Teun Remmers

40 | Chapter 3 Statistical analyses Characteristics of the population for analysis were evaluated with descriptive statistics. As PA behaviors were shown to differ considerably between boys and girls (39), we explored this with Chi-square tests and t -tests for categorical and continuous levels of measurements, respectively. To evaluate the crude association between PA enjoyment and PA behavior, we first modeled linear regression analyses, with several components of PA behavior (i.e. percentages of active, light, MVPA intensities, and number of PA bouts) as dependent variables. PA enjoyment (dichotomous) served as the independent variable. Second, we fitted linear regressions containing all main effects in the model (i.e. gender, age, BMI, and impulsivity) for each of the PA behaviors as dependent variables. Third, we inspected the interaction of each of these main effects separately. We checked linearity and homoscedasticity by plotting standardized residual versus standardized predicted residuals, and inspected the accompanying histogram. In all studied models, the assumptions were met. We tested for moderation by computing an interaction term between PA enjoyment and the potential moderators (i.e. age, BMI, and impulsivity) a priori for boys and girls separately. Based on these results, we stratified our final analyses based on statistically significant interaction terms ( p < 0.10). Results Participant characteristics The total sample consisted of 80 boys and 91 girls with a mean age of 9.17 (SD=0.63) years, as presented in Table 1. Average PA enjoyment was similar for boys and girls (PACES mean 4.19 versus 4.20). While time spent in light activity was similar between boys and girls (Table 1), boys showed a significantly higher percentage of time spent in MVPA than girls, amounting to a higher percentage of time spent active per day ( p <0.05). The average number of PA bouts per day was also higher in boys than girls ( p <0.05). Gender differences were small and not statistically significant for BMI and impulsivity (Table 1).