We examined teacher characteristics predicting likelihood of intervening in indirect bullying (N=55) and piloted a 45-minute educational presentation about its mental health impact. Teachers’ global empathy and perceived seriousness of indirect bullying vignettes were predictors of their likelihood of intervening, but knowledge of mental health impact was not. The presentation increased knowledge of impact and perceived seriousness, immediately and seven weeks later, compared with a treated control group. However, empathy for victims and likelihood of intervening did not increase. Recommendations include adding skills-based material, a self-efficacy measure and a focus on increasing global empathy rather than specific empathy for victims.

This paper reports the partitioning of variance in scale scores from the use of three classroom environment instruments. Data sets from the administration of the What Is Happening In this Class (WIHIC) to 4,146 students, the Questionnaire on Teacher Interaction (QTI) to 2,167 students and the Catholic School Classroom Environment Questionnaire (CSCEQ) to 2,211 students in Australian schools were analysed using MLwiN. Variance components models with three levels (viz. student, class and school) for each of the 22 scales were created. Results revealed that statistically significant proportions of variance in all scale scores could be attributed to the student and class levels. Apart from one scale, no school level variance in scale scores was statistically significant. Typically, variance proportions were above 75% at the student level, above 15% at the class level, and below 5% at the school level. Multilevel analyses should be conducted with classroom environment data.