Development of scales to assess children's perceptions of friend and parental influences on physical activity

Regular physical activity has many short and long-term benefits for children including lower body mass index [1] and lower mean values for cardiovascular risk factors [2–4]. Physical activity is also associated with higher levels of mental well-being among children [5] and helps children to develop social skills [6, 7]. Despite these benefits many children and adolescents do not engage in recommended amounts of physical activity [8, 9]. The mediating variable model suggests that in order to change a behavior (such as physical activity) we need to change the key factors or mediators of that behavior [10]. Therefore, in order to develop effective interventions to increase children's physical activity we need to understand the factors, or correlates of children's physical activity and then change those variables [11].

Parents and friendship groups are likely to be key influences on children's physical activity behaviors. Davison and colleagues reported that fathers' modeling of active behaviors and mothers' logistic support for physical activity (e.g., enrolling child in sport programs and going to sporting events with the child) were associated with the physical activity levels of 9 year old girls [12]. Salvy and colleagues reported that 10 year old boys and girls were more likely to engage in high intensity physical activity with friends [13] and that friends increased children's motivation for physical activity. Peer support for physical activity has also indicated an association with higher amounts of physical activity among fifth to eighth grade US students [14]. While these studies demonstrate the importance of the influence of parents and friends on children's physical activity they provide limited information about how these influences are manifested. More information about potential mechanisms through which interactions with parents and friends shape physical activity behaviors is therefore needed to develop effective strategies to increase children's physical activity.

Our research team have conducted extensive qualitative work to examine how friends and parents influence the physical activity behaviors of 10-11 year old children [15–17]. The data on the influence of friends showed that 10-11 year old British children have three types of friendship groups: school friends, neighborhood friends and other friends (e.g. children of their parents' friends and children from youth or community groups) with most children belonging to multiple groups [15]. Findings also suggested that friendship group members shared common attitudes and perceived normative values for physical activity and screen-viewing [15]. Children also reported that their reasons or motives for participating in physical activity included intrinsic appeal, increasing social affiliation and preventing isolation [15]. Motives for physical activity participation were also influenced by group affiliation [15]. Therefore, understanding group affiliation may be important for understanding physical activity attitudes, expectations and norms among children.

Our qualitative work indicated that parents exert considerable influence on children's behavior through provision of transport assistance, financial support, modeling, encouragement, and setting rules for activity [16, 17]. Several participants reported that their family structure and particularly the parents or guardians who they live with differ for weekday and weekend days. This suggests that measures should accommodate these differences when examining parental influence.

There is a need to understand how friends and parental factors influence children's physical activity, parental rules for physical activity, and physical activity and screen-viewing subjective normative values. Subjective normative values are a key component of the Theory of Planned Behavior and represent a person's perception of other peoples (i.e. other children's) preferences for engaging in a behavior such as physical activity [18]. However, although there are reliable measures that capture parental support for physical activity [12] there are no scales that also examine parent imposed activity-related rules, or scales that can address how associations may differ by weekday or weekend parent. Similarly, while some measures have included items that address the extent to which spending time with friends contribute to physical activity enjoyment [19] to our knowledge no current measure identifies friend influences on physical activity and particularly if influence differs by type of friend. Finally, although physical activity norms and motives for physical activity scales have been developed these scales have tended to focus on more global subjective norms and included questions on fellow students, teacher and parent norms [20] and have not included screen-viewing behaviors. To address these limitations this paper reports the development and reliability assessment of three new questionnaire scales to assess parental influence on physical activity, motives for activity with friends, and group norms related to physical activity and sedentary behaviors.

Participants were 173, 10-11 year old children recruited from 7 primary schools in Bristol, England. We initially approached 9 primary schools with 2 schools declining, one because of recent changes to the school management team and the other due to participation in a number of other, non-physical activity based research projects. There were 373, Year 6 pupils within the 7 primary schools and as such the recruitment rate was 46.4%. Participant sex and highest education within the household were obtained by parental report. The study was approved by a University of Bristol ethics committee and informed consent and assent were obtained for all participants [21].

The General parenting support scale provides an indication of the overall support that the child perceives their parent provides for physical activity (23.2% of the variance, alpha = .83) while the Active parents sub-scale provides a measure of the extent to which the child perceives his or her parent to be active (19.4% of the variance, alpha = .84). Past parental activity provides an indication whether or not the child perceives that the parental used to be active (12.5% of the variance, alpha = .80) while the Guiding support (12.5% of the variance, alpha = .82) scale captures the extent to which the child's parents have supportive rules for physical activity participation. Intra-class correlations for all items included in the Parental influence on physical activity were between 0.6 and 0.8 (7 > 0.7) suggesting satisfactory reliability.

The associations among the 11 sub-scales are presented in Table S1 (Additional File 1). There were a number of statistically significant associations among the sub-scales with the strongest associations within the same scale being the Avoid bullying and Social affiliation sub-scales of the motives for activity with friends scale (r = .521, p < .001). There were also significant associations between sub-scales derived from different main scales for example Teasing norms was significantly associated with the Avoid bullying sub-scale of the motives for activity overall scale (r = .427, p < .001).

Correlation analyses indicated that when the sample included all participants, the Active parent sub-scale of the parental influence scale was associated with light intensity physical activity (r = .178, p = .042) while Active norms was associated with minutes of MVPA per day (r = .181, p = .039) (data not in tabular form). Pearson correlations between each of the 11 sub-scales and the four physical activity variables are presented separately for boys and girls in Table S2 (Additional File 2). Among girls, Social sedentary was associated with sedentary minutes per day (r = .273, p = .024) and negatively associated with mean counts per minute (r = -.249, p = .040). Parental past activity was negatively associated with minutes of light activity per day (r = -.307, p = .011) and Avoid bullying was negatively associated with minutes of MVPA per day (r = -.255, p = .037) and accelerometer counts per minute (r = -.268, p = .028). For boys, Social affiliation was associated with minutes of MVPA per day (r = .253, p = .050).

