Physical activity is associated with lower levels of several cardiovascular risk factors among children and adolescents (11,15). Physical activity is also associated with improved mental well-being among youth and helps to develop social skills (1,21). These benefits are well established, and Western governments have national campaigns to promote physical activity across the life span (8,32,33). Despite these benefits, most children and adolescents do not engage in the recommended 60 min of physical activity per day (9,15,24,25). There is a need to develop new methods of increasing children's physical activity.
Recent reviews of physical activity interventions or obesity prevention interventions that include physical activity elements report weak and unsustained effects in the best of cases (10,16,17,28,29,35). We therefore need to reexamine our current practice and identify ways in which it might be possible to design better strategies to increase children's physical activity. The mediating variable model would suggest that, to promote children's physical activity, we need to identify the key predictors of behaviors and then work to manipulate those factors (2-4).
A relatively underexplored factor is how friends influence children's physical activity. Data from US middle school girls have highlighted that engaging in physical activity with a friend is correlated (r = 0.24) (26) with self-reported physical activity. Similarly, in the Trial of Activity in Adolescent Girls, friends' support for physical activity was a mediator of physical activity (19). Recently, a small number of studies have attempted to apply social network analyses to children's physical activity and obesity behaviors (7,34). These studies suggest that friendship groups are important influences on children's behavior but the influence may be limited to organized forms of activity as opposed to unstructured, undirected forms of physical activity (34). The social network studies have been limited (34) to self-reports of physical activity. Thus, there is a lack of information on the influence of friends on boys' and girls' physical activity. There is also no information about whether the influence is a function of direct modeling of activity, co-participation, or encouragement, nor any indication of where the activity takes place. Moreover, most of the current data are limited to "friends" in general and not specific friends who have been identified by children. Data on smoking behaviors have shown that best friends were a stronger influence on adolescents' smoking behaviors than the wider social circle (31), but there is a lack of data on the influence of best friends on youth physical activity. In this study, we addressed these limitations by examining the extent to which the attitudes and actions of named best friends were associated with the physical activity patterns of 10- to 11-yr-old UK children.
Sampling and participants.
Data are from the Bristol 3Ps Project, a cross-sectional survey conducted in Bristol (UK) between 2008 and 2009; information on sampling has been reported previously (13). Briefly, participants were recruited from 40 primary schools. Recruitment was based on the Index of Multiple Deprivation (IMD) score for primary school location. The IMD is an area-level measure of deprivation that includes income, health, educational, and employment status (20), with higher scores indicating higher levels of deprivation, that is, lower socioeconomic position. Schools were randomly recruited from tertiles of school IMD. In most schools, there was one year 6 class per primary school, but some schools had larger year groups, and in these instances, all year 6 children within the school were invited to take part in the study. In total, 1684 year 6 children were invited to take part in the study, and 986 children provided data (a recruitment rate of 58.6%). This study was approved by the University of Bristol's ethics committee, and informed parental consent was obtained from all participants.
Physical activity was assessed using GT1M accelerometers (ActiGraph, Pensacola, FL), which were set to record every 10 s. All participants were provided with instructions on wearing the monitor, and data were collected for five complete days. Periods in which ≥60 min of zero counts were obtained were interpreted as the time when the monitor was not worn, and these periods were removed from the analysis (30). Each day of accelerometer data was considered valid if data were obtained for at least 500 min (27). Participants were included in the analysis if they provided ≥3 d of valid accelerometer data. Mean accelerometer counts per minute (CPM), which provides an indication of the overall volume of physical activity, were calculated. Mean minutes of moderate-to-vigorous intensity physical activity per day (MVPA) were also obtained using a criterion of minutes with counts ≥2912 (23). This threshold was based on the 3200 cpm criterion that was developed for children using whole-body calorimetry (23), the most robust of all methods that have been used. The threshold was adjusted by 9% (3200 × 0.91 = 2912) because the GT1M monitors yield values that are 9% different from the original 7164 monitors on which the threshold was obtained (6).
