Bullying is defined as deliberate, repeated aggressive actions among children. It may involve physical contact and words, or it may be carried out by using body language in a negative way and excluding one person from a peer group. Also, it involves a power imbalance between the perpetrator and the victim due to size, strength etc. (Olweus, 1997; Monks & Smith, 2006). World Health Organization (WHO) (2015) indicates that bullying and physical fighting are frequent among young people. Approximately 42% of boys and 37% of girls in developing countries have been exposed to bullying.

There is no evidence of any studies on the prevalence of bullying in Turkey. But studies from different cities of the country indicate that the frequency of being a bully among adolescents and pre-adolescent varies between 2.8 and 30.2%, the frequency of being a victim range between 18.3 and 41.3% and the frequency of being both a victim and bully changes between 6.2 and 29.9% (Gökkaya & Tekinsav Sütcü, 2018; Kartal & Bilgin; 2007; Pişkin, 2010; Siyez & Kaya; 2011; Hesapçıoğlu & Yeşilova, 2015). A comprehensive study in Istanbul reveals that victimization rate is 51.26%, and bullying rate is 37.9% and younger students report victimization more often than the older ones (Dölek, 2002).

These rates are important when considering the negative consequences of bullying. Being a bully victim enhances the chance to show internalized and externalized behaviour problems and complaints about pain in both girls and boys (Lien, Green, Welander-Vatn, & Bjertness, 2009; Bond, Carlin, Thomas, Rubin, & Patton, 2001). Frequent excessive drinking and use of any other substance are more common among bullies and thereafter among bully-victims than victims (Kaltiala-Heino, Rimpelä, Rantanen, & Rimpelä, 2000). Different internalizing behaviours such as anxiety, depression, somatic complaints and withdrawn behaviours are predicted differently by direct (physical and verbal) or indirect (psychological and relational) types of bullying behaviours (Baldry, 2004). According to their bullying experiences students can be placed in different groups as bully, victim or bully/victimized and they can be affected negatively even if they are only the witnesses (Rivers, Poteat, Noret, & Ashurst, 2009). Therefore, understanding the bullying experience predictors, related developmental factors and social variables is important to decide and intervene the key points in prevention programs. In this study, attachment to fathers and moral development level of adolescents are considered important both in terms of their relationships with each other and the possibility of influencing different bullying experiences in different ways.

In the most general sense, moral development can be described as thoughts, actions, and feelings about norms of what is right and what is wrong (Gray, Culpepper, & Welsh, 2012). Moral development represents not only one's learning about cultural values, but also changes in his or her mind-set or way of thinking. As the person matures and progresses in the cognitive development stages, the reference point of the judgments he uses to solve moral dilemmas changes. The person refers to the personal consequences of his behaviour, such as reward or punishment, at the beginning of his development. In the following stages, this person focuses on the values and satisfaction of the family, community and society to which he belongs. At the last stage, regardless of a particular authority, he focuses on general rights and standards and ethical principles (Kohlberg & Hersch, 1977).

Students' level of moral maturity as one of the indicators of the capacity of to understand and evaluate the harmful consequences of a behaviour and usage of moral disengagement mechanisms such as comparing a harmful behaviour with more detrimental one to rationalize his/her harmful behaviour may be considered as a relevant factor in understanding bullying behaviour (Hymel, Schonert-Reichl, Bonnano, Vaillancourt, & Rocke-Henderson, 2010). Arsenio and Lemerise (2004) also say that relationship between moral reasoning and aggressive behaviour in normal population has not been investigated extensively. Some research has shown an association between moral disengagement and higher levels of aggressive behaviour (Menesini et al., 2003) and moral disengagement is a predictor of those kinds of behaviours (Hymel, Rocke-henderson, & Bonanno, 2005).

