Teun Remmers

Afterschool PA and the built environment using GPS, GIS and accelerometers | 131 Spatial analyses of multi-place environments We computed a 400-meter buffer surrounding a participant's school and residence. We also computed a 200-meter buffer surrounding the shortest path between participant's school and residence along the street network. Subsequently these three buffers were combined into an individualized multi-place environment using the dissolve function in ArcGIS. Relevant GIS-data from the municipality of 's-Hertogenbosch was extracted from these multi-place environments, using the spatial join and summarize functions in ArcGIS (see Figure 2). This GIS-data consisted of two levels of detail. In first detail-level, we identified the categories: vegetated terrain, water, buildings, and roads. Second level of detail consisted of road-categories (e.g. cycling path, rural road, highway, parking spots), and vegetated terrain appearances (e.g. woods, lawn, shrubs, agriculture, etc.). For each respondent, we first computed the total area (in square meters) of their multi-place environment, and the specific area (in square meters) that was assigned to each of the built-environment features (e.g. woods, roads). For each of the built-environment features, we computed the proportion of area assigned to that specific feature, relative to the total area. For example, a participant's exposure to bicycling paths was quantified as the area (in square meters) of bicycling paths within its individual buffer, relative to the total buffer-size of the participant's multi-place environment. Operationalization of the leisure time and active transport contexts Based on the GPS-derived context information and hierarchical decisions of Klinker et al. (2014), we first identified the domain ‘home’ by selecting records that were within 10 meters of each respondent's self-reported residential-parcel from all GPS points in the afterschool time-segment (Table 2). Second, we identified four other subdomains by identifying records within 10 meters from the school parcel, sports facilities, shopping centres or malls, or afterschool childcare. Third, for transport, we applied the above- described PALMS speed-thresholds. As bicycling and walking may be influenced by distinct environmental features, we investigated bicycling and walking separately. Fourth, all other non-defined records were identified as afterschool leisure time performed at other locations (e.g. at friend's homes, in parks, etc).