The influence of best friends on participants' physical activity was assessed using a modified version of the social network questionnaire developed for the Trial of Activity in Adolescent Girls (36). All participants were asked to identify their best friend within the school. Participants were also asked to indicate: 1) the frequency with which they participated in physical activity with their best friend (never, once a week, two times per week, three to four times per week, and five or more times per week, which were collapsed into zero to once a week, twice a week, three to four times a week, or five or more times per week); 2) the main location of activity with the best friend (not active with the friend, active at school, active at home or neighborhood, and other places, which were collapsed into school, home, or other); 3) if they were in a sports team/activity group with the friend (yes/no); 4) if the child had ever asked the friend to be active with them (yes/no); and 5) if the best friend had ever asked the participant to be active with them (yes/no).
Child height was measured using a SECA Leicester stadiometer (HAB International, Northampton, UK), and weight was measured using a SECA 899 digital scale (HAB International). Child body mass index (BMI = kg·m−2) was calculated and converted to an age- and sex-specific SD score (BMI SDS) (5). Pubertal status was self-reported using a validated scale (22) and five pubertal stages similar to Tanner stages derived.
The focus of the proposed research was the influence of perceived best friends on children's physical activity behaviors. As such, the analyses were limited to the influence of named individuals, and we did not consider attempting to model the broader social networks that exist within and outside of the school (12). Data were prepared for analysis by matching a child's physical activity data with those of their best friend. Because the focus of this study was on the influence of the respondent's perceived best friend, the children could nominate any child as their best friend. Only 170 (35.9% of the analyzed sample) nominations were part of reciprocated best friendships. Moreover, because the focus was on the influence of perceived best friends on the target child, it was possible for more than one child to nominate the same best friend. In cases where best friends were nominated from outside the study school or where complete physical activity data were unavailable, friendship pairs were excluded from analysis. Because of the small amounts of cross-sex friendships (n = 15), only same-sex friendships were considered. Descriptive statistics including means and SD were calculated for all continuous variables, with percentages used for categorical variables. Independent-sample t-tests and χ2 were used to examine whether there were any differences in the physical activity, BMI SDS, IMD, or sex distributions of participants retained or excluded from the analyses. As it has been shown that physical activity patterns in children differ by sex (10,25), all analyses were performed separately for boys and girls.
A series of univariate linear regression models were produced to examine whether a child's mean MVPA was associated with their best friend's physical activity (mean MVPA or mean CPM) or any responses from the friendship questionnaire. At this stage, models were not adjusted for potential confounders. After the analysis of univariate associations, those variables that had been found to be statistically significant (P < 0.05) for each sex were included in full linear regression models to examine the extent to which these variables were associated with a child's mean MVPA and mean CPM. Both boys' and girls' models were adjusted for BMI SDS and IMD score. Robust SE values were used to take account of clustering (nonindependence between students from the same school) in the computation of 95% confidence intervals and P values. Robust SE values use the residuals from each group of children in a school to control for the similarity of children within the same school (18). All analyses were undertaken in Stata version 10.1 (College Station, TX), and α was set at 0.05.
There were 472 participants with valid accelerometer and friendship group data, representing 52.5% of the overall sample. Boys and girls accounted for 41.9% and 58.1% of valid friendship data, respectively, which were significantly different from the proportions of the overall data set (boys = 46.4% and girls = 53.6%, P = 0.006). However, data were shown to be representative of the overall data set in terms of mean MVPA, mean CPM, BMI SDS, IMD score, and pubertal status. For the valid data, descriptive statistics are shown for all variables split by sex in Table 1. Independent-sample t-tests indicated that, compared with girls, boys engaged in more minutes of MVPA per day (42.19 vs 30.16, P < 0.001) and had a higher mean CPM (606.57 vs 513.23, P < 0.001). Girls, however, were at a further pubertal stage than boys were.
Results from the univariate linear regression models used to predict mean MVPA are summarized by sex in Table 2 and indicate that children's mean MVPA was positively associated with their best friends' in both girls (t = 2.25, P = 0.031) and boys (t = 4.28, P < 0.001). In addition, girls (t = 2.17, P = 0.036) and boys (t = 2.17, P = 0.037) who engaged in physical activity in the home or neighborhood with best friends undertook more minutes of MVPA than those who did not. Effects of other variables differed between sexes. For girls, increasing the frequency of physical activity with their best friend to twice a week or more was related to an increase in minutes of MVPA in all categories, whereas for boys, engaging together in a sports team or physical activity class was positively associated with mean MVPA (t = 2.17, P = 0.049).
