School bullying is a major problem in several countries (e.g., Borntrager et al. 2009; Eslea et al. 2003). Olweus (1993) defined bullying as, “a student is being bullied or victimized when he or she is exposed, repeatedly over time, to negative actions on the part of one or more other students” (p. 9). Negative actions refer to aggression, i.e., “intentional inflicts, or attempts to inflict, injury or discomfort upon another” (p. 9). In bullying, there is also an imbalance of power. The victim has difficulty in defending him/herself and is more or less helpless against the bully or bullies (Olweus 1993). Even if there is an enormous body of research on school bullying today, there are still only a few studies, in which school students’ perspectives on why bullying takes place in school has been investigated.

Bullying is a social psychological phenomenon (e.g., Salmivalli 2010), and according to a broad range of social psychological and social developmental theories, such as the social information processing (SIP) models (Crick and Dodge 1994; Dodge et al. 2006; Fiske and Taylor 2008), social representation theory (Deaux and Philogène 2001; Moscovici 2001), and symbolic interactionism (Blumer 1969; Charon 2007), the way in which children and adolescents interpret and make sense of social situations and individuals (including themselves) in these situations affects and guides their behavior. For example, in Crick and Dodge’s (1994) SIP model, children and adolescents’ cognitive processes of social information for solving social problems include a variety of constructs from the on-line processing of current social stimuli to latent knowledge structures or social schemas in their memory (Crick and Dodge 1994; Dodge et al. 2006). Their assumptions and conceptions of social situations and people involved in these situations influence how they process social information and thus how they respond and act in these social situations (Crick and Dodge 1994; Gifford-Smith and Rabiner 2004), including bullying situations (Terranova 2009). For instance, in addition to a child’s general social cognition (e.g., “aggression is effective” or “others are out to get me”), specific cognitions about a specific peer influence the frequency and type of aggression the child would use in relation to this peer (Gifford-Smith and Rabiner 2004). Furthermore, research indicates that the perception of provocation is far more important than the provocation itself in instigating aggression (Dodge et al. 2006). Children and adolescents’ social cognitions, at least in part and as a complement to situational or contextual factors such as social influence and reinforcements, can explain why they act as bullies or take on different bystander roles in bullying situations. Therefore, the task of investigating how students interpret, define and explain bullying situations is an urgent research problem.

Only a few studies have more specifically tried to explore students’ perceptions of the causes of bullying. Previous research indicates that students tend to attribute the causes of bullying to the victim by interpreting him or her as deviant or different (Bosacki et al. 2006; Buchanan and Winzer 2001; Erling and Hwang 2004; Frisén et al. 2007, 2008; Hamarus and Kaikkonen 2008; Hazler and Hoover 1993; Hoover et al. 1992; Teräsahjo and Salmivalli 2003; Thornberg 2010b; Varjas et al. 2008). Furthermore, boys seem to blame the victim more than girls in hypothetical bullying situations (Gini 2008), and in self-reports regarding bullying experiences (Hara 2002).

In addition to attributing the cause to the victims, Bosacki et al. (2006) found that some students also attributed the cause to the bullies in terms of instrumental motives (e.g., “He wants her lunch/money”) and psychological motives (e.g., “Because she might be jealous, because the other girl is prettier than her” or, “It makes him feel better about himself if the other feels bad”). In Frisén et al. (2007), in response to the question, “Why do some children and adolescents bully others”, 28% of the students reasoned that the bully had low self-esteem, 26% suggested that it made the bully feel “cool”, 15% suggested that the bully had problems, 9% referred to group pressure, 5% suggested that the bully was annoyed with the victim, 4% suggested that the bully was jealous of the victim, 4% referred to lack of respect, and 4% suggested that the bully was also a victim. According to Varjas et al. (2008), many students reported that bullies engaged in bullying to make themselves feel better or gain higher status. Some students also explained bullying as a way of having fun and breaking the boredom in their everyday life (Hamarus and Kaikkonen 2008).

In Thornberg’s (2010b) study, 56 Swedish schoolchildren were interviewed individually. According to the findings, the children operated with one or more types of explanation in order to make sense of bullying incidents they had seen, heard of or been involved in. Seven social representations of the causes of bullying were identified: (a) bullying as a reaction to deviance, (b) bullying as social positioning, (c) bullying as the work of a disturbed bully, (d) bullying as a revengeful action, (e) bullying as an amusing game, (f) bullying as social contamination, and (g) bullying as a thoughtless happening.

