Understanding variability in time spent in selected locations for 7-12-year old children.

This paper summarizes a series of analyses of clustered, sequential activity/location data collected by Harvard University for 160 children aged 7-12 years in Southern California (Geyh et al., 2000). The main purpose of the paper is to understand intra- and inter-variability in the time spent by the sample in the outdoor location, the location exhibiting the most variability of the ones evaluated. The data were analyzed using distribution-free hypothesis-testing (K-S tests of the distributions), generalized linear modeling techniques, and random-sampling schemes that produced "cohorts" whose descriptive statistical characteristics were evaluated against the original dataset. Most importantly, our analyses indicate that subdividing the population into appropriate cohorts better replicates parameters of the original data, including the interclass correlation coefficient (ICC), which is a relative measure of the intra- and inter-individual variability inherent in the original data. While the findings of our analyses are consistent with previous assessments of "time budget" and physical activity data, they are constrained by the rather homogeneous sample available to us. Owing to a general lack of longitudinal human activity/location data available for other age/gender cohorts, we are unable to generalize our findings to other population subgroups.