if the attack is intentional and nonanonymous and a lower probability if the attack is nonintentional and anonymous. The role of anonymity as a specific cyberbullying criterion has been stressed by several authors.14–16,21–23
Such studies have underlined the threatening nature of anonymity, but Nocentini et al.2
showed that although anonymity can raise insecurity and fear—if the perpetrator is familiar and he/she is someone who can be trusted—this can hurt the victim more. Overall, it seems that when anonymity is considered without any other criteria, it is perceived as more severe than when the act is done by a known person; at the same time when the act is anonymous and nonintentional, it is less representative of cyberbullying. Our results suggest that anonymity might change its impact on perception in relation to the other criteria and needs to be considered together with other criteria to be fully understood. Public versus private criterion did not show any relevance for the definition of cyberbullying; it seems that an act is defined as cyberbullying regardless of the fact that it is spread to a large audience or not. However, we cannot exclude that this criterion add something about the cyberbullying definition but at a lower level of relevance (i.e., as third dimension) or in combination with other criteria.
In terms of differences across types of behaviors, descriptive frequencies showed a more ambiguous role for exclusion as a form of cyberbullying, in line with traditional bullying literature.5,32
The other three types of behavior showed the same trend in terms of percentages of distribution. General support for the relevance of imbalance of power across all the types of behavior was found; intentionality (together with anonymity) is slightly more important in defining exclusion and impersonation than that of written-verbal and visual behaviors. However, given that differences are small we can assume structural equivalence of both dimensions across all the four types of behavior.†
In relation to the cross-country comparison, we need to underline some specificities related to the high frequencies of French participant's perceptions of the scenarios as cyberbullying, or cyberviolence; the French language does not have a direct translation of the term bullying and the term violence is generally used.5
Two possible explanations can help to understand these high frequencies of response. First, during 2010, a massive media campaign about school violence and cyberviolence was disseminated at the school level in France. Second, the term “cyberviolence” is very broad and can include a wider range of behaviors than the other terms used.5
In terms of the definition, general support for the relevance of imbalance of power was found across all the countries; intentionality (together with anonymity) is slightly less important in Italy and Germany as compared with the other countries. However, the differences are small, and are consistent with a structural equivalence of the second dimension across countries.
Finally, some limitations of the study have to be discussed, such as the randomized administration of the scenarios, the different terms used in each country, and the necessity of taking into account gender- and age-related differences.