Because bullying is a multifaceted phenomenon, researchers must consider family related variables in addition to developmental ones. Among these variables, attachment is an important psychological and relational quality that determines people’s relationship styles (Bartholomew & Horowitz, 1991). Bowlby (1980) defines attachment behaviour as “any form of behaviour that results in a person attaining or retaining proximity to some other differentiated and preferred individual”. Attachment is one of the variables that form the ground where a person learns moral judgment. By early communication with the caregiver, the child gains an internal representation of rules that regulate how wrong/right judgements are to be defined, used and understood (Govrin, 2014). If attachment behaviour deteriorates, as in children from neglectful and hostile backgrounds, they internalize these experiences as representations of the world and use them as the filter through which to make meaning of other forms of relationships. They become more prone to interpret ambiguous cues as hostile and threatening, have negative beliefs and use aggressive defences (Palmer, 2000). Researchers found out that tendency of attachment-related anxiety and avoidance negatively influenced compassion, personal distress, an altruistic behaviour and secure attachment produce foundation for care-oriented feelings and caregiving behaviours, whereas insecure forms of attachment interfere with compassionate caregiving (Mikulincer, Shaver, Gillath and Nitzberg, 2005). Although more anxious individuals have samples of prosocial attitude, this positive element includes a mixture of narcissistic motives, personal distress, self-focused fears and sometimes envious or hostile approach towards others (Shaver & Mikulincer, 2012).

Feelings of attachment to one or two parental figures in critical life periods such as adolescence may have specific significance in youngsters’ behaviours. In this period, developing peer relations will be affected by the quality of the attachment established with the parent or caregiver in previous years (Allen & Land, 1999). Insecure attachment may be related to whether adolescents show violent behaviour to their peers. Indeed, in both the bully and the victims, anxious attachment type is more common than those who have never experienced bullying (Kõiv, 2012). And securely attached adolescents have been involved less in bullying behaviour (Murphy, Laible, & Augustine, 2017). Nickerson, Mele, and Princiotta (2008) indicate that among middle school students, attachment to mother contributes whether a person is a defender who mediates the violence between the bully and the victim actively or an outsider who was not involved in bullying experiences. And insecure attachment in preadolescents is related to more involvement in both bullying and victimization (Kokkinos, 2013). The pattern of attachment to father contributes to the bullying behaviour differently according to gender and context (Guinn, 2015; Williams & Kennedy, 2012; Fang, 2018).

The aim of this study was to investigate whether moral maturity and attachment to father are associated with specific types of bullying experiences (such as being a bully, being a victim, being both bully and victim, or being neither). In addition, relationships of other socio-demographic and past experience factors were evaluated. The reason for choosing father attachment instead of mother bonding is the opinion that there are few studies on this issue (Sulejmanovic, 2017) and data showing that attachment to father may be a predictor of estimating adolescent's externalization behavior (Williams & Kelly, 2005).

Method

Sample and Participant Selection

The research was carried out in the district (Maltepe) of the university where the study was conducted. The reason for this was the ease of access to schools and students, as well as the possibility to use the data to be obtained from the region in the planning of future social responsibility projects of the university. Maltepe district secondary education institutions have a total number of students, universe of the study, approximately 10.000, and at least 644 participants must be reached in order to conduct the study with 99% confidence interval and 5% margin of error.

The research data were collected by convenient sampling, in terms of both the choice of school and the classroom, in which suitable ones were preferred for the conduct of the study. When the researcher went to a school, if there was no other study with the students and the school management helped organize the data collection and there was no reason for the students to be away from the school during the survey, that school was chosen. When the required number of people was reached, data collection was stopped.

Ten students refused to participate in the study and sixteen could not be categorized in four bully experience groups to incomplete information about their bullying experiences. These four group categorized according to the answers given to The Revised Olweus Bully/Victim Questionnaire. Those who only reported being bullied and not bullying anyone were called the victim group. Those who reported that they were just bullying and were not victims of bullying were called the bully group. Students who were both bullying and were bullied classified as both bully and victim group. Neither bully nor victim group consisted of adolescents who were neither bully nor victim. In the analysis made to compare these 4 groups, the number of participants is 632.

