Raising healthy Eating and Active Living Kids in Alberta (REAL Kids Alberta) is a population-based study of grade five children and their parents in the Canadian province of Alberta. In 2008, the REAL Kids Alberta survey included 148 randomly selected schools from across Alberta as described elsewhere 14 . In 2009, we selected and invited a convenient sample of 20 of the 148 schools, located in the city of Edmonton and surrounding areas, to participate in an additional survey that included objective measures of physical activity. We also invited grade five students from 10 schools participating in the Alberta Program Promoting active Living and healthy Eating in Schools (APPLE Schools) 15 . We repeated data collection among grade five students and parents in 2011 from the same schools that participated in 2009. In 2009 and 2011 combined, we sent 2,502 parent consent forms and surveys home with students for their parents to complete and return to school; 2,045 (81.7%) home surveys were returned, 2,028 (99.2%) students received parental consent to participate. Student assent to participate was obtained from 1,991 (98.2%) children, and 1,977 (99.3%) children completed student surveys, resulting in an overall participation rate of 79.0%. Pedometers were distributed to the 1,991 students with parent consent and student assent, 1,783 pedometers were returned, providing crude hourly step counts; 210 (10.5%) pedometers were lost or malfunctioned. The Health Research Ethics Board of the University of Alberta approved this study, including data collection and parental informed consent forms.

The parent survey included three validated questions related to beliefs and support that were adapted from the activity-related parenting practices scale by Davison et al. 16 . These included: 1) how much do you personally care about staying fit and exercising (a little bit; quite a lot; very much); 2) to what extent do you encourage your grade five child to be physically active (a little bit; quite a lot; very much); and 3) how often do you or another parent/guardian usually engage in physical activity together with your child (less than 1 time/week; 1–3 times/week; 4 or more times/week). The questionnaires we used are available on the project’s website: http://www.REALKidsAlberta.ca.

We used Omron HJ-720ITC (Ontario, Canada) pedometers to measure physical activity objectively. This pedometer records steps hourly, automatically resets at midnight, and can store data for 42 days. Further rationale for selecting this pedometer are described in detail elsewhere 3 . The accuracy and validity of the Omron pedometer has been demonstrated under various conditions 17 18 19 . Students wore pedometers for nine consecutive days on their right hip directly in line with their right knee during all waking hours unless showering, swimming, or taking part in activities in which an adult deemed it unsafe to wear. Pedometer recordings from the first and ninth days were not considered as, on these days, the students started or ended wearing the pedometers, and thus recordings are not available for the full day. Students also kept a diary of their daily activities, including the duration of each activity and whether or not the pedometer was worn (available at: http://www.REALKidsAlberta.ca). Trained evaluation assistants returned to schools on the ninth day to collect pedometers and download data to computers. Pedometer recordings were stratified into school day (Monday-Friday) and weekend day (Saturday, Sunday, and holidays) categories.

Evaluation assistants measured children’s standing height to the nearest 0.1 cm using stadiometers (Seca-Stadiometer, Germany) and body weight to the nearest 0.1 kg on calibrated digital scales (Health-o-meter, IL, USA). Children removed their shoes for both measurements. Body Mass Index (BMI) was calculated as kg/m². Overweight was defined using the International Obesity Task Force BMI cut-off point established for children and youth 20 . The cut-off point is based on the health-related adult definition of overweight (BMI ≥ 25), but is adjusted to specific age and sex categories for children. Analyses were adjusted for the confounding potential of parental educational attainment, household income, and year of data collection.

For our analyses we only considered pedometer recordings when worn for a minimum of 8 hours per day 21 . Additionally, we required pedometer recordings on at least two school days and one weekend day (weekend and/or holiday) 22 . When these conditions were met, students’ pedometer-measured steps were complemented with step-equivalents of non-ambulatory and non-wear time activities recorded in students’ activity diaries. Briefly, we assigned each activity recoded in activity diaries a youth-specific metabolic equivalent task (MET) unit 23 . Next, we categorized activities by intensity (i.e., moderate, moderate-to-vigorous, vigorous) 24 25 and assigned a step per minute value to each category 26 . Adult METs units were used when youth specific values were not available 24 . When students forgot to wear their pedometer and complete their activity diary, we imputed information from the same hour (s) on other randomly selected valid days. Steps were only imputed within an individual and within school days and weekend days. This method of imputation has been shown to replace data more accurately than traditional group-centered methods that replace missing data with the group mean 27 . These procedures are described in further detail elsewhere 3 .

In 2009 and 2011 combined, we had information of pedometer step counts and parental reporting of physical activity-related beliefs and support for 717 girls and 638 boys.

