A Taxonomy of Anger-Related Behaviors in Young Adults

In 2 studies, we investigated the occurrence of anger-related behaviors and their relationship to emotional, performance-related, and situational variables. In the first study, we constructed a comprehensive taxonomy of behaviors associated with anger, and we examined the occurrence of the resulting behavior categories as a function of several independent variables. A total of 8 distinct behavior categories were identified, 3 aggressive and 5 nonaggressive. Our results also demonstrated that fight (including both verbal and physical aggression) and flight behaviors occurred most frequently. Physical aggression, however, occurred most frequently in an inhibited form, in response to the emotion of anger (as compared to the emotion of irritation), and when the anger was intense. A second study was conducted to replicate and extend the findings of Study 1. The results suggest that the taxonomy, as derived in Study 1, is comprehensive and allows for a reliable categorization. Moreover, it appeared that fight and flight behaviors occurred most frequently if the target at whom one is angry was present. The following reduction rules were used: (a) responses that differed only in the order of the words were put together; (b) singular and plural items were joined; (c) responses that differed only in verbs such as “to be,” “to become” or “to do” were grouped together; (d) responses that differed only syntactically (verb, noun or adjective) were grouped together; (e) responses that differed only with respect to an article, a preposition, or a possessive word were grouped together; (f) and a few specific rules were used (e.g., yourself = myself; self = auto). The complete coding manual is available upon request from the first author. This number differed across proportions because it depended on the number of participants assigned to each of the experimental questionnaires. Our initial analyses contained a participant-specific random intercept, so as to take into account individual differences in the propensity to use each of the behavior categories. Inclusion of a random intercept appeared to yield a significant improvement in model fit for two of the eight behavior categories. However, because analyses with a random and with a fixed intercept yielded similar conclusions, we decided to select the most parsimonious model, namely the one without a random intercept. The total number of participant–performance–instrumentality combinations equals 504 (i.e., 84×6) because all 84 participants had to answer six questions.