Martijn van Teffelen

General discussion 161 7 Despite the interesting nature of the present findings, it must be stressed that in the present dissertation, effect sizes on aggressive behavior were small. That is, CBM-I had a small effect ( d = -.) on physical aggression; CR had a small effect of d = -.35 on aggressive inclinations (irrespective of the nature of aggression); and I-CR had a small effect of d = -.48 on aggressive inclinations. I-CR and CR had no effect on physical aggression, as measured by the voodoo doll task (VDT). One explanation for the physical aggression discrepancy between (I-)CR and CBM-I could be that our (I-)CR study included a sample with more women than men, whereas the CBM-I sample included more men than women. Women tend to express their aggression more socially than physically compared to men (Björkqvist et al., 1992). An alternative explanation could be that the people in the clinical sample of the CBM-I study had higher levels of physical aggression at baseline than people in the sample of the I-CR study and were therefore more susceptible for change. A third explanation could be that one session of (I-)CR is just not potent enough to elicit change in physical aggression. Although both interventions reduced aggression to a small but significant extent, the current dissertation underlines the importance of separating physical, verbal, and relational dimensions of aggression ( chapter 2 ). Two general limitations are worth mentioning. First, verbal, and relational aggressionwere not operationalized or analyzed consistently across our studies. That is, chapter 3 did not operationalize verbal and relational aggression, and chapters 4 and 6 did operationalize, but did not separately analyze verbal and relational aggression. Second, hostility dimensions were not measured consistently throughout this dissertation. For example, chapter 2 was solely based on self-report, whereas chapter 6 measured hostile intent using vignettes and word-sentence associations, anger using self-report and aggression using both self-report and behaviorally using the VDT. In general, we found no differences within studies when multiple measures were used per construct. However, chapters 4 and 6 showed that both interventions differentially impacted state and trait levels of hostile intent and aggression. Specifically, CBM-I impacted hostility traits, whereas (I-)CR impacted hostility states. One explanation for this divergence is a time effect. CBM-I consisted of eight 30-minute training sessions (i.e., four hours), while (I-)CR lasted 90 minutes. Hopefully, the present work stimulates future research to investigate the effects of treatment on aggressive behavior beyond physical aggression. Next to the importance of precise hostility measurement, the findings in this dissertation indicate that it may be beneficial to personalize interventions. The personalized medicine account holds that a patient’s unique characteristics play an important role in tailoring their therapies (Hamburg & Collins, 2010; Ozomaro et al., 2013). Following this line of reasoning, mapping a patient’s characteristics that are relevant to hostility may provide an avenue for tailored, more effective treatment. That is, chapters 2 and 3 revealed that it is important to differentiate hostility into cognitive, affective, and behavioral dimensions and that these dimensions are differentially impacted by psychopathic and narcissistic personality traits under