Study was conducted with 9th, 10th and 11th grade students from six high schools in Maltepe district in Istanbul. 12th grade students usually prepare for the university exam in the spring semester when the study is conducted and their work schedules can be intense or often obtain permission from the school. Due to this they were excluded from the study. Mean age of the participants (n = 648) was 15.86 (SD = 0.94) and 62.7% (n = 410) of them were females. Distribution of the participants according to their grades was as follows: 34% (n = 222) were in grade 9; %41.3 (n = 270) were in grade 10 and 24.7% (n = 161) were in grade 11.

Procedure

Informed consent forms were prepared for both participants and parents in the format desired by the university ethics committee. These forms briefly include the purpose and content of the research, the fundamental rights such as not participating in the study, leaving the study, and no negative consequences will arise in the condition of leaving the study.

In the guidance and psychological counselling courses, where surveys are applied, activities related to students’ getting to know and accept themselves are carried out. In addition, training is provided on the development of skills such as problem solving, communication, coping with stress. Guidance is made in terms of professional planning and hobby development.

Students participated in the study voluntarily and the questionnaires were obtained anonymously. It took about 20 min for a student to complete the questionnaire. School counsellors were allowed to see a summary of the results (not individual questionnaires) to evaluate the school's situation. District National Education Directorate and University Ethical Committee provided necessary permissions.

Statistical Package for the Social Sciences 14.0 (SPSS 14.0) was used for descriptives, frequencies, correlation, Chi-square tests, Kruskal–Wallis Test, Spearman correlation and Multinomial Regression Analysis.

Measures

Variables included in the analysis are different bully experiences (as independent variable), demographics, moral maturity score and attachment to father score (dependent variables).

Independent Variables

Bullying Experiences

The Revised Olweus Bully/Victim Questionnaire was translated and got back translated by Dölek (2002) and its internal consistency values were 0.71 for victimization and 0.75 for bulling. This scale was chosen because it was used safely in many studies in Turkey and it gave information not only about bullying but also victimization. In addition, using this scale, it is possible to distinguish both the bullied and the victims of bullying according to the answers given. In this study, a part of the questionnaire (16 questions) was used to classify study groups and this part of the questionnaire was to depict the frequency of different bully and victimization behaviours. In this part of the questionnaire bullying was first described and then the participants were asked about the frequency of showing such behaviour during the semester they were in and how often they were the victims of such behaviour. Secondly, the types of bullying behaviour are listed and the frequency of showing these behaviours and being the victim of these behaviours was asked.

This questionnaire does not have a special cut-off point or categorization criteria. The higher the score the more frequency of a behaviour.

Other questions that we did not use in this study were about teachers’, parents’ and students’ reactions to bully behaviour etc. (Atik, 2006) (Sample item: “I was called mean names, was made fun of, or teased in a hurtful way.”).

Dependent Variables

Demographics and Other Related Variables

Twelve questions were asked to identify gender, perceived economic condition, family characteristics and experience of different life events. Witnessing any kind of violence in a life-time was measured by one yes–no question.

Paternal Attachment

The revised version of Inventory of Parent and Peer Attachment (Father Form-Short) is to describe adolescents’ perceptions of the positive and negative affective and cognitive dimension of relationships with their parents and close friends. This 12-item instrument is a self-report questionnaire with a five point Likert-scale (1-never, 5-always) response format. For parental attachment Cronbach’s Alfa, internal consistency coefficient was 0.82. If the score was higher, the relationship with the father was described better (Sample item: “My father accepts me as I am”) (Kumru & Carlo, 2004). The scale does not have a special cut-off point or categorization criteria. Only the height of the score indicates that the measured feature is higher.

Moral Maturity

The Scale of Moral Maturity consists of 66 items, and prepared in the form of Likert Scaling, and it aims to test the moral maturity of the adolescents. Validity and reliability study was conducted with high school students. Higher scores indicate higher moral maturity. Item-total correlation, criterion related validity, and different types of reliability analyses yielded that the scale is valid and reliable. Higher score showed higher moral maturity (Sample item: “I reflect on my behaviours and try to correct my mistakes.”) (Şengün & Kaya, 2007). This scale also does not have a special cut-off point or categorization criteria. Only the height of the score indicates that the measured feature is higher.