The present study adopted a mixed methods (i.e., combining qualitative and quantitative methods) approach to investigate students’ representations of the causes of bullying. Hence, the aim of this study was to explore how teenagers themselves explain why bullying takes place at school in order to better understand their actions as participants or bystanders in bullying situations. Research on children and adolescents’ explanations of bullying indicated that they typically explain bullying in individualistic terms (blaming the victim or the bully). Nevertheless, this has not been tested statistically, since most studies have been conducted by qualitative methods. A second aim in the present study was therefore to test whether teenagers use individualistic explanations more often than nonindividualistic explanations. Furthermore, a few studies have indicated that boys more than girls tended to blame the victim in bullying situations. A third aim was therefore to test whether male teenagers explain bullying by blaming the victim more compared with female teenagers. Finally, earlier studies have not investigated whether young people's explanations of bullying could be linked to their previous bullying experiences. Nevertheless, social developmental and social psychological theories such as the SIP models (e.g., Crick and Dodge 1994), the social representation theory (e.g., Moscovici 2001), and the symbolic interactionism (e.g., Charon 2007) claim that prior experiences of a specific type of social situation have some influence on how children and adolescents interpret people, actions, and other information in a similar actual situation. Therefore, a fourth aim was to investigate whether there were any differences in explaining bullying due to previous bullying experiences.

Method

Participants

The original sample consisted of 185 students (97 females and 88 males) in Grade 9 attending three junior high schools in a medium-sized Swedish town. Out of 185 students, nine (four females and five males) were excluded from the study because of incomplete responses to the questionnaire. Consequently, 176 students (93 females and 83 males) participated in the present study. Hence, the drop-out rate was 5%. Socioeconomic and ethnic background data were not gathered on an individualistic level. Nevertheless, the locations of the three schools in the town ensured that the sample represented students from working-class and middle-class families. The mean age was 15.3 years and ranged from 15 to 16 years old.

Procedures and Materials

This study was a part of a Swedish research project, “School bullying as a social process”, funded by the Swedish Research Council, and approved by the regional ethical review board. The participants were given the questionnaire and filled it out in their ordinary classroom setting. According to ethical guidelines for social science research in Sweden, the researcher has to obtain informed consent by those who participate in the study, and if participants are younger than 15 years old, parental consent has to be obtained as well (HSFR 2009). In our study, the participants were 15–16 years old, and in line with ethical recommendations, informed consent was only obtained by the participants, and not by their parents. One of us was present in every classroom during the data gathering and the definition of bullying, according to Olweus’ (1993) definition, on the very front page of the questionnaire to the participants was read aloud.

Bullying means that a student, or a group of students, repeatedly say or do nasty and unpleasant things to another student. It’s also bullying when others repeatedly exclude a student. But it’s not bullying when two students of about the same strength quarrel or fight. Jokes between friends are not bullying if everyone involved thinks it’s okay. But it’s bullying if a student, or a group of students, repeatedly makes fun of or teases another student in a way he or she doesn’t like.

School Bullying Experiences

Three items tapped in student reports of their past experiences of school bullying (“Have you ever seen someone or some people bullying another person during your school years?”, “Have you ever been bullied during your school years?”, and “Have you ever bullied another person during your school years?”). Student responses to these items were either “Yes” or “No”. Based on their responses, the students were categorized into one of five categories: (a) bystander experience only, (b) victim experience, (c) bully experience, (d) victim and bully experience, and (e) nonbullying experience. Note that students, who were categorized into (b), (c), and (d), might have also reported bystander experience.

Accounts of the Causes of Bullying

Furthermore, an open question was used to ask the participants to explain why they thought bullying occurs, “How come bullying takes place at school? I think bullying takes place because ___________?”, with 16 empty lines provided for their answers (and an explicit suggestion to continue writing on the next page if they needed more space).

