An Empirical Classification of Children Involved in Bullying and Bully Victimization Using Latent Class Analysis

Background and Purpose: Aggression among children in elementary school is frequently characterized by bullying behavior. Research indicates that approximately 25% of students bully other students or experience victimization during elementary school. Bullying represents a complex set of behaviors and social interactions among children who are involved, and uninvolved, in bullying and bully victimization. Physical bullying is characterized as hitting, kicking, shoving, and intimidation techniques. Relational bullying involves indirect forms of aggression, including gossiping, rumor spreading, and exclusionary behaviors. Longitudinal studies of bullying and aggression indicate that children who bully other students and/or are victimized by their peers are at an increased risk for subsequent mental health problems, delinquency, peer rejection, depression, low self-esteem, and anxiety. While previous research has suggested that bullying involves four distinct groups of children – bullies, victims, bully-victims, and bystanders – this classification system, to date, has not been empirically examined or supported.

Methods: The current investigation applied latent class analysis (LCA) to a sample of 671 (52% female, 59% Latino/a; mean age = 11 yrs) largely minority, low-income fifth grade students attending urban public elementary schools. Using wave four data from a group-randomized trial of a bullying prevention intervention in 28 schools, LCA provided an empirical classification of students based on reported rates of involvement in bulling and bully victimization as measured by the Olweus Bullying Questionnaire. LCA assumes that the relationship among the bullying indicator variables can be explained by a finite number of latent categorical variables. Analysis of variance (ANOVA) tests were then conducted to describe the nature of the classes by examining class differences on a number of covariates including level of school commitment, antisocial peer influence, antisocial attitudes, perspective-taking empathy, delinquency, and depression.

Results: LCA results indicated that a three class solution best fit the data based on several fit indices including the Bayesian Information Criterion (BIC). Findings revealed distinct victim, bully-victim, and bystander classes. Interestingly, a separate bully class was not identified. ANOVA results indicated significant group differences between the victim, bully-victim, and bystander groups on levels of school commitment, antisocial peer influence, antisocial attitudes, perspective-taking empathy, delinquency, and depression.

Conclusions and Implications: The findings of this investigation are somewhat contrary to prior research suggesting that children often fall into four groups in the larger context of bullying and victimization (bully, bully-victim, victim, and bystanders). Our results suggest that children who bully other students also experience some level of bully victimization during elementary school. This finding challenges the common assumption that childhood bullies are somehow a separate and influential group in school settings. Similar to other studies, our results do indicate that levels of school commitment, antisocial peer influence, antisocial attitudes, perspective-taking empathy, delinquency, and depression differentiate bully-victims from victims and from those uninvolved in bullying and victimization. Implications for preventing bullying and victimization are delineated.