Anne Fleur Kortekaas-Rijlaarsdam

CHAPTER 5 104 Generalization of learning Children who did not meet the predefined criteria for successful acquisition of the stimulus-response relationships were identified and their distribution across groups (ADHD versus TD) and treatments (placebo versus MPH) was assessed using χ 2 testing. Effects of group and treatment were tested using GLMM with group and treatment as fixed factors. Reversal learning GLMMs with a random intercept were used to test the effects of group (ADHD using placebo versus TD) and treatment (MPH versus placebo), entering stimulus pair and either group or treatment as fixed factors. Medication condition sequence For all dependent variables medication condition sequence effects (comparing children who were treated with placebo first to children who were treated with methylphenidate first) were investigated using GLMM. R E SU LT S One case was excluded from our ADHD group in the MPH condition due to missing data and five cases were excluded in our TD group due to missing data (for the TD group the learning task was dropped for these cases due to time restrictions), resulting in a total of 63 children with ADHD in the placebo condition; 62 children with ADHD in the MPH condition and 62 TD children. Demographics of our ADHD and TD group are shown in Table 5.1. The ADHD and TD groups did not differ in age, but estimated IQ was lower in the ADHD group. IQ did not significantly correlate with our dependent variables (all r <.17, all p>.05) and was therefore not included as a covariate in the analyses. Learning task performances are shown in Table 5.2. Learning performance was not affected by mediation sequence (all p s>.13), thus ruling out carry-over effects. Learning Task Performance Effect of group on learning task performance The residuals of learning task performance were not normally distributed and therefore, a square root transformation was applied to the data resulting in normal distribution of the residuals. There was a main effect of stimulus pair ( F (246)=22.3, p <.001). Post hoc analyses with Bonferroni correction indicated that learning task performance for