Data Analysis

First, grounded theory methods were used in order to qualitatively analyze students’ written responses to the open question about the causes of bullying. Initially we analyzed fifty of the questionnaires by conducting initial or open coding (Charmaz 2006; Glaser 1978). During the initial coding, we read and analyzed the data word for word, asked analytical questions (e.g., What is happening in the data? What is going on? What is the main concern or concerns? What process or processes are at issue here?), and named “segments of data with a label that simultaneously categorizes, summarizes, and accounts for each piece of data” (Charmaz 2006, p. 43). We developed codes by constantly comparing data with data, data with codes, and codes with codes. During the focused coding (Charmaz 2006), we used the set of focused codes (i.e., the more significant and frequent codes) to sift through the large amounts of data in the remaining questionnaires. The focused codes condensed the data. This qualitative analysis generated a set of main categories and subcategories of accounts of the causes of bullying. During the end of the analysis, the broad concept “attribution”—referred in literature to how people infer causes of events (e.g., Fiske and Taylor 2008)—was used as a heuristic tool, meaning that it was employed as a “conceptual frame which helps to understand empirical phenomena found in the research field” (Kelle 2007, p. 2008). There were only few minor disagreements during the coding, and we discussed them until a consensus was reached and the consensus was then coded. Coder reliability was then checked on a random subsample (40 students) of the total sample. Students’ written explanations of bullying in these cases were coded by a third coder, guided by a short manual that consisted of our developed coding scheme, with category labels and definitions. The interrater reliability according to this procedure was 97% agreement.

Second, we calculated percentages of reported bullying experiences and bullying explanations within the total sample as well as within the target subgroups. A dependent t-test was also conducted to investigate if there was a significant difference between the frequency of individualistic explanations and the frequency of nonindividualistic explanations among the students. Finally, Chi-Square tests were conducted to investigate (a) gender differences in school bullying experiences, (b) gender differences in explaining bullying, and (c) differences in school bullying experiences in relation to explaining bullying. Because the participants were divided into five subgroups (bullies, victims, bystanders, victim-bullies, and none) regarding school bullying experiences, when conducting (c), we had to conduct a Chi-Square test in which all subgroups’ means were compared first, and then we had to construct and conduct a stepwise post hoc test controlling for α = .05 (cf., Thornberg 2006). The first step in the post hoc test procedure was to rank all the mean differences from the largest to the smallest between all five groups. Then the largest difference was tested by a Chi-Square test. If this difference was significant, the next step was to test the second largest difference with a Chi-Square test. If this difference was significant, the third largest difference was tested with a Chi-Square test, and this procedure continued until nonsignificance was detected. In order to maintain α = .05 across all the comparisons, the significance was tested according to α′ = α/c in which c refers to the number of operated comparisons. In the first step c = 1 and therefore α1′ = .05. In the next step (when the second largest difference was tested) c = 2 and therefore α2′ = .025. In the third step c = 3 and therefore α3′ = .01667, and so on. This procedure (the initial Chi-Square test and then the stepwise post hoc test) was conducted twice in order to investigate differences in bully attributing and victim attributing as well.

Results

According to an initial descriptive statistical analysis of the teenagers’ reports of school bullying experiences, 25% of them had been a victim of bullying, 19% had acted as a bully, 11% had experiences of both being a victim and a bully, 30% reported that they had only been involved in bullying as a bystander, and 15% claimed that they had no experiences of school bullying at all (see Table 1).

Table 1 Reported experiences of school bullying
Full size table

Even if Table 1 indicated that more boys reported that they had bullied others as compared to girls, this difference was not significant (χ2 = 2.646, p = .104). Neither were there any significant differences between boys and girls regarding their reports of being victimized (χ2 = 1.339, p = .247), or bystanders only (χ2 = 3.341, p = .068). The next step was to analyze students’ accounts of the causes of bullying with GT methods, which generated five main categories (bully attributing, victim attributing, peer attributing, school attributing, and human nature/society attributing) and 26 subcategories (see Table 2). A dependent t-test, in which individualistic attributions (an index of bully and victim attributing) and nonindividualistic attributions (an index of peer, school and human nature/society attributing) were compared, revealed that the students explained bullying much more in individualistic terms than in nonindividualistic terms (t = 13.643, p = .000, r = .72).