Results

Demographics and Life Experiences

Mean number of siblings was 2.52 (SD = 1.14) and the participants were mostly living with their nuclear family (97.9%, n = 639). Some life experiences of the participants are shown in Table 1.

Table 1 Critical life experiences of the participants
Full size table

Categorization of students by their bullying experiences as follows: 40.5% (n = 256) were neither bully nor victim; 13.3% (n = 84) were only bully; 20.3% (n = 128) were only victim and 25.9% (n = 164) were both bully and victim. Further analyses were executed to compare these groups.

Bullying Experiences and Demographics

Kruskal–Wallis Test indicated no differences among groups in the mean rank of siblings (p > 0.05) and perceived socio economic status (p > 0.05). Chi-square Test did not reveal any significant difference in the rates of school course failure (p > 0.05) and parents’ divorce (p > 0.05). Table 2 presents gender distribution of the groups [X2(3) = 13.88, p = 0.003].

Table 2 Bullying experiences by gender
Full size table

Group distributions were varied by grade level (9th, 10th or 11th) [X2(6) = 17.71, p = 0.007, (Table 3). In the only bully group the rate of 9th grades was the lowest.

Table 3 Bullying experience by grade level
Full size table

Groups showed significantly different distribution of witnessing violence (such as theft, physical conflict, murder etc.) [X2(3) = 18.86, p = 0.000, While neither bully nor victim groups had the highest rate of not witnessing to a violent act, students in the both bully and victim group had the highest rate of witnessing to such an event (Table 4).

Table 4 Bullying experiences by witnessing a violent event
Full size table

Bullying Experiences, Attachment to Father and Moral Maturity

Kruskal–Wallis test indicated differences among bullying experiences in attachment to father scores, X2(3, n = 541) = 10.61, p = 0.014, and in moral maturity scores, X2(3, n = 426) = 57.15, p = 0.000. Mann–Whitney U test was used for pairwise comparisons (Table 5, 6).

Table 5 Pairwise comparisons of attachment to father score between bullying experience groups
Full size table
Table 6 Pairwise comparisons of moral maturity score between bullying experience groups
Full size table

Correlation Between Attachment and Moral Maturity

Positive correlation was found between attachment to father score and moral maturity score (r = 0.265, p = 0.001).

Predictors of Bullying Experiences

In the multinomial logistic regression, the bullying experience classification (neither bully nor victim group was the reference category) was the outcome variable. And age, gender, attachment to father score, moral maturity score and witnessing a violent event were the predictors. The overall chi-square for the model was found to be significant (X2(15) = 97.75, p = 0.000). Variables other than attachment to the father predicted belonging to bullying experience groups (Table 7).

Table 7 Multinomial logistic regression results
Full size table

Discussion

Bullying definition requires the power imbalance between the perpetrator and the victim, and this study shows that bullying is more common in the 10th and 11th grades than in 9th grades. According to the study of Ayas and Pişkin (2011), 9th grade students exhibit less bullying behaviours than the 11th grades. It is also stated that both bullying and victimization increase with age (Haynie et al., 2001; Turkmen et al., 2013). Being older may bolster above-mentioned power imbalance by providing more psychosocial equipment to the bully. While 9th grades are at the beginning of the high school process and need time to adapt to the new environment, the older ones have already learned what school rules are and how they would manipulate them. This transition period may make the younger students more vulnerable to be bullied. In addition, older students already have the chance to create their own social environment that provides them with functional or dysfunctional support.

Gender distribution in bullying experiences varies from one study to another due to different definitions and study designs. While some studies indicate that boys exhibit more bullying behaviour and are more likely to be bullied than girls (Griezel, Finger, Bodkin-Andrews, Craven, & Yeung, 2012; Hoertel, Strat, Lavaud, & Limosin, 2011) others suggest that both sexes have aggressive behaviours but in different forms (Crick & Grotpeter, 1995; Rivers & Smith, 1994).