The univariate linear regression models for mean CPM as the outcome are summarized by sex in Table 3 and indicate that, for girls, physical activity five or more times a week (t = 2.35, P = 0.024) or at home or in the neighborhood (t = 2.68, P = 0.011) with best friends was associated with increased mean CPM. For boys, best friends' mean CPM (t = 2.57, P = 0.015), activity in the home or neighborhood (t = 3.22, P = 0.003), activity in other places (t = 2.21, P = 0.034), and best friends' who initiated physical activity (t = 2.23, P = 0.032) were positively associated with increased mean CPM.
Full regression models comprising the significant variables identified in the univariate analysis are presented in Tables 4 and 5. For girls, mean MVPA was associated with the frequency of activity with the best friend (P ≤ 0.02 for all categories) and taking part in activity with the best friend at home or in the neighborhood (t = 2.27, P = 0.030), whereas BMI SDS was negatively associated (t = −2.75, P = 0.010) in a model that accounted for 17.8% of variance. Similarly, girls' mean CPM was associated with a greater frequency of activity of their best friend (t = 2.05, P = 0.047), location of activity (t = 2.71, P = 0.010), and lower BMI SDS (t = −2.43, P = 0.020) in a model that accounted for 10.1% of variance. Boys' mean MVPA was associated with their best friend's mean MVPA (t = 3.68, P = 0.001) and whether they engaged in physical activity with the best friend at home or in the neighborhood (t = 2.52, P = 0.017), in a model that accounted for 22.4% of total variance. Boys' mean CPM was associated with whether the boy predominately engaged in physical activity at home or in the neighborhood (t = 3.53, P = 0.001) or in other places (t = 2.26, P = 0.031) when compared with school-based activity, in a model that accounted for 22.2% of the variance.
The data reported here indicate that the physical activity patterns of 10- to 11-yr-old children are associated with the physical activity of their "best friends." For boys, MVPA is directly associated with the MVPA of their best friend. Moreover, for boys, being active with their best friend at home or in the local neighborhood was associated with higher levels of physical activity (both MVPA and CPM) than just being active with best friends at school. Our findings therefore extend the previous research with adolescent girls (19,26) by demonstrating not only that best friends are important influences on boys' behavior but also that taking part in activity with a best friend at home or in the local neighborhood is associated with higher levels of physical activity. Collectively, these findings indicate that promoting physical activity with best friends and particularly activity outside of school hours could be an effective means of increasing the physical activity of 10- to 11-yr-old boys.
For girls, the frequency of engaging in physical activity with their best friend was associated with more minutes of MVPA per day and higher mean CPM, indicating a greater volume and intensity of physical activity. Girls who took part in physical activity with their best friend at home or in the local neighborhood also had higher physical activity levels than girls who only took part in physical activity with their best friend at school. Thus, although the patterns of association were slightly different for girls compared with boys, the overall message for girls would be that engaging in physical activity with your best friend often and outside of school hours is associated with higher levels of physical activity. Therefore, promoting this message or providing opportunities outside as well as within school for physical activity could be an effective means of promoting physical activity in children of this age group.
In a previous qualitative research, we have reported that friendship groups influence children's physical activity via co-participation, modeling, and verbal encouragement to be active (12). In this study, we found little evidence, for either asking a best friend to be active or being asked by a best friend to be active, influencing the physical activity patterns of either boys or girls. Thus, our data may indicate that, among best friends, modeling and spending time being active together are important ways that friendship influence is demonstrated and continually reinforced. This assertion is supported by the positive associations of target child and best friend MVPA among boys and implies that a greater understanding of the mechanisms by which modeling and coparticipation occur is needed to understand how best to use the influence of best friends to promote physical activity.
In the adjusted model, we found no evidence that being on a sports team with a best friend was associated with physical activity among either boys or girls. Because being on a sports team has been associated with higher levels of physical activity among adolescent girls (26), the absence of a finding here may reflect that the participants were 10-11 yr and thus perhaps less likely to be members of sports teams than older children. Alternatively, the absence of an association may indicate that the influence of best friends on activity is not exerted via organized sport clubs but via more unstructured physical activity or "active play" at home or in the local neighborhood.