Table 2 Main categories and subcategories of bullying explanations (the percentage of the total sample using each category is reported in the brackets)
Full size table

Bully Attributing

Bully attributing refers to attributing the cause of bullying to the bully. Sixty nine percent of the teenagers did so. The analysis resulted in 10 subcategories. The account of bully’s inner flaws, means that the bully feels “bad” or insecure, has poor self-confidence, low self-esteem, psychological problems or a weak mind, which cause him or her to bully others (e.g., “The bullies feel bad or have poor self-confidence”, “Weak minds”, “The reasons why they bullied us were their problems with self-esteem and self-confidence”). According to the account of social positioning, bullying takes place because those who bully others want to manifest, maintain or enhance their power, status or popularity (e.g., “Some people want to show off as ‘popular’. They think they are so cool, and really want to show that by bullying someone else”, “You want power”), or they try to protect themselves from social exclusion, harassment or bullying (e.g., “they want to prove that they are ‘tough’ so nobody else bullies them back”). Inner flaws and social positioning were the two most common subcategories of bully explanations (32% each of the total sample). According to the account of bully’s hostile feelings, the bully is driven by hostile feelings or emotions, such as jealousy, a lot of inner anger or just dislikes the victim, and therefore begins bullying (e.g., “they take out their anger on other people”, “I also think jealousy is a reason for many kids becoming bullies”). The account of boosting well-being means that the bully bullies others to feel better or enhance his or her self-confidence or self-esteem (e.g., “It’s mostly about boosting their self-confidence, self-esteem”).

According to the account of problematic family of bully, the bully is a product of his or her problematic family background, such as poor parenting, lots of quarrelling or conflicts at home, divorce, abuse, or harsh or nonloving parents (e.g., “I also think bullies mostly had an insecure and sometimes loveless childhood”, “Those kids who are bullies are usually the ones with a tough situation at home”). The account of bully’s bad personality is a kind of immoral character explanation. The bully is a bad, immoral or mean person, has an “immoral” self-concept, or has no empathy, and therefore he or she starts bullying others (e.g., “They don’t understand how the person being bullied feels and have absolutely no feelings for the victim”, “Because they are bad”, “Certain persons think they are better than others, which gives them the right to do what they want to other people”). The account of doing it for fun means that bullying occurs because the bullies think it is a funny or amusing thing to do, and they just want to have some fun (e.g., “One or more kids want to have some fun”, “Others who tease other kids with jokes might want to make the other kids laugh and have some fun themselves too”). According to the account of deviancy-intolerant bully, the bully is intolerant of deviant or different people and therefore bullies students he or she perceives as deviant, odd or different. Here the original cause is not the deviant victim (see “deviant victim” in victim attributing) but the bully’s negative attitude to deviance (e.g., “They don’t accept the way other people are”). The account of bully’s prior victim history means that the bully has been bullied him/herself, and as a consequence he or she now bullies others (e.g., “Those people who bully others have probably been bullied when they were younger”). Finally, the account of media influencing means that the bully has been exposed to and influenced by computer games and movies (“Some young people, who have been badly brought up, create their own concept of what’s right from computer games and movies”).

Victim Attributing

Victim attributing refers to attributing the cause of bullying to the victim. Forty-two percent of the teenagers operated with victim explanations. The analysis resulted in six subcategories of victim attributing. The most common one was the account of deviant victim, which means that the victim is deviant, different or odd, which in turn leads to being bullied (e.g., “He/she is different, maybe wears special clothes”, “Those kids who have an odd personality or look a bit weird get bullied”). This was actually the most represented subcategory of all the 26 generated subcategories across the five main categories in the data (37% of the students in the total sample used the explanation of deviant victim). According to the account of irritating victim, the victim is irritating, which provokes bullying (e.g., “It’s someone who is irritating, and then you snap back at them with a word or two, maybe more”). The account of weak victim means that the victim is weak in some way, such as physically weak, poor self-confidence, insecure, shy or has no friends, which leads to bullying (e.g., “Those who are weak and can’t defend themselves”). According to the account of mean victim, the victim has done something bad or mean, and bullying takes place as a revenge or payback (e.g., “If you snitch, you might become a victim”, “Many kids who are bullied have done or said something mean to the person who bullies them, I think”). The account of victim’s prior victim history means that the victim has a history of being bullied, and therefore he or she continues to provoke bullying, even in a new class. The account of problematic family of the victim means that bullying occurs because the victim has a bully-attracting behavior caused or produced by his or her problems at home.

