Predictors of aggressive behavior

A multiple factor approach was used to test additive and multiplicative models as well as to isolate a best predictive model of physical aggression. The variables of aggressive learning history, provocation, sex of target, sex of subject, sex‐role orientation, and aggressive tendencies were selected. Eighty‐three males and 117 females participated in the experimental session. Multiple regression analyses indicated that multiple predictor models were able to account for significantly more variance than were single predictor models; however, multiplicative models were unable to increase predictive efficacy. A model composed of sex of target, masculinity, and aggressive tendencies was established as the best predictive model for unprovoked aggression; provocation, masculinity, and aggressive tendencies made up the best predictive model of provoked aggression. © 1992 Wiley‐Liss, Inc.