The proportion of variance explained by the models was different for boys and girls. The proportion of variance in boys' MVPA that was accounted for by the model was 22.4%, whereas it was only 17.8% for girls. There was also a more stark difference for mean CPM whereby the model accounted for 22.2% of the variance for boys but only 10.1% for girls. The considerable differences in the proportions of variance in mean CPM that was accounted for by the models likely reflect the well-established differences in boys' and girls' physical activity (9,14), which were also evident in this data set. Moreover, in this data set, the influence of best friends on youth physical activity also differed by sex, and collectively, these results highlight that factors that explain youth physical activity differ by sex. Efforts to change youth physical activity therefore need to consider these factors and indicate a need to develop sex-specific strategies to change youth physical activity.
Strengths and limitations.
The major strength of this study is the objective assessment of physical activity via accelerometers for both target children and their best friend, which has made it possible to look at direct associations using reliable methods. Supplementing the objective assessment with self-reported measures of how often and where activity takes place with the best friend has also facilitated an understanding of how best friends influence other children's physical activity behavior. However, the study is limited by a sample of 10- to 11-yr-olds drawn from a single English city, which makes it difficult to generalize conclusions about the nature of these associations to children of different ages from different locations. It is also important to recognize that children have three types of friendship groups: school friends, neighborhood friends, and other friends (from organized groups, family friends, etc.), but this study has been limited to best friends who attend the same school (12). Thus, we have not accounted for the influence of best friends who do not attend the same school, are in a different year group, or who were nonparticipating year 6 children. Although it seems likely that our approach captured the majority of best friends, we will not have been able to assess the influence of non-school best friends, and therefore, it is possible that associations have been attenuated by this omission. Finally, because the data are cross-sectional, it is not possible to identify the direction of causality in associations and particularly whether the best friend is influencing the target child, vice versa, or if the associations are bidirectional. Further examination of the influence of best friends in larger data sets, from different locations, and with multiple data points is therefore warranted to further examine these issues.
Boys who have best friends who are physically active engage in greater amounts of physical activity. Girls who frequently take part in physical activity with their best friend obtain higher levels of physical activity than girls who do so less frequently. Boys and girls who take part in physical activity with their best friend at home or in the neighborhood where they live engage in higher levels of physical activity. Therefore, for 10- to 11-yr-old children, engaging in physical activity with their best friend often and outside of school hours is associated with higher levels of physical activity. Collectively, these findings indicate that interventions that focus on building support for physical activity among friendship groups and encouraging friends to be active together, particularly outside of school, may yield important changes to children's physical activity.
This project was funded by a project grant from the British Heart Foundation (ref PG/06/142). This report is also from research arising from a Career Development Fellowship (to Dr. Jago) supported by the National Institute for Health Research. The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the National Institute for Health Research, or the Department of Health.
The authors thank all of the children and schools that participated in this study.
The authors have no conflicts of interest to declare.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Bailey R. Evaluating the relationship between physical education, sport and social inclusion. Educ Rev
2. Baranowski T, Anderson C, Carmack C. Mediating variable framework in physical activity interventions. How are we doing? How might we do better? Am J Prev Med
3. Baranowski T, Cerin E, Baranowski J. Steps in the design, development and formative evaluation of obesity prevention-related behavior change trials. Int J Behav Nutr Phys Act
4. Baranowski T, Jago R. Understanding mechanisms of change in children's physical activity programs. Exerc Sport Sci Rev
5. Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the UK, 1990. Arch Dis Child
6. Corder K, Brage S, Ramachandran A, Snehalatha C, Wareham N, Ekelund U. Comparison of two ActiGraph models for assessing free-living physical activity in Indian adolescents. J Sports Sci
7. de la Haye K, Robins G, Mohr P, Wilson C. Obesity-related behaviors in adolescent friendship
networks. Soc Netw
8. Department of Health PA, Health Improvement and Prevention. At Least Five a Week: Evidence of the Impact of Physical Activity and Its Relationship to Health: A Report from the Chief Medical Officer
. London (UK): Department of Health PA, Health Improvement and Prevention; 2004. pp. i-vi.