Peer Attributing

Peer attributing refers to attributing the cause of bullying to the peer group. Twenty-one percent of the teenagers used this type of explanation. The category peer attributing in turn consisted of four subcategories. The most frequent subcategory of peer explanation was the account of group pressure (13% of the sample), which means that bullying is simply a consequence of group pressure (e.g., “feels like a kind of group pressure”). According to the account of conflicts, bullying arises out of conflicts or competition between peers or peer groups (e.g., “if you’ve been involved in a conflict, then some kids take certain sides and pick on some special kids”, “Conflicts often create bullying”). According to the account of group reinforcing, bullying takes place because the peer group reinforces the bully to torment someone by giving him or her positive attention or other kinds of rewards (e.g., “If a lot of people bully another kid, then the others in the group might give you the respect you want. If they give you respect when you bully someone else, of course it will continue”). The account of group norm of deviant intolerance means that bullying occurs because the group dislikes and reacts negatively to a peer if he or she is perceived as deviant. The group does not tolerate deviance.

School Attributing

School attributing refers to attributing the cause of bullying to the school setting. This was quite a rare explanation among the teenagers. Only seven percent operated with school explaining. This category consisted of two subcategories. The account of boredom in school means that bullying takes place because students have nothing to do or are bored as a result of the school life or structure (e.g., “Mostly when there isn’t so much to do”; “When the school makes people bored”). According to the account of poor antibullying practices, bullying occurs because teachers do not care or do not intervene, poor teacher supervision during recesses, or the school has poor antibullying practices in general (e.g., “If there aren’t enough adults to supervise the students, bullying might take place”, “There aren’t enough school staff who can stop the bullying. If there were more staff in the corridors, the bullying would probably be much less”).

Human Nature/Society Attributing

Human nature/society attributing refers to attributing the cause of bullying to human nature or society. This was also a quite rare explanation among the students. As in the previous category, seven percent of the teenagers explained bullying in terms of human nature or society, which in turn consisted of four subcategories. The humans are different account refers to an idea that the very nature of differences among humans results in friction, conflicts, violence and bullying (e.g., “People are different … Differences create conflicts, violence and bullying”). According to the account of biological instinct of deviancy-rejection, people tend to reject deviant individuals as a result of biological instinct (“People often reject anyone who is deviant in some way. It’s our animal instinct”). Hence, the explanation here is not just about specific characteristics or attitudes in specific bullies, but human characteristics rooted in the common biology that all humans share. The account of appearance-focused society means that our society strongly values good-looking people or a pleasing appearance, which in turn causes bullying when people are not good-looking enough (e.g., “Today, people just care about appearances and start to spread gossip about someone before they even know the person in question”). Finally, another reported cause of bullying is the account of poor mental health among young people.

Bullying Explanations and Gender

Table 3 shows the percentages of how many boys and girls using the different main categories of bullying explanations. In order to ascertain whether gender differences existed in bullying explanations, two Chi-Square tests were conducted. We investigated gender differences in relation to victim attributing (Test 1) and bully attributing (Test 2). We dropped testing gender differences in relation to peer attributing since Table 3 displayed very small percentage differences. In addition, gender differences regarding the other main categories of bullying explanations were not investigated because of the small number of individuals in these cells. The analysis revealed that significantly more boys (52%) attributed causes of bullying to the victim compared to girls (33%, χ2 = 6.143, p = .013). In contrast, significantly more girls (80%) attributed causes of bullying to the bully compared to boys (57%, χ2 = 1.075, p = .001). Hence, more girls explained bullying by blaming the bully and fewer explained it by blaming the victim, compared to boys.

Table 3 Percentages of girls and boys using different bullying explanations
Full size table

Bullying Explanations and Bullying Experiences

Table 4 shows the percentages of how many teenagers with different school bullying experiences (been a victim but not a bully, been a bully but not a victim, been a victim and a bully, been a bystander only, and lack of bullying experience) using the different main categories of bullying explanations. In order to ascertain whether differences between these groups existed, two Chi-Square tests (including follow-up post hoc tests) were conducted. We investigated whether differences in school bullying experiences were related to victim attributing (Test 1) and bully attributing (Test 2). We dropped testing the other main categories of bullying explanations because of the small number of individuals in these cells.

