Teun Remmers

62 | Chapter 4 This study has demonstrated a negative association between parent-perceived difficulty in improving their child’s OP and OP at child age five and seven. This indicates that parents are able to indicate difficulties, and that the presence of these difficulties was indeed associated with relatively low levels of OP. Future studies should investigate these difficulties more thoroughly to identify what the exact difficulties are that parents are struggling with (e.g. time constraints, child’s friends to play with, etc.). Strengths and weaknesses A strength of the present study is its longitudinal design, including a relatively large sample. In addition, this study assessed PA through a whole year, subsequently adjusting for the effect of seasonality. The present study’s selective analytical approach resulted in sufficient model stability. This can be seen in the stability of parameters across model- variations (e.g. stability of parental age between models 1 and 4 of Table 2) and the relatively large sample size. In addition, multi-collinearity seemed to be a minor issue, as all individual predictors showed variance inflation factors of < 10 (24). Our drop-out analyses showed that relatively lower educated parents were more likely to have one or more missing values on their OP. This may be due to misunderstanding the Dutch translation of “OP” or the relative complexity of the question assessing OP. In addition, as we are aware that in some of our cross-sectional results reversed causation may have played a role, we urge future studies to use longitudinal designs in order to unravel this, especially regarding the relationship between parental and family attitude and OP. To date, OP can only be assessed by parental report, as often-used single objective measurements (e.g. accelerometers or heart-rate monitors) cannot distinguish OP from other types of PA. However, future studies need to assess OP using objective measures. In this regard, special attention should be directed towards combining global positioning system (GPS), GIS, and accelerometers, whose methodologies yield an objective assessment of domain-specific PA (25, 26). The present study used parental perception of their PE while nowadays more detailed objective assessment of the environment is also available: for example, with the use of geographic information systems (GIS) (27). Studies that directly compared objective and perceived PE suggest that these two concepts are different but interrelated (18, 28, 29). Although studies using perceived physical environment have shown relatively weak associations with objective PA (30, 31), consistency was higher when PA was also measured by parental reports (18). This is supported by several conceptual frameworks, which postulate that perceived PE may be a more proximal function of the objective environment (28, 29), as the influence of objective PE is moderated by personal factors and selective daily mobility (32, 33). Therefore, future studies need to include both objective and subjective measures of PE to unravel these phenomena.