Racial Tension and School-Based Violent Aggression and Victimization: A Multilevel Latent Profile Analysis

Introduction: Numerous studies have demonstrated that the psychological strain of racial discrimination can lead to increased risk of violent behavior among youth. However, few studies have examined whether racial discrimination can predict behavioral patterns that include both aggression and victimization, and even fewer have analyzed this relationship within a multilevel mixture regression framework. Using a multilevel latent profile model, we determined patterns of co-occurring aggression and victimization among a sample of urban middle school students. We then examined students’ perceptions of racial tension with demographic control variables as predictors of class membership.

Methods: We surveyed 4,025 middle school students from 34 Charlotte-Mecklenburg middle schools during Fall 2016. The survey included items from the Aggression and Victimization Scales for Middle School Children and the Inventories of School Climate for Students. We fit multilevel latent profile models with six indicators of students’ past-week aggression and victimization, including pushing or shoving, slapping or kicking, and threats of harm. Along with students’ perceptions of racial tension and cultural plurality in their schools, we added individual- and school-level demographic covariates, including gender, race and ethnicity, and schools’ socioeconomic profile. We obtained all predictor variable estimates using multivariate logistic regression with maximum-likelihood bias correction to account for classification error.

Results: Results revealed a three-class solution representing patterns of past-week aggression and victimization: Low Aggression/Low Victimization (86.5%), High Aggression/Low Victimization (7.3%), and Moderate Aggression/High Victimization (6.8%). Net of most control variables, perceptions of high levels of racial tension significantly predicted both High Aggression/Low Victimization (b=0.290, SE=0.065, p=0.000) and Moderate Aggression/High Victimization (b=0.327, SE=0.066, p=0.001) membership. Students who identified as Black or African American were more likely to be in the High Aggression/Low Victimization (b=1.025, SE=0.203, p=0.000) and Moderate Aggression/High Victimization (b=0.547, SE=0.231, p=0.018) classes, and girls were less likely than boys to be in the Moderate Aggression/High Victimization class (b=-0.776, SE=0.144, p=0.000). The typicality of each profile also varied significantly between schools (variance estimate=0.442, SE=0.142, p=0.002), with schools that had higher proportions of economically disadvantaged students showing higher likelihoods of having students in the High Aggression/Low Victimization (b=1.779, SE=0.409, p=0.000) and Moderate Aggression/High Victimization (b=2.021, SE=0.401, p=0.000) classes.

Conclusions: Our analysis suggests that students’ perceptions of racial tension can significantly predict patterns of co-occurring aggression and victimization in schools. However, this relationship does not exist in a vacuum. Surveillance of school-based violence should examine racial tension as one of several individual- and school-level stressors that may predict aggressive behaviors and likelihood of victimization.