Dorien Bangma

48 | CHAPTER 3 age was also a significant predictor of the IBQ, explaining an additional 3.4% of variance. Further investigation of the three IBQ components indicates that the relation of age is particularly evident for the affective component. Again, gender explained a large proportion of the variance (i.e., 25.7%; b = -3.27 [-4.70, -1.85], p < .001) of the IBQ affective component, but age was an additional significant predictor of this component and explained 6.3% of the variance. Both the internal as well as external validation analyses support the effects of age on the IBQ (Table 3.2). The cognitive and situational components of the IBQ were not significantly influenced by age. Regarding the TDT, no relation was found with any of the control variables, but 5.8% of the variance was explained by age (Table 3.2). The internal validation analyses support this finding, but these age effects lack external validity since these results could not be replicated in sample 2 (Table 3.2). Mediating effect of cognition After controlling for gender ( β = -0.15, p = .018) and years of education ( β = 0.15, p = .015), standard measures of cognition explained additional and significant variance of the performance on the CDR ( ΔR 2 = .40, ΔF (17, 142) = 8.77, p < .001). Numeracy (Arithmetic) was the strongest predictor of the performance on the CDR ( β = 0.27, p < .001), followed by planning (TOL) and working memory (Digit Span backward) ( β = 0.19, p = .003 and β = 0.16, p = .023, respectively). The remaining twelve regression models focused on the effects of cognition on each of the FDM tests did not reach statistical significance (.056 ≤ p ≤ .915). For numeracy (WAIS-IV Arithmetic), planning (TOL) and working memory (WAIS-IV Digit Span backward) the relation with age was investigated. Age was significantly negatively related to planning (TOL; ΔR 2 = .14, β = -0.49, p < .001) and numeracy (WAIS-IV Arithmetic; ΔR 2 = .05, β = -0.30, p = .002), whereas no significant relation was found between age and working memory (WAIS-IV Digit Span backward) ( β = -0.15, p = .120). A mediation analysis was therefore performed for the relation between age and the CDR controlling for TOL and WAIS-IV Arithmetic. The results indicate that, after controlling for these measures of cognition, the strength of the relation between age and the CDR decreases but remains significant ( β = -0.30, p < .001 after controlling for TOL and WAIS-IV Arithmetic compared to β = -0.50, p < .001 in the original analysis). According to the Sobel test, the decrease in the association between age and CDR can be explained by a mediating effect of both numeracy (WAIS-IV Arithmetic, p = .003) and planning (TOL, p = .012). Related to these findings, only a few weak correlations between standard neuropsychological measures and FDM tasks were found (see Appendix 3-2). Correlation analyses performed to determine the coherence between different FDM tests showed also only weak to negligible correlations (see Appendix 3-3).