Martijn van Teffelen

Chapter 4 76 for increasing the efficacy of CR is in line with the ideas of Beck (1985) that “hot” (i.e., affectively valenced) cognitions should be more modifiable than “cold” (i.e., affectively unvalenced) cognitions (Beck, 1985). In sum, enhancing CR for transdiagnostic hostility with mental imagery, may potentially lead to an increase of its ‘efficacy. The primary aim of the present study is to compare the efficacy of one session ‘imagery- enhanced CR’ (I-CR) for hostility with traditional CR and an active control (AC) condition in a sample of participants with increased hostility levels at pre- and post-intervention and at one-week follow-up. The main hypothesis is that I-CR is more efficacious than both traditional CR and the AC condition in primarily (H1) reducing the believability of hostile cognitions and secondarily (H2) reducing aggressive tendencies; (H3) state anger and (H4) hostility traits. Moreover, both interventions, I-CR and CR, are hypothesized to be more efficacious in reducing these variables than an AC condition. We differentiated between (T1) immediate intervention efficacy and sustained intervention efficacy at one-week follow-up (T2a) before and (T2b) after being ‘provoked’ by imaginarily re-exposing the participants to an idiosyncratic anger-provoking situation. METHODS Participants Participants were recruited using flyers on the campus of Maastricht University, the Netherlands, and from the university’s online participant database. We included participants between 18-60 years and with a raw score above .67 on the hostility scale of the Personality Inventory for DSM-5 (PID-5) (Van der Heijden et al., 2014). This cut-off equals the observed PID-5 mean in a Danish population (a comparable population to the Netherlands) (Bach et al., 2016). Sample size was determined a priori using the formula N 1 = N 2 = N 3 = (Z 1 - b + Z 1- α /2 ) 2 x (2 / d 2 ) / .90 with α = .05 and b = .20, anticipating a drop-out of 10%. We reasoned that we would need less participants in the AC condition, because the expected difference would be larger between the active conditions (I-CR and CR) and the control conditions. Based on the current literature, CR and I-CR were compared with an expected difference of d = .72 (McEvoy et al., 2015). CR and I-CR were compared with an AC condition with a minimally expected difference of d = .98 (Norton & Abbott, 2016). Hence, sample size was N 1 = N 2 = 34 in the CR and I-CR conditions and N 3 = 19 in the AC condition. The total final sample consisted of N = 87 participants. The CONSORT Flow diagram is presented in Figure 1. Sample characteristics are shown in Table 1. The groups did not significantly differ in age, gender, nationality, education level and work situation.