Table 4 Percentages of teenagers with different school bullying experiences using different bullying explanations
Full size table

The analysis indicated a significant difference between bullying experience and victim attributing (χ2 = 1.988, p = .001). The follow-up post hoc test revealed that significantly more bullies (74%), compared to bystanders (27%) (χ2 = 1.802, p = .000), victims/bullies (35%) (χ2 = 7.743, p = .005), and victims (36%) (χ2 = 1.062, p = .001), attributed causes of bullying to the victim. Moreover, a significant difference between bullying experience and bully attributing was found (χ2 = 2.787, p = .000). The follow-up post hoc test showed that significantly fewer bullies (35%), compared to victims (84%) (χ2 = 1.955, p = .000), bystanders (81%) (χ2 = 1.820, p = .000), and victims/bullies (75%) (χ2 = 7.941, p = .005), attributed causes of bullying to the bully. Hence, bullies were more likely to explain bullying by blaming the victim and fewer by blaming the bully, compared to victims, bystanders, and victims/bullies.

Discussion

In the present study and in line with our first aim, we identified several accounts among school teenagers of why bullying occurs. The qualitative analysis revealed several ways youth explain bullying. The five main categories were bully attributing, victim attributing, peer attributing, school attributing, and human nature/society attributing (with 26 subcategories)—showing that teenagers might operate with a variety of explanations of bullying. Whereas earlier studies (e.g., Erling and Hwang 2004; Frisén et al. 2007, 2008) have indicated that children and young people typically used individualistic explanations in relation to bullying, these studies have not tested this tendency statistically, which the present study actually has done. Thus, in relation to our second aim, the present findings add to the current body of research by showing that the teenagers explained bullying significantly more in individualistic terms (bully or victim attributing) than in nonindividualistic terms (peer, school or human nature/society attributing). While this study did not examine the reasons for this, one might speculate that this tendency might be explained in terms of the fundamental attribution error (or correspondence bias), which is the most commonly documented bias in social perception, and is about overattributing behaviors and social events to individual dispositional causes and at the same time overlooking other forces such as social norms or social pressure (Fiske and Taylor 2008). Fundamental attribution bias may be responsible for the tendency among the teenagers to attribute bullying to the victims or bullies rather than situational, social or cultural conditions.

Some of the subcategories in the present findings resemble student accounts from other studies. For example, the account of the deviant victim is found in many studies investigating student perspectives (e.g., Bosacki et al. 2006; Frisén et al. 2007, 2008; Hamarus and Kaikkonen 2008; Hoover et al. 1992; Teräsahjo and Salmivalli 2003; Thornberg 2010b; Varjas et al. 2008) and can be associated with the social misfit hypothesis (Wright et al. 1986), stigma theory (Goffman 1963) and labeling theory (Becker 1963; Phelan and Link 1999). Ethnographic research on bullying clearly indicates that such theories (i.e., a peer group defines or constructs the victim as deviant—a label that results in stigma and rejection) might be adopted to understand the social process of peer harassment and social rejection (for a review, see Thornberg in press). One or more of the other subcategories generated in the present study, such as the teenagers’ accounts of bully’s inner flaws, social positioning, hostile feelings, boosting well-being, and bad personality, could also be related to findings from earlier studies (e.g., Bosacki et al. 2006; Frisén et al. 2007; Thornberg 2010b; Varjas et al. 2008).

In relation to our third aim, the present findings demonstrated that many more girls attributed causes of bullying to the bully and fewer attributed the causes to the victim, compared to boys. This confirmed previous research findings showing that boys blamed the victims more than girls in hypothetical bullying situations (Gini 2008) and in self-reports regarding bullying experiences (Hara 2002). Such gender differences in how to explain bullying and thus tendencies to blame the victim versus blaming the bully, might at least in part contribute to a better understanding of why girls are more likely to show positive attitudes towards victims (e.g., Menesini et al. 1997; Rigby 1996; Rigby and Slee 1993), and are more likely to support or defend victims and less likely to reinforce bullies than boys (O’Connell et al. 1999; Salmivalli et al. 1996; Salmivalli and Voeten 2004). In hypothetical bullying situations, whereas girls endorsed defending behavior more than boys, boys endorsed the behavior of assisting the bullies more than girls (Gini et al. 2008). In a recent study, among those students who reported that they had witnessed bullying, girls were significantly more likely to report that they told the bully to stop, talked to the bully’s friends, helped the victim, got friends to help solve the problem, and talked to another peer as compared to boys (Trach et al. 2010).

