Teun Remmers

182 | Chapter 10 methodologies and generating stronger hypotheses for future research. Namely, chapter 3 generated stronger hypotheses for individual-level motivational determinants of PA, using complex conceptual models involving moderation mechanisms. In chapter 7, an alternative methodology was explored by testing the strength of environmental determinants on afterschool PA using various filters of time and place. In chapter 8, this time-place methodology was further improved by investigating relationships between multi-place environmental determinants and additional context of PA during afterschool hours. The studies in chapters 4, 5, 6 and 9 used longitudinal designs, which provided valuable insights into how environmental determinants may influence PA over time. Determinants of PA, especially environmental determinants, are typically relatively stable over time. The value of longitudinal observations may be even further increased in case of changes in the environment as a determinant of PA behaviour. For example, as current evidence is emerging on promising features of PA-supporting environments such as greenspace and street connectivity (22, 23), the effectiveness of these environmental changes on children's PA over time can be investigated using natural experiments (24-26). By doing so, causal relationships between these environmental attributes and PA may be revealed. Measurement of (context-specific) PA Previous studies have traditionally focused on investigating aggregated levels of total PA, performed daily or across several days. For example, chapters 2, 3 and 6 investigated such associations with daily objectively measured PA. Numerous studies indicated the importance of objective measurements for daily PA, as self-reported or parent-reported questionnaires tend to significantly overestimate PA levels (27, 28). In addition, objective measurements such as accelerometers are independent of recall- or social desirability bias (29). However, accelerometers have their own challenges and limitations, such as varying thresholds for classification of intensity-levels (30), general incorrect measurement of cycling and static upper-body activities (31), varying definitions of non-wear time (32), and relative representativeness of measurements with respect to total habitual PA (33). Nevertheless, the use of accelerometers is still the recommended methodology to measure daily PA in free-living conditions. In contrast to aggregated daily PA, accelerometers alone are unable to measure the context in which PA occurs (e.g. school from organized sports participation) (34-36). Contextual information is essential in understanding environment-behavior relationships, as environmental determinants, such as studied in chapters 7 and 8, have been shown to be context-specific (37). For example, accessibility to sports grounds may influence children's sports participation, but not their active transport to school. Therefore, previous studies have used self-reports or parent-reports to be able to focus on context- specific PA (38-40). In this thesis for example, chapters 4 and 5 investigated such associations with parent-reported child's outside play. Studies with clearly conceptualized assessments of the duration and frequency of specific PA-contexts and appropriate theoretical- and analytical frameworks, can still contribute to the evidence-base regarding determinants of specific PA-contexts.