In this paper we have presented information on the factor structure and reliability of three new scales: the Parental influence on physical activity, Motives for activity with friends and Physical activity and sedentary norms. Items were only included in the scales if they had acceptable test - retest reliability and variance in responses and thus all scales can be considered reliable. The alphas for the Parental influence on physical activity and Motives for activity with friends scales as well as the alphas for all of the sub-scales of these measures were >0.7. Alpha values >0.7 are considered satisfactory for non-clinical instruments [39] and therefore we can be confident that the items included in these scales were measuring coherent concepts. The alpha for the overall Norms scale was .64 indicating that caution is required when attempting to use all of the items on this overall scale to describe friend related physical activity norms. Collectively, these analyses therefore highlight that we have developed new scales that have good test re-test reliability and internal consistency.

The associations between the sub-scales and physical activity were different for boys and girls, suggesting that the extent to which new sub-scales predicted physical activity differed by sex. For example, the correlation between Parental past activity and light intensity physical activity was -.307 for girls but there was no association for boys. This would suggest that parental physical activity influences girl's physical activity only. Similarly, the Social sedentary scale correlated .275 with minutes of sedentary time for girls but there was no association for boys. These findings are consistent with previous research which has shown that the association between correlates of children's physical activity differs by gender. For example, gender differences in the associations between self-efficacy, social norms, beliefs, and outcomes of children and adolescents have been reported [37, 40]. Findings suggest that the parental and friendship factors derived from our new questionnaire could be important predictors of behavior, but associations may well be sex specific and thus further research that examines these differences is required.

Although not statistically significant (p < .05), a number of the sex stratified associations between the sub-scales and physical activity variables were in excess of 0.15 and often above 0.20. Such associations are comparable to the associations between physical activity self-efficacy and physical activity [40–42]. A number of physical activity interventions have been designed in which self-efficacy is hypothesized to be a key mediator of physical activity behavior change [43, 44]. Although there is a shortage of studies that have employed mediating variable analyses of self-efficacy based interventions [11, 45] this is likely to be a function of the lack of success in changing youth physical activity which hampers the detection of mediation effects [45]. The comparison is salient because it suggests that these sub-scales could explain a considerable amount of the variance in children's physical activity behaviors. Therefore an appropriately powered study is needed to fully examine the associations between these constructs and children's physical activity.

The Prevent bullying sub-scale accounted for the highest amount of variance (18.6%) of the four factors on Motives for activity with friends scale, suggesting that this concept is particularly salient for some participants. Interestingly, Teasing was a specific sub-scale on the Norms scale suggesting that this factor was also a salient normative value. Being physically active is associated with a reduction in the likelihood that 13-15 year old adolescents are bullied [46]. Furthermore, athletic identity is associated with increased physical activity among children [47] and many children, particularly girls, are socialized out of physical activity by teasing or peer victimization on the basis of poor sporting ability [48]. As such these two sub-scales may identify children who are concerned about or perhaps at risk of peer teasing. The scales could also be utilized as a means of identifying particular groups of children who do not engage in key behaviors because of fears of social isolation and teasing.

The General parenting support and Active parents sub-scales include similar items to Davison's logistic support factor [12] but utilize more specific examples and have resulted in two factors rather than one. An interesting area of future research would therefore be to consider how the new sub-scales and Davison's measure are related to each other and whether utilizing all scales increases our understanding of how parents can help to support physical activity.

The Neighborhood friends sub-scale suggests that there is something specific about how children identify with this group of friends. As the concept of neighborhood friends and their influence on physical activity behaviors is new [15], exploring the role of this friendship group and how neighborhood friends can help promote activity in less active children is likely to be essential in fully understanding why different groups of children are active.

In the Motives for Activity with Friends scale there were two social sub-scales: Social sedentary which captured preferences for engaging in sedentary behaviors with friends; and Social affiliation which captured engaging in activity to spend time with friends. A number of studies have reported that social factors are associated with participation in physical activity [15, 49–51], and these scales extend that work by indicating that the social aspects of screen-viewing and physical activity are likely to be different. The increased specificity of these sub-scales suggest that they may be able to explain more of the variance in the behaviors to which they relate than current measures and research that focuses on this possibility is needed. Moreover, like the teasing and prevent bullying sub-scales, these two new sub-scales may be useful in developing profiles of children, particularly children who have preferences for either physical activity or screen-viewing behaviors.

The analysis presented in this paper has focused on the reliability of the new scales and not the "validity" of the scales [52]. However, as the development of the scales and items were informed by extensive qualitative work and the items assess the issues raised in the qualitative work the items can be considered to have "face and content validity" [52]. The items included in the scales address new constructs that were identified in the qualitative work and which have not been reported before and therefore there is no existing scale against which to compare these items. As such we are unable to assess the "criterion validity" but in order to provide an indication of the potential utility of these scales we have provided information on the associations with physical activity the key behavior to which they are hypothesized to relate.

Three scales that assess how parents, friends and group normative values may be associated with children's physical activity have been shown to be reliable and internally consistent. Initial analyses suggests that these measures will provide new information on the factors that influence children's physical activity but more research in a larger dataset is required to identify how associations may differ by sex and participant characteristics. They also provide further indication of the importance of conceptualizing physical activity as social-context specific with different parental and peer influences in action in different settings and time periods in the week.