Teun Remmers

Investigating moderation between determinants of children's outside play | 71 Walkability Scale but was modified to reflect the Dutch built environment, including items relevant to children (e.g. playgrounds, school yards, and dog waste) (37). A full description and reliability statistics of all scales can be found in Appendix 2, Table S1. Response scales were constructed according to level of agreement with statements like “Most streets in the neighborhood have cycle paths” or “The neighborhood is a real community”, and consist of five-point scales ranging from “I strongly disagree” to “I strongly agree”. Composite variables were created from individual items by the inspection of Cronbach’s alpha (between 0.70 and 0.80) and principal component analyses for potential constructs with a Cronbach’s alpha of at least 0.60 (38). The following constructs of the perceived physical environment were identified: accessibility (6-item sum score), functionality (6- item sum score), attractiveness (7-item sum score), satisfaction (3-item sum score), and traffic safety (4-item sum score). The use of each of these constructs is recognized in studies investigating the supportiveness of the environment for PA (11, 39, 40). Social environment Factors of the social environment were expressed as 1) parenting influences and 2) social capital. Parenting influences were assessed by a set of nine questions, based on the Dutch translation of the Child Feeding Questionnaire (CFQ) (41). The nutrition-related items of the CFQ were translated into a PA-related parenting questionnaire (42). Parenting influences were defined as the 1) perceived influence of the respondent and their partner on the PA behavior of their child (both 3-item sum scores), 2) attitude towards child PA (5- item sum score), 3) perceived responsibility regarding child PA (2-item sum score), 4) concern regarding child PA (3-item sum score), 5) restriction of child screen time (6-item sum score), 6) pressure towards child to be active (3-item sum score), and 7) monitoring of child PA (2-item sum score). Average reliability (Cronbach’s alpha) of these scales was 0.67, with a range of 0.57 – 0.93. A complete description and reliability statistic of these scales is presented in Appendix 2, Table S1. Social capital was assessed by a set of five items, based on earlier empirical research on the influence of social capital on obesity and PA (43). Cronbach’s alpha for this scale was 0.87 (see Appendix 2, Table S1). Outside play Outside play was defined as the total duration of unstructured outside play in an average week, without organized sports, school physical education, and active transport. Outside play was assessed by questionnaire, both at child age five and seven years using an identical set of questions. First, parents were asked on how many days their child played outside in an average week for the last four weeks. The eight response categories ranged from “never or less than one day on average” till “seven days per week”. Second, parents were asked to indicate the average duration of outside play. The five response categories were: shorter than half an hour (computed as 15 minutes), half to one hour (computed as 45 minutes), one to two hours (computed as 90 minutes), two to three hours (computed as 150 minutes) and three hours or more (computed as 210 minutes). The frequency and duration of child outside play were multiplied to arrive at an average minutes of outside play per week. The date of completing the questionnaire was used to classify the season (i.e. winter, autumn, spring, and summer).