The Archives of Pediatrics & Adolescent Medicine have an interesting metaanalysis article available this month regarding the success of School-Based bullying interventions. The analysis found that the vast majority of anti-bullying programming was ineffective including classroom discussions, role playing, or detention.
There were positive results found when a more holistic model was involved that utilized a whole school approach with involvement from administration, teachers, and social workers (or counseling staff). Unfortunately such a large scale targeted model is expensive and may suggest the need for the utilization of a more appropriate and accurate method of identifying bullying incidents. I've used the Reynolds Bully Victimization Scales for Schools (Reynold's of APS, RADS & RASI fame) a few times. The tool is useful for both externalized and internalized aspects of bullying (but victimization and bullying behaviour) and could be a useful solution for targetting more accurately.
CLICK HERE to read a review of the Article
CLICK HERE to read the journal abstract
CLICK HERE for a link to Reynolds bio