Martijn van Teffelen

Chapter 7 156 “When people annoy me, I may tell them what I think of them” (measuring both anger and verbal aggression), or “Given enough provocation, I may hit another person” (measuring both hostile intent and physical aggression). A search on Web of Science on 28 August 2020 reveals that the AQ is cited 2995 times since its initial construction. This shows that currently the attitude in the hostility field regarding self-report instrument is too permissive. A premise to the theoretical understanding, prevention and treatment of hostility begins with accurate measurement. When the validity of hostility items is already reduced at face level, we should not be surprised that there is a lack of consensus on the structure of hostility in psychometric studies – reflecting the garbage in, garbage out concept. Consequently, the existence of cross-capturing items in hostility measures reduces their construct and face validity. In chapter 2 we examined whether a hierarchical model structure can account for differences reported by psychometric studies on the number of underlying hostility dimensions. We specifically selected instruments from the literature that included as little cross-capturing items as possible. Similar to constructs such as ‘narcissism’ and ‘agreeableness’, we predicted that hostility can be expressed as a construct with a hierarchical structure or, in other words, as a construct that can be interpreted at different levels of specificity. The findings of chapter 2 confirmed this hypothesis. Specifically, hierarchical factor analysis revealed five specificity levels of hostility. At the highest specificity level (most abstract), hostility can be expressed as one dimension characterized by a low threshold to experience angry affective states and react harmfully upon them (i.e., physically, and verbally). At the lowest level (most specific), hostility is characterized by one cognitive, one affective and three behavioral dimensions (i.e., social, verbal, and physical aggressive behavior). These dimensions align with the content of the solutions that were reported in previous psychometric studies. Theoretically our findings are in line with work by Averill (1983), who demonstrated already almost four decades ago that angry affective states and aggressive behavior can occur independently of each other in everyday situations. In other words, a person may experience angry affect, but this does not necessarily result in the expression of aggressive behavior. Vice versa, a person may behave aggressively without experiencing angry affect. The present work adds that the same goes for the cognitive dimension of hostility, and similarly, that a person can be prone to express aggression either socially, verbally, or physically. In other words, interpreting a situation in a hostile way does not necessarily result in angry affect or aggressive behavior. Similarly, being verbally aggressive can exist in isolation, without necessarily resulting in physical aggression. The finding that hostile affect, behavior, and cognition can occur independently from each other implies that these may have different antecedents and consequences. Perhaps, hostile affect (i.e., anger) is influenced more by internal psychophysiological changes (e.g., blood pressure or cardiovascular reactivity) (Zawadzki et al., 2017), whereas hostile behavior (i.e., aggression) is more influenced by