Teun Remmers

42 | Chapter 3 gender and impulsivity), all three pairs of two-way interaction terms, and the three-way interaction term. We tested this with all types of PA. Results showed statistically significant three-way interaction terms for active ( p = 0.002), light, ( p = 0.10), MVPA ( p = 0.001), and PA bouts ( p = 0.002). Figure 2 gives a graphical presentation of this interaction. Consequently, we stratified our analyses for gender and impulsivity (Table 3). Table 2. The association between physical activity enjoyment and physical activity behavior for boys and girls. % Active % Light % MVPA # PA Bouts Boys Crude 0.81 (-1.44 to 3.07) 0.66 (-0.86 to 2.19) 0.14 (-1.17 to 1.46) -0.14 (-0.49 to 0.22) Adjusted ‡ 0.69 (-1.60 to 2.97) 0.90 (-0.59 to 2.39) -0.21 (-1.59 to 1.16) -0.25 (-0.61 to 0.11) Girls Crude 2.21 (0.37 to 4.05) 1.51 (0.13 to 2.89) 0.70 (-0.39 to 1.79) 0.15 (-0.20 to 0.50) Adjusted ‡ 1.45 (-0.31 to 3.20) 1.06 (-0.28 to 2.40) 0.39 (-0.75 to 1.52) 0.10 (-0.25 to 0.45) Values in the table are regression coefficients (with 95% confidence intervals) from linear regression analyses with physical activity (PA) enjoyment (dichotomized over the median) as independent variable and physical activity levels (% time spent per day) as dependent variable, ‡ adjusted for child’s age, recruitment group, maternal education, season, child’s BMI, and impulsivity. Bold numbers represent statistical significance ( p <0.05). In boys, associations between PA enjoyment and PA behavior were stronger with relatively high impulsivity. For example, in boys who scored high on impulsivity, PA enjoyment had a stronger positive relation with percentage of time spent in MVPA than in boys who scored low on impulsivity (regression coefficient 1.30 versus -1.85). This difference is represented by the statistically significant interaction term of 2.84 (0.27 to 5.41). By contrast, in girls, associations between PA enjoyment and PA behavior were stronger with relatively low impulsivity. For example, in girls who scored low on impulsivity, PA enjoyment had a stronger positive relation with percentage of time spent in MVPA than in girls who scored low on impulsivity (regression coefficient -0.83 versus 1.19), respectively. This difference is represented by the interaction term of -2.00 (-4.29 to 0.29). The average percentage explained variance (R-square) of all models was 20.8% (SD = 0.7%) with a minimum of 6.8% (regarding number of PA bouts in boys with high impulsivity), and a maximum of 31.2% (regarding active behavior in boys and girls with relatively low impulsivity). No major differences in explained variance were found between PA intensities and gender.