Teun Remmers

Daily weather and children's PA patterns | 97 as preliminary results indicated that children were most active in schools with policies to provide opportunities for active indoor activities during wet weather (14). Given our longitudinal design across four school terms, substantial variations in day length were recorded in our study; however, no associations between day length and PA were observed. Unlike the study of Goodman (12), we entered day length as a continuous variable in our models. When we computed tertiles based on equal frequency distributions however, some small differences in MPA were found but results were inconsistent (data not shown). Initial multi-collinearity statistics provided indications for potential multicollinearity between day length, temperature and solar radiation. Hence, we investigated this in sensitivity analyses. For example, removing temperature or solar radiation from our models led to significant decrease in model fit ( p < 0.05 for MPA and VPA), while the association between day length and MPA remained unchanged (data not shown). Based on these sensitivity analyses, we decided to retain day length in the models as we deemed multicollinearity not to have significantly influenced our results. However, in line with the present study, previous studies reported that associations between day length and PA were only strong during the late afternoon and early evening (12), and only small increases in overall daily PA were shown between days with large differences in sunset times (i.e. before 5 PM versus after 9 PM) (11). Consequently, day length may independently increase PA in late afternoons and evenings, but associations between day length and overall PA across the whole day seem to be weak (11). According to the results of Goodman et al., this may be due to an increase in light PA during late afternoon or early evenings (12). In addition, it is possible that day length might have less influence on PA performed as part of organized sports versus discretionary PA (e.g. sports facilities may be more likely to be lit than playgrounds). Future studies are encouraged to further examine these distinct PA patterns. Strengths and weaknesses To our knowledge this is the first study to use a longitudinal design to investigate the influence of weather elements on children’s intra-individual PA patterns. The repeated measurements distributed across four school terms within one year ensured variation in weather elements and are representative of yearly PA patterns among Australian children. The present study’s approach of PA patterns within individuals is different from earlier studies. These studies typically analyzed between-individual differences, aggregating PA over days. Within-individual analyses, such as in our study, are better able to characterize day-to-day variation in children, helping us to understand PA patterns in children. In our study, the lowest average temperature recorded was 10.7 degrees Celsius. Therefore, these findings may not be generalised to countries with lower winter temperatures. In addition, weather elements also show within-day variability (e.g. short periods of heavy rainfall may have a different influence on PA than prolonged light rainfall) (12). Future studies are encouraged to also analyze within-day variability of weather elements in relation to PA patterns, particularly in regions where the weather is