Teun Remmers

94 | Chapter 6 Figure 1: Relationship between Temperature and Physical Activity. Mean Unstandardized Predicted Values of Moderate and Vigorous Physical Activity were adjusted for rainfall, humidity, day length, day type, gender, age, and BMI. For Moderate Physical Activity p-value temperature < 0.01; p-value temperature 2 < 0.01. For Vigorous Physical Activity p-value temperature = < 0.86; p-value temperature 2 < 0.01. Moderation across weather elements, and between socio-demographics and day type There was statistically significant moderation for solar radiation in the relationship between temperature and MPA and VPA (improvement of the model by including interaction-terms p < 0.01 for both MPA and VPA, likelihood ratio test with 2 df ). This means that the previously computed temperature that corresponded to highest PA, depends on the level of solar radiation. We dichotomized solar radiation to graphically demonstrate this. It should be noted that during days with low solar radiation, maximum temperatures only reached 29.1 degrees Celsius, while in days with high solar radiation, temperatures were recorded up to 39.1 degrees Celsius (see Figures SDC2 and SDC3 in Appendix 3). Consequently, temperature had a smaller influence on children’s MPA and VPA during days with lower solar radiation; while temperatures (above 25 degrees Celsius) were strongly associated with lower MPA levels during days with high solar radiation (see Figure SDC2 and SDC3 in Appendix 3). This can also be deducted from the negative interaction-term solar*temperature, while the interaction-term of the quadratic term solar*temperature squared was positive (see Supplemental Table 4 in Appendix 3) In addition to the interaction described above, moderation was also found between day type and temperature in the relationship with VPA (see Supplemental Table 5 in Appendix 3; moderation of weather elements and day type in relationship between weather elements and vigorous PA). We found a strong curvilinear relationship between temperature and VPA during weekdays; while during weekends temperature was linearly related with lower VPA (see Figure SDC6 in Appendix 3). We found no indication for moderation by gender, age, and BMI.