The stakes of choosing the best theory as a basis for violence prevention and offender rehabilitation are high.
However, no single theory of violence has ever been universally accepted by a majority of established
researchers. Psychiatry, psychology and sociology are each subdivided into different schools relying upon
different premises. All theories can produce empirical evidence for their validity, some of them stating
the opposite of each other. Calculating different models with multivariate logistic regression on a dataset of
N = 21,312 observations and ninety-two influences allowed a direct comparison of the performance of
operationalizations of some of the most important schools. The psychopathology model ranked as the best
model in terms of predicting violence right after the comprehensive interdisciplinary model. Next came the
rational choice and lifestyle model and third the differential association and learning theory model. Othermodels
namely the control theory model, the childhood-traumamodel and the social conflict and reaction model turned
out to have low sensitivities for predicting violence. Nevertheless, all models produced acceptable results in
predictions of a non-violent outcome.