Martijn van Teffelen

Chapter 6 144 study does converge with findings of a recent meta-analysis across 85 studies showing that CBM-I for anxiety ( SMD = -0.30) and depression ( SMD = -0.26) demonstrates small effects on symptoms compared to a sham condition (Fodor et al., 2020). Importantly, this meta-analysis showed that studies with lower methodological quality and therefore higher risk of bias generally found higher effect sizes. We believe that the present work fits within its conclusions, given that our effect sizes on aggression in Study 2 are small, while this study fulfills criteria for high methodological quality and low risk of bias (e.g., intention to treat analysis, random sequence generation, allocation concealment, blinding). Still, we observed no significant efficacy of CBM-I on hostile bias in Study 1, which most probably can be attributed to low power. Next to an observed efficacy on bias, Study 2 revealed secondary efficacy of CBM-I on behavioral aggression. The finding that CBM-I reduces behavioral hostility suggests an information processing pathway towards aggressive behavior that operates distinctly from non-behavioral aspects of hostility. This is in line with the Social Information Processing model of aggression (Crick & Dodge, 1994), which implies that the hostile interpretation of external (i.e., situational) and internal (e.g., emotional) cues results into a narrowing of potential behavioral response patterns. This, in turn, increases the likelihood of aggressive response patterns. CBM-I for hostility thus seems to provide people with an interpretational strategy (Clifton et al., 2016) that ‘re-opens’ previously suppressed but more adaptive behavioral response patterns, while leaving the more temperamental, emotional cue generation processes intact. Importantly, changes in aggressive behavior were observed using a well-validated behavioral measure, but not self-reported measures. On the one hand, this suggests that changes may occur outside of participants’ awareness. On the other hand, the self-report/ behavior discrepancy may reflect measurement problems that are related to aggression, such as paradigm flexibility (Elson et al., 2014; Hyatt et al., 2019), social desirability, and construct identity confusion (van Teffelen et al., 2020). The explorative finding that reductions in aggressive behavior were related to increased working alliance levels, but that working alliance levels in subsequent psychotherapy did not differ significantly between conditions could be explained by a non-differential motivation effect. That is, ‘doing something’ while waiting for therapy and experiencing a change in one’s aggressive behavior may facilitate beliefs about one’s ability to change. In line with this, experimental research shows that when people are led to belief that their emotions are highly controllable, this facilitates the regulation of subsequent emotional responses (Bigman et al., 2016). Notwithstanding, statistical power in the present explorative sample is simply too small to conclude that there is no true effect of CBM-I on the perceived quality of the working alliance in subsequent psychotherapy. We cannot rule out that CBM-I may have augmenting effects when provided prior to treatment, but this issue is largely neglected in the field and requires further experimental evidence.