T-tests and chi-square tests were used to test for data collection year and sex differences. Because observations of students are clustered within schools, we applied random effects models with students nested in schools. For each of the parent belief and support variables, we first applied univariable linear regression models to determine their associations with girls’ and boys’ step-counts taken on weekend days and on school days. Second, we applied multivariable linear regression models to adjust for the confounding potential of parental educational attainment, household income, and year of data collection (referred to as Model 1). Last, we considered parent beliefs and support variables simultaneously while adjusting for the above confounders to quantify their independent importance for children’s step-counts (referred to as Model 2). For each of the parent belief and support variables we used the middle response category as the reference group for analyses.

In a combined analysis, we used an interaction term (defined as the product of school day/weekend day and the parent belief and support variables) in the adjusted linear regression models to quantify the differential effect of beliefs and support on physical activity during school days versus weekend days.

Missing values for parental education attainment and household income were treated as separate covariate categories however we do not present their estimated values. We used Stata Version 12 (Stata Corp, TX USA) to perform the statistical analyses.

On school days, increased parental encouragement of physical activity, and increased parental engagement in physical activity together were significantly and positively associated with girls’ daily step counts (Table  2: Model 1). Girls whose parents encouraged physical activity “very much” took an additional 632 (95% CI: 108, 1155) steps per day on school days relative to girls whose parents encouraged them “quite a lot”. Additionally, on school days, girls whose parents engaged in physical activity with them more than four times per week achieved an additional 890 (95% CI: 67, 1712) steps per day relative to girls whose parents engaged in physical activity with them one to three times per week. On weekend days, increased parental encouragement of physical activity was the only positive association with girls’ daily steps counts (β = 997, 95% CI: 130, 1864) that appeared to be statistically significant. Girls whose parents encouraged them “very much” to be physically active took an additional 997 (95% CI: 130, 1864) steps per day on weekend days than girls whose parents encouraged them “quite a lot”. Model 2 of Table  2 reveals that the three parental behaviours (parental care for staying fit and exercising, parental encouragement and parental engagement) are correlated such that none of the three behaviours has a statistically significant effect on girls’ step counts over and above that of the other two behaviours.

¹– pedometer-measured steps adjusted for non-ambulatory activities, non-wear time activities, and missing data.

²– Model 1 is adjusted for household income, parental educational attainment and year of data collection.

³– Model 2 is adjusted for parental care about staying fit and exercising, encourage physical activity, engage in physical activity together, household income, parental educational attainment, and year of data collection.

⁴– reference category.

^*– considered statistically significant at P < 0.05.

On school days, increased encouragement of physical activity was associated with daily step counts. This association was statistically significant (Table  3). Boys whose parents encouraged physical activity “very much” took an additional 1373 (95% CI: 606, 2139) steps per day on school days relative to boys whose parents encouraged physical activity “quite a lot”. This association was independent of parental care for staying fit and exercising and parental engagement in physical activity (Table  3: Multivariable Model 2). On weekend days, increased parental care about staying fit and exercising was positively associated with boys’ daily step counts (Table  3). Boys whose parents reported to care “very much” about staying fit and exercising took an additional 1381 (95% CI: 85, 2676) (Table  3) steps per day on weekend days relative to boys whose parents reported to care “quite a lot”. Also, on weekend days, decreased parental engagement in physical activity was negatively associated with boys’ daily step counts (Table  3). That is, boys whose parents engaged in physical activity with them less than once per week took 1367 fewer steps per day on weekend days relative to boys whose parents engaged in physical activity with them one to three times per week (95% CI: -2643, -90). This association was independent of parental care for staying fit and exercising and parental encouragement (Table  3: Multivariable Model 2).

¹– pedometer-measured steps adjusted for non-ambulatory activities, non-wear time activities, and missing data.

²– Model 1 is adjusted for household income, parental educational attainment and year of data collection.

³– Model 2 is adjusted for parental care about staying fit and exercising, encourage physical activity, engage in physical activity together, household income, parental educational attainment, and year of data collection.

⁴– reference category.

^*– considered statistically significant at P < 0.05.

There were no significant interactions between parental care for staying fit and exercising, parental encouragement of physical activity, or engagement in physical activity together and girls’ “school day/weekend day” step counts. Parental engagement in physical activity and boys’ “school day/weekend day” physical activity was the single statistically significant interaction. This interaction remained significant after adjusting for potential confounders. Boys whose parents engaged in physical activity with them less than once per week took 1475 (95% CI: -2609, -341) fewer steps per day on weekend days than they did on school days.