46 Chapter 2 3.2 Intervention effects on substance use frequency, substance use severity and binge drinking Table 2 and Figure 2 present the intervention effects (bivariate correlations are shown in Appendix A). Visual inspection of model diagnostics plots reveals good model fit for all models without violations of statistical assumptions. The results showed a stronger decrease in substance use frequency in the intervention condition compared to the control condition, as the interaction time × condition was significant, F(1, 50.43) = 9.27, p = 0.004. Similarly, a stronger decrease in binge drinking was found for adolescents in the intervention condition compared to those in the control condition, F(1, 48.02) = 8.63, p = 0.005. For substance use severity, the interaction time × condition was not significant, F(1, 42.09) = 2.20, p = 0.145, indicating no differences between conditions over time on adolescents’ most severely used substance at baseline. Intervention effects were thus found for substance use frequency and binge drinking, but not for substance use severity. 3.3 Intervention effects per substance separately Table 2 presents effects of frequency and severity on individual substances. These additional analyses on separate substances reveal, in addition to intervention effects on alcohol and cannabis frequency, a stronger decrease in the severity of alcohol use in the intervention group compared to the control group, F(1, 48.26) = 5.37, p = 0.025. 3.4 Power and effect size Results should be seen in the light of our sample size (n = 66), that was smaller than intended (Schijven et al., 2015). Nevertheless, post-hoc power analyses using 100 Monte Carlo simulations revealed a 90, 45 and 85% chance of finding a statistically significant effect (α = 0.05) for the interaction time × condition in models for, respectively, substance use frequency, substance use severity and binge drinking. Marginal R2 was 0.16, 0.13, 0.17 for the combined fixed effects in models for, respectively, substance use frequency, substance use severity and binge drinking, reflecting the medium effect sizes (Cohen, 1988) we aimed for with our a priori power analysis.
RkJQdWJsaXNoZXIy MTk4NDMw