9. Jago R, Anderson C, Baranowski T, Watson K. Adolescent patterns of physical activity: differences by gender, day and time of day. Am J Prev Med
10. Jago R, Baranowski T. Non-curricular approaches for increasing physical activity in youth: a review. Prev Med
11. Jago R, Baranowski T, Baranowski JC, Thompson D, Greaves KA. BMI from 3-6 y of age is predicted by TV viewing and physical activity, not diet. Int J Obes Relat Metab Disord
12. Jago R, Brockman R, Fox KR, Cartwright K, Page AS, Thompson JL. Friendship
groups and physical activity: qualitative findings on how physical activity is initiated and maintained among 10-11 year old children. Int J Behav Nutr Phys Act
13. Jago R, Fox KR, Page AS, Brockman R, Thompson JL. Parent and child physical activity and sedentary time: do active parents foster active children? BMC Public Health
14. Jago R, Froberg K, Cooper AR, Eiberg S, Andersen LB. Three-year changes in fitness and adiposity are independently associated with cardiovascular risk factors among young Danish children. J Phys Act Health
15. Jago R, Wedderkopp N, Kristensen PL, et al. Six-year change in youth physical activity and effect on fasting insulin and HOMA-IR. Am J Prev Med
16. Kipping RR, Jago R, Lawlor DA. Obesity in children. Part 1: epidemiology, measurement, risk factors, and screening. BMJ
17. Kipping RR, Jago R, Lawlor DA. Obesity in children. Part 2: prevention and management. BMJ
18. Kirkwood BR, Sterne AC. Essential Medical Statistics
. London (UK): Blackwell; 2006. p. 354-70.
19. Lytle LA, Murray DM, Evenson KR, et al. Mediators affecting girls' levels of physical activity outside of school: findings from the trial of activity in adolescent girls. Ann Behav Med
20. Noble M, McLennan D, Wilkinson K, Whitworth A, Barnes H, Dibben C. The English Indices of Deprivation
. London (UK): Department of Communities and Local Government; 2007. p. 1-129.
21. Parfitt G, Eston RG. The relationship between children's habitual activity level and psychological well-being. Acta Paediatr
22. Petersen AC, Crockett L, Richards M, Boxer A. A self-report measure of pubertal status: reliability, validity, and initial norms. J Youth Adolesc
23. Puyau MR, Adolph AL, Vohra FA, Butte NF. Validation and calibration of physical activity monitors in children. Obes Res
24. Riddoch C, Andersen LB, Wedderkopp N, et al. Physical activity levels and patterns of 9- and 15-yr-old European children. Med Sci Sports Exerc
25. Riddoch CJ, Mattocks C, Deere K, et al. Objective measurement of levels and patterns of physical activity. Arch Dis Child
26. Sharma SV, Hoelscher DM, Kelder SH, Diamond PM, Day RS, Hergenroeder AC. A path analysis to identify the psychosocial factors influencing physical activity and bone health in middle-school girls. J Phys Act Health
27. Steele RM, van Sluijs EM, Cassidy A, Griffin SJ, Ekelund U. Targeting sedentary time or moderate- and vigorous-intensity activity: independent relations with adiposity in a population-based sample of 10-y-old British children. Am J Clin Nutr
28. Stice E, Shaw H, Marti CN. A meta-analytic review of obesity prevention programs for children and adolescents: the skinny on interventions that work. Psychol Bull
29. Summerbell CD, Ashton V, Campbell KJ, Edmunds L, Kelly S, Waters E. Interventions for treating obesity in children. Cochrane Database Syst Rev
30. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer
. Med Sci Sports Exerc
31. Urberg KA. Locus of peer
influence: social crowd and best friend. J Youth Adolesc
32. US Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans
. Washington (DC): US Department of Health and Human Services; 2008. p. 1-61.
33. US Surgeon General. US Surgeon General's Report on Physical Activity and Health
. Washington (DC): US Department of Health and Human Services; 1996. p. 1-9.
34. Valente TW, Fujimoto K, Chou CP, Spruijt-Metz D. Adolescent affiliations and adiposity: a social network analysis of friendships and obesity. J Adolesc Health
35. van Sluijs EM, McMinn AM, Griffin SJ. Effectiveness of interventions to promote physical activity in children and adolescents: systematic review of controlled trials. BMJ
36. Voorhees CC, Murray D, Welk G, et al. The role of peer
social network factors and physical activity in adolescent girls. Am J Health Behav
Keywords:©2011The American College of Sports Medicine
NEIGHBORHOOD; ACCELEROMETER; PEER; FRIENDSHIP; MODELING