Whole-day and segmented-day physical activity variability of northwest England school children Stuart J. Fairclough a,b, ⁎ , Zoe H. Butcher a,b , Gareth Stratton a,c a REACH Group, Liverpool John Moores University, UK b Centre for Excellence in Physical Education, Sport, Dance, and Outdoor Education, Liverpool John Moores University, I.M. Marsh Campus, Barkhill Road, Liverpool L17 6BD, UK c Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, UK Available online 18 January 2007 Abstract Objective. This study aimed to (1) assess the day-to-day variability of children's weekday physical activity for the whole-day, and when segmented into discrete periods of the day; and (2) compare boys' and girls' physical activity variability. Method. Fifty-eight children (aged 7–11 years; 31 boys) from a northwest England city wore accelerometers for 4 consecutive weekdays during November and December 2005. Intra-class correlations (ICCs) were calculated for 1 and 4 monitored days and segments of days to determine reliable estimates of moderate-to-vigorous physical activity (MVPA). The Spearman–Brown prophecy formula established the number of monitoring days required for 80% reliability. Results. Higher ICC coefficients were typically observed among boys. Eight days of boys' whole-day monitoring were needed to achieve a reliability of 0.8, while 10 days were required for girls. The pre-school segment was the most stable (boys' ICC = 0.862; girls' ICC = 0.770). Compared to girls, fewer days of boys' monitoring would be required during all segments other than the school day. Conclusion. Day-to-day physical activity variability was sex-specific, with boys' MVPA generally more stable than girls'. The greatest reliability occurred between 7 am and 3 pm, suggesting that physical activity behaviors are more consistent in the school environment. © 2007 Elsevier Inc. All rights reserved. Keywords: Boys; Girls; MVPA; Weekday; Segmented day; Sex-specific variability; Accelerometer Introduction Monitoring of children's physical activity is an important public health surveillance measure. This can, however, present some challenges as children's activity is typically intermittent (Bailey et al., 1995) and varies between days, and between weekdays and weekends (Trost et al., 2000), meaning that several days of monitoring are required to gather a reliable estimate of ‘typical’ physical activity behaviors. Such extended monitoring periods can be financially costly to undertake and onerous on the participants. To address this issue, investigators have applied statistical procedures to estimate the number of days monitoring needed to reflect children's habitual physical activity (Janz et al., 1995; Murray et al., 2004; Treuth et al., 2003; Trost et al., 2000). Based on whole-day physical activity measurement, it is recommended that 4 to 5-days of monitoring are necessary to reliably estimate children's activity (Trost et al., 2005). Yet, an unexplored line of enquiry is how many days of monitoring are needed if physical activity is being assessed during discrete periods of the day. This may be particularly important in relation to studies conducted in the school environment, which is a setting that is significantly related to children's physical activity (Fein et al., 2004). If reliable physical activity data were demonstrated in different contexts, this might allow shorter monitoring periods to be undertaken. To our knowledge, only one previous study, on physical activity during school recess has reported day-to-day variability in this way (Ridgers et al., 2006a). Boys' and girls' physical activity motivations (Whitehead, 1993), attitudes (Cardon et al., 2005), and patterns (Riddoch et al., 2004; Trost et al., 2002) differ, yet previous studies of Preventive Medicine 44 (2007) 421 – 425 www.elsevier.com/locate/ypmed ⁎ Corresponding author. Centre for Excellence in Physical Education, Sport, Dance, and Outdoor Education, Liverpool John Moores University, I.M. Marsh Campus, Barkhill Road, Liverpool L17 6BD, UK. Fax: +44 151 231 5357. E-mail address: s.j.fairclough@ljmu.ac.uk (S.J. Fairclough). 0091-7435/$ - see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2007.01.002