The experience and role of teachers in school bullying incident identification and intervention has only just begun to be studied. The literature regarding the factors which influence a teacher’s intervention in school bullying have not focused on teachers’ understanding of bullying or the impact of administrators and other teachers on their intervention. In this study Latane and Darley’s (1970) framework for bystander intervention was applied to school bullying to better understand how teachers identify and intervene in school bullying incidents. An on-line survey of middle school teachers, which utilizes video scenarios, collected data specific to teacher accuracy in identifying bullying, intended responses to bullying incidents, administrator support, peer response, and self-efficacy. Quantitative methods were utilized in analysis.

Results indicate that 25.35% of teachers consistently identify bullying behavior across all five video scenarios. The assignment of grades to accuracy rates indicate that most teachers earn an A (25.35%) or a B (46.48%). Accuracy varied by type of bullying behavior, with teachers least likely to accurately identify social bullying. Training had a medium effect on accuracy (d = 0.331), as did years of teaching experience (d = 0.4505). Results indicate accuracy in identifying bullying behavior is correlated with direct intervention, r = 0.293, p ≤ .05. Administrator Support and Peer Response variables were calculated. Chi-square analysis indicates that peer response scores are associated with teacher self-efficacy, X2 (36, N=61) = 52.561, p = 0.037, and response-efficacy, X2 (21, N=37) = 44.412, p = 0.002. Additionally, years of teaching experience has a large effect on peer response (d = 0.762). Results of analysis of variance indicate that peer response does have a significant impact on a teacher’s direct intervention in incidents of school bullying F(8, 58) = 6.067, p=0.014. Administrator support also has a significant impact, F(14, 58) = 6.515, p=0.009. Finally, the effect of school building was also significant, F(3, 58) = 8.014, p=0.012. While the interaction between peer response and administrator support was significant, F(4, 58) = 5.610, p=0.024, other interactions between variables in the model were not.