Research on aggression in educational settings has produced numerous factors and complex interrelated effect patterns. Results of research have been used to establish emotion-regulation-programs, like anti-aggression-trainings or other counseling activities. However, many of such activities are intended to be applied long time after problems with aggressive behavior have occurred. In order to be able to prevent aggression in schools, assessment tools are necessary that could help to assist educational decision makers and teachers in prognosticating negative emotional outbursts. In medical or criminal diagnosis, computer-based Bayesian probability networks are used to establish such a decision support. Within such networks, the relationships between different variables can be used to predict other variables based on conditional probabilities. Especially in respect to other statistical methods like, for example multiple regression procedures, these networks have several advantages: a) they allow to integrate different sources of information (existing data sets, results of reviews or meta-analysis, or expert estimates); b) they can be used for prediction (given independent variables and computing dependent ones), but at the same time also for diagnosis (given dependent variables and finding values of independent ones); or c) they allow to model complex mechanism that would overcharge researchers and experts when using traditional statistical methods in a relatively simple and practicable manner. Because of these advantages, it is not surprising that Bayesian Networks are used in medical diagnosis, in clinical decision support, in crime risk factors analysis, in forensic science, in student modeling, and many other fields.  However, in order that decision support is valid, these systems must be developed in several steps, using certain procedures, and focusing on research-based criteria. It is the aim of this paper to show how such networks can be designed and used for assisting aggression risk assessment and management in schools. Theoretical frameworks concern integrated theories on aggression (prevention) and models on computer-based decision support.

Boxer, P., & Dubow, E. F. (2002). A social-cognitive information-processing model for school-based aggression reduction and prevention programs: Issues for research and practice. Applied & Preventive Psychology, 10, 177-192. Jensen, F.V., & Nielsen, T. D. (2007). Bayesian networks and decision graphs (2nd ed.). New York, NY: Springer. Merrell, K. W., Gueldner, B. A., Ross, S. W., & Isava, D. M. (2008). How effective are school bullying intervention programs? A meta-analysis of intervention research. School Psychology Quarterly, 23, 26-42. Peterson, R. A. & Brown, S. P. (2005). On the use of beta coefficients in meta-analysis. Journal of Applied Psychology, 90, 175-181. Pourret, O., Naim, P., & Marcot, B. (Eds.). (2008). Bayesian Networks. A practical guide to applications. Chichester: Wiley.