Classroom-Level Differences in Child-Level Bullying Experiences: An Application of Multilevel Latent Class Analysis

Introduction: Research documents the importance of classroom norms or classroom climate in either reducing or increasing bullying among students. Importantly, classrooms where the composition of bully and victim subgroups favors those engaged in bullying may contribute to a higher likelihood of peer contagion. In other words, classrooms with a high number of bullies may create a classroom climate where youth become more tolerant of bullying and may gravitate toward those who bully as they hold positions of power and status. However, no study to date to our knowledge has examined the composition of bully / victim subgroups at the classroom level.

To that end, we examined whether children's experiences of bullying and victimization systematically differ at the classroom level. Conceptualizing classroom-level differences as a manifestation of classroom climate, we use a multilevel variant of latent class analyses (MLLCA) to characterize the composition of classrooms vis-à-vis children's reported experiences of verbal, relational, and physical victimization and bullying.

Methods: A total of 644 students in 3^rd – 5^th grades participated in the study from six elementary schools (51.3% female; M age = 9.3, SD = 1.2). Students completed measures of the frequency of involvement in bullying and peer victimization. Using MLLCA, a child-level model of bullying experiences was estimated by a series of standard single-level LCA models. Second, we tested the hypothesis that the identified child-level bullying classes coalesce at the classroom level. Decisions about model selection were governed by a combination of statistical and substantive criteria.

Results: Five distinct child-level victimization-bullying classes were identified: (a) few or no experiences of victimization or bullying, (b) moderate levels of victimization and bullying, (c) high levels of victimization and low levels of bullying, (d) low levels of victimization and high levels of bullying, and (e) high levels of victimization and bullying. Further, at the classroom-level, two distinct mixtures of child-level classes were identified. These classroom-level mixtures differ with respect to the proportions of children with few or no experiences of victimization or bullying and children who reported high levels of victimization. Finally, the proportion of classroom-level mixtures was found to differ significantly across sampled schools, suggesting that classroom-bullying climate may be partly a function of school-level phenomena. 

Conclusions: Classroom level differences in the composition of bully / victim subgroups as well as the differences in the proportion of classroom-level mixtures across schools point to the need for unique prevention and intervention approaches.