This study investigated whether independent mobility, perceptions of the environment and distance from home to school were related to children's self-reported physical activity in three different contexts - outdoor play, structured ex/sport and active commuting to school. Boys reported playing and taking part in structured ex/sport more often than girls, but there was no gender difference in active commuting. This highlights the importance of active commuting as a source of physical activity for girls and that increasing opportunities for outdoor play and structured sport and exercise in girls may be warranted.
The results show that overall environmental correlates related differentially according to activity context and by gender. A higher frequency of outdoor play was related to higher scores for Social Norm ('there were other children around to play with' or 'you often see children playing on the street') for both boys and girls. This is supported by qualitative studies in which parent's reported that having children to play with related to an increased likelihood of their children playing outside [17, 40] and Jago et al.  reported that different friendship groups at school and home are key influences on the location and type of physical activity in which children engage.
Greater independent mobility (IM) was related to higher frequency of play for both boys and girls. This is consistent with other studies where greater freedom to play out unsupervised by an adult was linked to higher levels of children's outdoor play and that the dependence on a parent's availability to take children to play spaces was a significant barrier to their outdoor play [41, 42]. Wen et al.  in one of the few other quantitative studies that has investigated IM in relation to outdoor play reported that those who were 'mostly allowed to walk on their own', near where they lived were more than two and a half times (OR 2.56, CI 1.84-3.58) more likely to spend at least 30 minutes outside after school compared to those who were never allowed to walk on their own near where they lived. Wen et al.  also reported that more positive perceptions of neighbourhood safety (measured by a single item) were related to higher levels of outdoor play. This study does not fully support this finding as personal safety was not a significant predictor of outdoor play. However more positive perceptions of traffic safety (pollution, traffic density, crossing places) were related to higher frequency of girls play. Comparative data relating children's perceptions to play is limited but studies investigating parents perceptions show that higher density and speed of traffic inhibits their willingness to let their children play outdoors, particularly unsupervised [40, 41].
Higher levels of IM were also related to higher levels of active commuting but this was only significant for males. This is similar to other UK data where children who were rarely or never allowed to go outside without an adult to walk to school or for leisure were twice as likely to be driven to school . Similarly in Australia, children with lower levels of independent mobility to walk alone in their neighbourhood were less likely to walk to school . Thus, there is growing evidence that higher independent mobility is an important correlate of active commuting in children of this age . This may be because parents' direct involvement in active commuting with their child is less common at this age.
Consistent with other studies [20, 46] a longer route from home to school was significantly related to decreased likelihood of active commuting to school. For boys, perceived ease of accessibility to school and wider destinations remained significant in the model. This supports other studies which show that both objective and perceived measures of the environment relate to active commuting in children [46, 47]. Ease of access to a range of facilities was also a significant correlate of structured ex/sport, particularly for girls, which is consistent with other studies where access to facilities is related to participation in structured physical activity . Independent mobility in the local (Local-IM) and wider area (Area-IM) was also a significant correlate for structured ex/sport but only for boys. This may be because boys range further than girls so are able to access facilities for structured sport and exercise unsupervised .
The strengths of this study include the measurement of three distinct physical activity contexts and environmental correlates in the same study allowing the relative importance of the different correlates to be investigated. The weak but significant positive correlations between the three physical activity contexts supports the view that children's participation is often specific to a particular physical activity context and that this specificity needs to be considered in the design of interventions as changes to increase physical activity in one context may not necessarily transfer to another . The finding that different correlates generally relate to different physical activity contexts has been found in a small number of other studies [22, 48] but this study includes a wider, more robustly measured range of correlates and adjusts for powerful confounders in analyses. Although this study has measured specific activity contexts in line with the recommendations by, for example Giles-Corti et al. , future work should measure both context specific behaviours alongside context specific correlates. Some progress in this area has been made with active commuting where studies have included some measures specifically tailored to active commuting in children [34, 42, 46] but there are few measures available to investigate the specific correlates of children's outdoor play.
The finding here that independent mobility was the only correlate related to all three physical activity contexts adds to the recent literature on independent mobility and physical activity in children [25, 40, 42]. Further work investigating independent mobility is warranted as this has been in decline over recent decades . This cross-sectional study cannot determine the direction of relationship or causality. Longitudinal data is required to indicate whether high independent mobility is a precursor to or a consequence of higher physical activity levels, whether change in independent mobility is related to change in specific physical activity behaviours and whether independent mobility is temporally linked to other perceived and objective measures of the environment. Prezza and Pacilli  showed that greater independent mobility and associated play in public areas during childhood was related to a stronger sense of community, and less fear of crime. Temporal data are required to determine if independent mobility and children's consequent familiarity with their environment leads to more positive perceptions of their environment or whether positive perceptions are a precursor of children's desire and parent's willingness to afford greater independent mobility . Further work should also consider factors not included here, such as car ownership, as it may moderate the relationships between independent mobility, perceptions of the environment and participation in different physical activity contexts, particularly in more rural settings. Also due to the relatively small number of participants from minority ethnic groups in this sample, ethnicity was not included in the analysis. Further work in a more diverse sample is required to determine if participation in different physical activity contexts varies by ethnic group.
Longitudinal data are also important to determine how the relationship between independent mobility, environmental perceptions and participation in different activity contexts change over time. For example, Sener et al.  reported that participation in unstructured free play decreased with age whereas participation in structured out of home activities was higher in older (12 to 15 years) compared to younger children (5-11 years). Further work should also investigate both child and parental perceptions of the environment in relation to children's participation in different physical activity contexts as these may exert independent and interactive effects. Whilst parents are still important gatekeepers of children's physical activity opportunities children's perceptions may increasingly influence their physical activity behaviour as they age.