Martijn van Teffelen

Imagery-enhanced cognitive restructuring: Efficacy 85 4 Table 2 Continued Condition Baseline Stat. Dif. I-CR ( n = 31) CR ( n = 32) AC ( n = 20) F / H (p) M0 (S1 baseline) 1.11 (.51) .96 (.40) 1.02 (.39) .53 (.592) M4 (S2 baseline) a .81 (.52) .78 (.41) .75 (.41) STAXI-2T M0 (S1 baseline) 17.81 (4.76) 18.38 (2.95) 17.50 (3.49) .35 (.703) M4 (S2 baseline) a 17.00 (4.60) 16.09 (3.43) 15.95 (2.80) AQ-H M0 (S1 baseline) 19.65 (7.79) 20.63 (5.63) 21.35 (7.22) .39 (.677 M4 (S2 baseline) a 17.16 (6.87) 18.25 (6.58) 17.80 (6.30) FOA M0 (S1 baseline) 57.10 (11.01) 58.91 (8.78) 59.15 (11.54) .33 (.718) M4 (S2 baseline) a 54.13 (10.28) 56.88 (9.84) 58.40 (14.38) Note . a non-normally distributed. S1 = session 1; S2 = session 2. Aggressive tendencies and state anger To test the hypotheses that the efficacy of I-CR is larger compared to CR and AC in reducing aggressive tendencies (H2), anger VAS and state anger (POMS) (H3) three separate mixed regression models were run. Fixed effects are shown in Table 3. First, in the model with aggressive tendencies (H2) as the dependent variable the random effect of therapist was removed from the model due to non-convergence. Results revealed that I-CR ( b = -0.48, t = - 3.48, p = .001) and CR ( b = -0.35, t = -2.51, p = .014) both effectively reduced aggressive tendencies from pre- (M1) to post-intervention (M3) compared to AC. When the CR condition was selected as reference category, there was no significant difference between I-CR and CR at post-intervention (M3) ( B = -3.23, SE = 2.88, t = -1.12, p = .264). Second, in the model with anger VAS as the dependent variable (H3), the random effect of therapist was not significant ( p = .398). Findings demonstrated elevated anger VAS levels at baseline in both intervention conditions compared to AC. This finding deviates from our earlier analysis of baseline anger VAS scores. To correct for baseline differences, anger VAS levels at baseline were entered as a covariate (as opposed to earlier models). Moreover, results revealed that both I-CR ( b = -0.57, t = -4.05, p < .001) and CR ( b = -0.35, t = -2.50, p < .014) led to a greater reduction in anger VAS levels at post-intervention (M3) compared to AC. When the CR condition was selected as reference category, there was no significant difference between I-CR and CR at post-intervention (M3) ( b = -11.22, SE = 6.32, t = -1.78 , p = .080).