Analyses were carried out using STATA/SE 11 [36 ]. Demographic characteristics of children with (n = 593) and without valid physical activity data (n = 492) were compared using independent t-tests. A significance level of 0.05, set a priori, was used for all tests.
Descriptive characteristics of the sample and overall daily minutes spent in each activity intensity were calculated, along with the percentage contribution of each intensity to total activity.
To investigate the influences on children’s MVPA, LPA and sedentary activity, a series of two-level random intercept models were used. Daily observations at level 1 were nested within participants at level 2. For each outcome, average daily activity level and activity segmented across the day (morning, afternoon and evening) were assessed. Due to non-normality, the logarithm of MVPA was used for regression analyses. As these regression coefficients refer to the log-transformed outcome variable, beta values were exponentiated to give a ratio of the geometric means (GMR). A GMR can be interpreted similarly to a risk or odds ratio: any deviation from 1 indicates a% difference in MVPA relative to the respective reference category in the exposure variable. All models were adjusted for sex, weight status, age child’s mother left full-time education, time of the week (weekday vs. weekend) and season. Two sets of sensitivity analyses were conducted to explore the impact of including children with differing numbers of valid days: first including and excluding children with only one or two days of valid physical activity data (n = 49), and second children with and without activity data for both weekday and weekend days (n = 85).
Descriptive characteristics of the sample and overall daily minutes spent in each activity intensity were calculated, along with the percentage contribution of each intensity to total activity.
To investigate the influences on children’s MVPA, LPA and sedentary activity, a series of two-level random intercept models were used. Daily observations at level 1 were nested within participants at level 2. For each outcome, average daily activity level and activity segmented across the day (morning, afternoon and evening) were assessed. Due to non-normality, the logarithm of MVPA was used for regression analyses. As these regression coefficients refer to the log-transformed outcome variable, beta values were exponentiated to give a ratio of the geometric means (GMR). A GMR can be interpreted similarly to a risk or odds ratio: any deviation from 1 indicates a% difference in MVPA relative to the respective reference category in the exposure variable. All models were adjusted for sex, weight status, age child’s mother left full-time education, time of the week (weekday vs. weekend) and season. Two sets of sensitivity analyses were conducted to explore the impact of including children with differing numbers of valid days: first including and excluding children with only one or two days of valid physical activity data (n = 49), and second children with and without activity data for both weekday and weekend days (n = 85).
Free full text: Click here