In this study, the proportion of boys in the bully group was higher than in the girls’ and the rate of girls in the victim group was higher than boys’, and being a girl was a predictor of being in the victim group. Seals and Young (2003) indicate that percentage of males in bullying group is high and Turkmen et al. (2013) says that the probability of a male student being involved in violence was 8.4 times more than a female student. Although Carlyle and Steinman (2007) indicate enhanced male perpetration rates for 7th–9th grades, gender effect varies by ethnicity in their study. Since being a girl predicts being in the victim group, it can be a good start to prioritize the needs of this gender when creating a prevention program. First, determining whether the bullying behaviour they are exposed to is relational, direct, or indirect, may be important for the program to focus on. Identifying their own behaviours, thinking styles that contribute to the victimization of individuals benefit the determination of areas to be intervened.

Experiencing or witnessing any form of violent act during childhood is a risk factor for future aggressive act of the victim. For example, community violence exposure is related to enhanced number of aggressive acts (Gorman-Smith & Tolan, 1998). Physical abuse victimization in adolescence has immediate and enduring effects on the prevalence and frequency of a variety of self-reported offenses and intimate partner violence (Fagan, 2005). In this context, this research reveals that students who are neither bully nor victim have the lowest percentage of witnessing a violent event. Besides, witnessing a violent event predicted being in both bully and victim group. Due to lack of detailed description of the violent event in the questionnaire, it is not possible to comment on the mechanism of bully experience as a function of witnessing a violent act. On the other hand, as the above studies suggest, exposure to various violent events is related to the person's aggressive behavior. Based on this information, although we do not know the details in our own sample, the knowledge that the person has a violent experience can be taken into account while developing the intervention program. Understanding variables related to these students is important because the literature indicates that bully/victim group is a separate group of children that has its own characteristics (Juvonen & Graham, 2001). In a study, bully/victims who are both bully and are victim, become different from other groups by their high level of Machiavellianism and negative self-esteem and their low level of social acceptance (Andreou & Andreou, 2000). In addition, they may be subject to specific form of violence such as physical dating violence and emotional abuse in dating relationship (Holt & Espelage, 2007). Bully/victim group is also more vulnerable to be rejected socially and have tendency to show maladjustments in various domains (Schwartz, 2000).

Besides, these groups are not static in their nature: Hanish and Guerra (2004) indicate that approximately one fifth of aggressive victims who have behavioural characteristics similar to both only aggressive and only victimized groups became bullies after 2 years. In addition, this aggressive victims group had significantly higher scores in exposure to violence than stable ones. In this study bully/victim group resembles only bully group in its lower level of father attachment and moral maturity scores than neither bully nor victim group but showed a distinctive characteristic as having a violent event history as a predictor. Detecting the characteristics of this group is also important to predict and intervene with their group change and group participation behaviours.

As can be recalled from the introduction, the way the child attaches to the caregiver is associated with schemes of how he understands and uses moral values and this may have effects on attitudes in friendships. In this study, positive correlation between attachment to father and moral maturity was found and neither bully nor victim group’s attachment quality is better than victim group’s and both bully and victim group’s. Kokkinos' (2013) shows that anxious type of attachment scores increased in both bully and victims. Aggressive victims reported less perceived parental warmth than other groups and victims have emphasized the overprotection of the parent. Kõiv’s (2012) results indicate the lower secure scores of the victims and higher avoidant attachment of bullies that is in accordance with this study. Walden and Beran (2010) also indicated the increased parental attachment quality of uninvolved group than bully victim and victim groups. Some studies emphasize the relationship between attachment to father pattern and bullying behaviour. While females who have anxious bonding with their fathers are more likely to be aggressive physically, males who show anxious attachment with their fathers have higher levels of relational aggression (Williams & Kennedy, 2012). If attachment to father is high, males involve less in in-person bullying (Fang, 2018). In addition, attachment to father related to victimization directly (Innamorati et al., 2018). Although secure attachment to parents, significantly and negatively related to bullying, attachment to father sometimes predicts adolescent aggression (Gallarin & Alonso-Arbiol, 2012) and sometimes not (Maximo Sabrina, & Loy, 2014).