In contrast to earlier studies on children and adolescents’ perspective on bullying causes, in the present study we have investigated if bullying explanations could be related to prior bullying experiences. In relation to our fourth aim, our findings showed that many more bullies attributed the causes of bullying to the victim and much less to the bully (cf., Hara 2002). These findings might be explained in terms of a self-serving bias among bullies. With reference to the attribution theory, Gini (2008) argues that “in the case of potentially harmful events, blaming other individuals is a very real self-serving attribution and, in particular, blaming victims for their fate allows people to distance themselves from thoughts of suffering” (p. 337). In their study, Hymel et al. (2005) actually found that students who did not report engaging in bullying reported the lowest levels of moral disengagement whereas those students who repeatedly bullied others reported the highest level of moral disengagement. Among the moral disengagement items, those that emerged as significant predictors of bullying were efforts to justify bullying as “okay” and efforts to blame the victims.

Limitations of the Study

Some limitations of this study should be noted. First, we could not test for interaction effects between gender and bullying experience because of the variable scale level (nominal scale). Second, the use and construction of the stepwise post hoc test in investigating whether differences in school bullying experiences were related to victim attributing and bully attributing could be problematized since the power of the test decreases as the number of comparisons increases, and consequently the probability of making a Type II error (i.e., concluding that there is no difference when, in reality, there is) is higher. However, we used 5% instead of 1% as the level of significance in order to reduce the risk of making a Type II error, which of course increases the risk of making a Type 1 error instead (i.e., concluding that there is a difference when, in reality, there is not). Nevertheless, conducting post hoc tests is in itself a way of minimizing the probability of making a Type I error for the set of all possible comparisons by correcting the level of significance for each test so that the overall Type I error rate (α) across all comparisons remains at .05 (Field 2005). Third, only using “Yes” or “No” as answers for prior bullying experiences could be criticized as a rather blunt measurement, not saying anything about intensity, duration, and type of bullying (physical, verbal, or indirect). Nevertheless, at least the item tapped into the participants’ subjective perceptions of whether they had been bullied, had bullied others, or had been witnessing bullying during their schooling.

Fourth, an open question was used to ask the students to explain why they thought bullying occurred, which might result in just replying with what spontaneously came into their minds without any further elaboration and more explanations that might be elicited with follow-up questions, which would had been possible in, for example, qualitative interviews. However, with reference to SIP models (e.g., Crick and Dodge 1994; Gifford-Smith and Rabiner 2004; Fiske and Taylor 2008), the participants’ own formulated answers in the questionnaire might be interpreted as a result of easy accessible information in their long-term memory, and hence indicating it was more influential on their on-line processing of social information in bullying situations than other explanations of bullying, which they might have in their long-term memory. Furthermore, the risk of a social desirability bias is much higher in qualitative interviews as compared to filling out questionnaires anonymously, especially considering the delicate topic of bullying. Fifth, a note of caution needs to be sounded regarding the generalization of the findings. This sample of teenagers from a particular area of Sweden may or may not be similar to the population of teenagers with whom the readers primarily work.

Implications for Practitioners

The findings in this study have some important implications for prevention and intervention efforts designed to reduce bullying among children and adolescents. The findings suggest that bullying prevention efforts and interventions should investigate and target teenagers’ conceptions of the causes of bullying, since such mental representations are a source of interpretations, attitudes and behavior in real situations (cf., Crick and Dodge 1994). First, practitioners have to investigate and reflect upon how teenagers think and reason about why bullying occurs. They have to build their instructions, explanations, conversations, and practices on teenagers’ contemporary repertoire of knowledge, conceptions, and skills regarding bullying and other social and moral issues. “Moral growth comes about through the child’s progressive construction of ways of understanding the world, and not just an accommodation to the positions and practices of adults and society” (Nucci 2006, p. 663). Research has shown, for example, that effective teachers adapt their teaching to their learners’ prior experiences, knowledge, and conceptions, and encourage their learners to use their own experiences to actively construct and reconstruct understandings that make sense to them and for which they can “take ownership” (see Borich 2007).

