The minimum number of days required to establish reliable physical activity estimates in children aged 2-15 years.

The purpose of this study was to determine the minimum number of days needed to assess reliable estimates of step-count data for children based on age and gender. A total of 428 developing children (at least 30 boys and 30 girls in seven two-year intervals from 2-3 years to 14-15 years of age) wore a StepWatch accelerometer for 7 consecutive days. Following data screening, the 422 children's data were finally analyzed by age and sex groups using the Generalizability theory. Single-facet crossed designs (i.e. Participant (P) × Days (D)) were applied for each of 14 datasets. G-studies were performed to quantify the percentage of variance associated with the facet and interaction in the model. Follow-up D-studies were performed to determine the minimum number of days of step-count data collection needed to achieve a desirable reliability coefficient (G ≥ 0.80). The results from the G-studies show that P explained a large percentage of the total variance (26-71%) while D had little effect on the total variance (0-5%). A relatively large percentage of variance was unidentified (i.e. the P × D interaction). The minimum number of days necessary to achieve a desirable reliability coefficient (G ≥ 0.80) ranged from 2 to 12 d. On average, boys required less days of monitoring than girls. Researchers should use the findings of this study to design data collection that ensures reliable data.