As indicated, parental attachment in general or attachment to father may have direct and indirect effect on adolescents’ behaviours: secure attachment has indirect effect on bullying by reducing aggressive attitudes (Eliot & Cornell, 2009). Nikiforou, Georgiou, and Stavrinides (2013) indicated the negative correlation between trust to father and communication with him, and positively correlated with fathers’ estrangements for both victims and bullies. This study shows that attachment to father score was not the predictor for belonging to one of the bully experience and its possible indirect effects are not investigated here.

In this study, moral maturity score is a significant predictor and only bully group had the lowest moral maturity score, neither bully nor victim group had the highest. Only victim group’s score was between the bully and bully/victim groups. Mature moral behaviour requires evaluating the other peoples’ rights, psychological states and effects of social interactions etc. (Cooley, Elenbaas, & Killen, 2012). A lack of moral values and a lack of remorse (Perren & Gutzwiller-Helfenfinger, 2012) and immoral, disengaged behaviours (Menesini, Nocentini, & Camodeca, 2013) are the predictors of both traditional and cyber bullying. Morally disengaged thinking style partially mediates the relationship between aggressive behaviour and moral competence (Grundherr, Geisler, Stoiber, & Schäfer, 2017). Bullies show more morally disengaged reasoning than not-involved students did and bully-victims more frequently indicate that violating moral rules is right (Perren, Gutzmiller-Helfenfinger, Malti, & Hymel, 2012). The authors also support that victims produce more victim-oriented justifications such as empathy but fewer moral rules.

In summary, gender, age, moral maturity scores and witnessing a violent event contribute differently to the prediction of different bully experiences. This implies that being female is a vulnerability to be a victim and prevention programs must support students’ moral maturity. Also, to screen the students' past traumatic events and provide psychological support are other important implications.

Implications

The result of the study indicates that different experiences of bullying are seen in different age groups (i.e. victimization rate is highest in 9th grades). School mental health professionals should keep in mind that being bullied and/or victimized during development can manifest themselves in different ways at different stages of adolescent psychosocial development. In particular, it may be beneficial for younger age groups to follow up periodically in terms of personal, familial, and socio-cultural characteristics that make them vulnerable.

Based on the idea that the bullying phenomenon is a dynamic process, professionals should be aware that individuals who fall into the victim category at some stage of their psycho-social development may be likely to fall into the bully category in the coming years. Therefore, they may need to evaluate adolescents periodically in order to capture possible transformations in the bullying experience.

School environment and managerial policies can be updated according to the data obtained by monitoring the interaction contexts of different age groups in the school so that effective primary prevention of bullying can be possible.

According to this study, having no violent experience is high in neither bully nor victim group. Based on this data, it may be meaningful for professionals to investigate children's experiences of exposure to violence and to identify and support the victims of violence quickly in order to protect the mental health of adolescents and prevent bullying.

Although the predictive effect of attachment to the father was not verified, it was observed that those who did not experience bullying had stronger attachments to their fathers than those in the victim group and in both the bully and victim group. It can be suggested to take this finding into consideration in activities that develop the parent–adolescent relationship.

Activities supporting moral development can be integrated into the general education program, and included in skill development programs for the management of adolescent peer relationships.

Limitations

In this study, "witnessing to bullying", which has an important place in bullying experiences, could not be addressed. Addressing this issue with this dimension will serve to understand this phenomenon more fully. This study had limitations which are related to use self-report measures. By using these types of measures we assumed that the participants were honest and open in their answers. Besides, their answers might be subject to response bias etc. For some questions the response rate was changed due to skipped questions, although the number was low. Finally, selecting the population participating in the study by convenience sampling may be an obstacle to generalizing the results.

Conclusion

Understanding the bullying-related and predictive factors in adolescence is important to focus the targets of the prevention programs. While moral maturity predicted bullying negatively, age and past violent act exposure predicted it positively. Although attachment to father was not a predictor, it varied among study groups, with higher scores for uninvolved. Those who have never participated in bullying action have higher attachment scores and morale maturity.