Second, the current findings show that individualistic explanations were the most prevalent bullying explanations among the teenagers. By almost exclusively attributing the causes of bullying to the bully or the victim, teenagers risk overlooking other factors that might cause or influence the process of peer harassment. Bullying prevention efforts and interventions have to help children and adolescents to discover, understand and consider the complexity of bullying and factors such as social participation roles in bullying and the power and responsibility of bystanders (O’Connell et al. 1999; Salmivalli 2010; Salmivalli et al. 1996), the everyday process of making and maintaining friendships by defining and excluding nonfriends (Besag 2006; Mishna et al. 2008; Owens et al. 2000; Thornberg in press), instability in peer networks (Besag 2006; Neal 2007), the power of group norm settings by popular classmates (Dijkstra et al. 2008), social hierarchy (Frisén et al. 2008; Kless 1992; Neal 2007; Thornberg in press), probullying norms (Duffy and Nesdale 2009; Salmivalli and Voeten 2004), group processes and group pressure (Bukowski and Sippola 2001; Burns et al. 2008; Hamarus and Kaikkonen 2008), social representations or peer discourses about victims and bullying (Teräsahjo and Salmivalli 2003; Thornberg 2010b), gender (Kless 1992; Neal 2007; Phoenix et al. 2003; Stoudt 2006), heterosexual hegemony (Phoenix et al. 2003; Ringrose 2008), and intolerance of diversity (MacDonald and Swart 2004; Merton 1994; Thornberg in press) in peer culture. Hence, it appears to be urgent to deepen children’s and young people’s insights into peer attributing, school attributing, and human nature/society attributing, since these are atypical bullying explanations among the young people in our current findings.

Third, teenagers’ gender and social roles in prior experiences of bullying at school have to be considered when designing and conducting bullying prevention and intervention practices among young people, as the current findings indicate that these factors appear to affect how they attribute causes and attribute blame for bullying. Since we found that many more female teenagers explained bullying by blaming the bully and fewer of them by blaming the victim, compared to male teenagers, practitioners have to be aware of these gender differences, and especially targeting boys’ tendency to blame the victim. Furthermore, initiating classroom and small group discussions about bullying across gender as well as discussing prior experiences of peer harassment might be a way of improving young people’s insights and sensitivities regarding bullying as well as challenging their taken-for-granted assumptions of why bullying takes place and who to blame.

Fourth, the present study revealed that former bullies most often explain bullying by blaming the victims as compared to others who most often blamed the bullies. Bullying prevention and intervention programs have to deal with these differences based on the teenagers’ own experiences of participation roles in bullying. Blaming the victim as a self-serving bias among bullies helps them to minimize or diffuse their own role and responsibility as well as hindering their empathic arousal and moral concerns regarding the bullying situations, the victim’s distress, and their own actions (cf., Bandura 2002; Hoffman 2000). Moreover, the predominant victim attributing by explaining bullying by referring to “deviance” of the victim among the teenagers who reported experiences of bullying others (62% of the bullies as compared to 25% of the bystanders, 32% of the victims, 30% of the victims/bullies, and 46% of those who reported no bullying experiences) indicates an underlying logic of conformism and intolerance, and can be challenged by pointing out and inviting bullying teenagers (as well as other teenagers) into a deliberative discussion about the mechanisms of stigma, labeling, and dehumanization, and the values of multiplicity, heterogeneity, social inclusion, a caring community, and tolerance.

Finally, the tendency to blame victims increases the risk of moral disengagement (Bandura 2002; Hara 2002; Hymel et al. 2005), which in turns makes it easier to act as bullies, reinforcers (actively supporting the bullies by laughing and so on), and passive bystanders. Furthermore, it is reasonable to assume that blaming the bully easily results in a responsibility transfer—instead of feeling a personal responsibility to help the victim, the responsibility of the situation is transferred to the transgressor, resulting in a responsibility loss among bystanders (cf., Thornberg 2010a).