Thesis (PH.D.) -- Syracuse University, 2004. / "Publication number AAT 3136458."

Despite research evidence that social context and personal characteristics are related to girls' violent behavior, little is known about the relative contribution of such antecedents. Therefore, the purpose of this study was to assess the relative strength of predictors of school violence among a sample of middle school girls. Of special interest were the intervening variables, because knowledge of their relative strength could enable schools to design targeted interventions to reduce school violence. Social learning theory formed the theoretical foundation for the study. A four-part survey consisting of sociodemographic items, the Rosenberg Self-Esteem Scale, an amended version of the Attitudes Toward Violence Scale, and the School Violence Inventory (used to assess engagement in offenses that could result in school suspension) was administered to 229 girls enrolled in a middle school in a southern U.S. state. Data were analyzed using hierarchical multiple regressions in which intervenable variables were entered first as a block, followed by nonintervenable variables. The results indicated that the predictors of school violence (from strongest to weakest) were observation of school violence, gang membership, favorable attitude toward violence, school suspension, grade level, and drug use. This finding suggests that female middle school students may be learning to behave violently by observing others engaged in such behavior at school and through the influence of gangs. Implications for positive social change are that the results could be used by educators and other school officials develop specific interventions that more effective target known predictors of school violence among middle school girls (for example, increased student monitoring, after-school programming, and guided classroom discussions on the nature of violence and its motivations).