Teun Remmers

44 | Chapter 3 This study adds to the current literature indicating that impulsivity attenuates the relationship between PA enjoyment and PA behavior in girls. In line with our hypothesis, this can be explained by the nature of impulsive behavior itself; since performance of leisure time PA occurs relatively automatically in impulsive children (40), and thus relatively independent of intrapersonal motives such as PA enjoyment. In girls with relatively low impulsivity on the other hand, leisure time PA occurs more deliberately and is thus more likely to be regulated by personal motives such as PA enjoyment. This is supported by studies that demonstrated impulsivity to be a key moderating factor of influences on child snacking behavior (22,23). In boys by contrast, we found that impulsivity strengthened the relationship between PA enjoyment and PA behavior. As we hypothesized that impulsivity would attenuate the relationship between PA enjoyment and PA behavior, findings in boys are contra-intuitive. One can however cautiously speculate that predominantly in boys, impulsivity may be likely to be associated with hyperactivity (41-43). Impulsivity (and thus hyperactivity) may strengthen the relationship between PA enjoyment and PA behavior in boys as 1) hyperactive boys may be deliberately provided with more PA opportunities by their parents (e.g. membership of sports clubs), thus positively reinforcing impulsive boys’ PA enjoyment. In this way, one may cautiously suggest that hyperactive boys may engage in PA more than their non- hyperactive peers, and also score relatively high on the enjoyment of these behaviors. Future studies with a longitudinal design are however warranted. In addition, future studies should incorporate a larger sample allowing for more flexible analytical strategies such as structural equation models in order to analyze various direct, moderating mechanisms simultaneously. Strengths and weaknesses In the usage of accelerometers, considerable differences exist in thresholds of defining PA intensity in children. In the present study, we used the thresholds of Evenson because of the similarity of the age of our children compared with those of Evenson’s study, and the diverse content of the testing protocol (e.g. free-play activities) (32). In these thresholds, counts above 100 counts per minute were already considered as light PA. As these thresholds were notably lower compared to other thresholds (32), our results may have overestimated the percentage of light PA in children, and underestimated the percentage of sedentary behavior. In this study, sedentary time is defined as any waking time where the accelerometer registered less than 100 counts per minute (often with a sitting or reclining posture) (32). In addition, one may suggest that differences exist between the two Actigraph accelerometer models (i.e. uniaxial ActiGraph 7164 and GT1M) used in our study. The question relevant to the present study is if these differences are large enough to have implications for activity intensity classification. Three studies showed that this was not the case in time spent in the moderate or vigorous category. If one did suppose that differences existed, then this potential bias would be randomly distributed over the entire sample, as we were unaware of any child characteristics when handing out the accelerometers. It is however questionable whether this potential misclassification affected our results, as information from parent diaries showed that 100 children (58,4%) had not swum during the accelerometry measurement. From the 71 children that did