Abstract
Participation is defined as involvement in a life situation (World Health Organization, 2001). Recent reviews suggest two key conceptual foci in the participation construct: attendance and involvement (Adair, Ullenhag, Keen, Granlund, & Imms, 2015; Imms, Adair, et al., 2016). Attendance relates to the number of activities a child participates in (diversity of participation), the frequency of taking part, or both. Involvement relates to a person’s affect; motivation; and, in some situations, social engagement while participating (Imms, Adair, et al., 2016).
Previous research has suggested that the determinants of participation in children with disability include child activity preferences, age, gender, severity of disability, family factors, and environmental factors such as access to activities (Imms, Reilly, Carlin, & Dodd, 2009; King, Law, Hurley, Petrenchik, & Schwellnus, 2010; King, Petrenchik, et al., 2010). One study investigated the determinants of participation of children without impairments and identified differences based on gender and age; however, it was limited to children ages 6–14 yr who lived in Canada (King, Law, et al., 2010). Descriptive information about the characteristics of Australian children who participate in physical activities for recreation, exercise, and sport has been reported (Australian Government, 2012). These data suggest differences in participation might exist for children based on age, gender, and language spoken at home. However, the relative impact of personal (e.g., age, gender, body mass index [BMI], preference) or environmental (e.g., socioeconomic status, living location, school attended, cultural background) characteristics on participation diversity has not been examined.
Participation in activities outside of mandated school activities is important for children and adolescents to help them develop skills, make friends, and establish patterns of participation that may or may not be health promoting (Law, 2002). Health-promoting activities provide opportunity for physical activity and support social connections (Murphy & Carbone, 2008) and are important to prevent obesity, mental health problems, and social isolation.
The current study aimed to further understanding of the determinants of the diversity of participation of children and adolescents without impairments by investigating what influences their participation in out-of-school activities. These data can be used by occupational therapy practitioners, physiotherapists, educators, researchers, community developers, and policy makers to understand patterns of participation and inform programs and strategies for fostering community engagement and participation. Our exploratory research question was “Which environmental and individual characteristics most strongly influenced the diversity of participation in recreational, active physical, social, skill-based, and self-improvement activities in typically developing children and adolescents?”
Method
Design
We conducted a cross-sectional survey of typically developing children and adolescents from Victoria, Australia. Ethical approval was obtained from the human research ethics committees of relevant universities and the Victorian Department of Education and Early Childhood Development. Participants and their parents gave written informed consent or assent to take part.
Participants
Eligible participants were ages 6–18 yr; enrolled in mainstream schools in Victoria, Australia; and had sufficient English-language skills to complete the study questionnaires. Children and adolescents were excluded if they attended a special school or if they had a physical, cognitive, or developmental disability as defined by their need for an integration aid at school.
Outcome Measures
Each child completed the Children’s Assessment of Participation and Enjoyment (CAPE) and Preferences for Activity of Children (PAC) questionnaires (King et al., 2004) with or without assistance from a parent, teacher, or other person (e.g., older sibling). The CAPE comprises 55 items that are grouped into Recreational (12 items; e.g., playing board games, watching TV, playing with a pet), Active Physical (13 items; e.g., track and field, water sports, playing nonteam sports), Social (10 items; e.g., talking on the phone, visiting, going to the movies), Skill-Based (10 items; e.g., swimming, learning to dance, playing a musical instrument), and Self-Improvement (10 items; e.g., reading, doing homework, chores) activity types. The PAC comprises the same 55 items, and the child is asked about his or her preference to participate in each activity.
In this study, diversity and preference data were analyzed according to these five activity types. Diversity scores represent the number of different activities a child participates in per activity type during the previous 4 mo. Diversity scores provide information about the range of activities the child does, regardless of the frequency of participation. Preference scores represent the child’s preference for each activity type as really like, sort of like, or not like at all, regardless of whether the child participated in the activities.
Evidence supporting the validity of the CAPE and PAC questionnaires for children and youth ages 6–21 yr is available (Imms, 2008; King et al., 2004, 2007) and includes content validity based on a literature review, expert review, and pilot testing (King et al., 2007); convergent and discriminate validity based on relationships with standardized measures of environmental, family, and child factors; and concurrent validity based on comparing enjoyment and preference scores (King et al., 2007). Both questionnaires have adequate retest reliability for diversity scores (King et al., 2004) with α values ranging from .67 to .77 and adequate internal consistency (King et al., 2004) with α values ranging from .30 to .65 for the five activity types.
The participants (or their parents) completed a demographic questionnaire about their gender, age, height, weight, home address, school, year level, country of birth (Australia or other), parents’ country of birth (Australia or other), languages spoken at home, and the need for an education support teacher and whether any significant events in the past 4 mo had influenced usual participation. Socio-Economic Indexes for Areas (SEIFA) advantage/disadvantage scores at collector district level were calculated based on census information from an Australian Bureau of Statistics (2006b) database. Higher SEIFA scores indicate greater socioeconomic and resource advantage. Home location (address), documented as major city or regional, was determined using an Australian Bureau of Statistics database. BMI was calculated for each participant according to the standard equation weight in kilograms/height in meters squared. These values were compared with the thresholds provided by Cole, Bellizzi, Flegal, and Dietz (2000) to determine whether the child or adolescent was underweight/healthy or overweight/obese when standardized for age and gender.
Procedure
Participants were recruited using a combination of stratified random and convenience sampling. The reason for using a combination of sampling methods was to balance rigor with feasibility to recruit a large enough sample to ensure data from the study were representative and meaningful. A stratified random sample of 21 government schools in Victoria was selected according to the Like Schools Group ranking and agreement to participate. This ranking rates schools from 1 to 9 based on government funding received. Schools were categorized into three strata (high, middle, or low Like Schools Group rank), and a random selection of schools from each strata was made using a random numbers table. If a school declined to participate, then the next school in that stratum was approached. In addition to the random selection of schools, 4 schools (1 government, 1 independent, and 2 Catholic) were invited to participate using methods of convenience.
Participating schools distributed information about the study to their students. Interested students contacted the researchers directly and were sent a survey package. This process of recruitment was used in 22 of the 25 included schools. In the other 3 schools, researchers or teachers conducted an in-school data collection session for students who returned a signed consent form to the school. Missing data were followed up where possible by telephone or mail.
Data Analysis
Demographic characteristics were summarized using descriptive statistics (Table 1). Diversity and preference scores were calculated for each activity type according to the CAPE/PAC manual (King et al., 2004; Table 2). Higher scores indicated higher diversity and greater preference for activity types. Diversity and preference scores were considered continuous variables (King, Law, et al., 2010). Scores were calculated if at least 80% of data in each level were available (King et al., 2004).
Participant Characteristics (N = 422)
Note. BMI = body mass index; M = mean; SD = standard deviation.
BMI categories standardized by age and gender.
Country of birth listed as other when both parents were born in a country other than Australia or when one parent was born in Australia and one elsewhere (countries other than Australia included the United Kingdom, Ireland, New Zealand, China, and Vietnam).
Mean SEIFA, Preference, and Diversity Scores
Note. M = mean; SD = standard deviation; SEIFA = Socio-Economic Indexes for Areas.
Linear regression analyses were used to examine whether individual characteristics (age, gender, BMI, preference) and environmental factors (SEIFA, home location, school type, parent’s country of birth) influenced participation diversity for the five activity types. Independent variables were included in the regression models where there was evidence of a univariate relationship for at least one activity type (Pearson correlation, r > .1; p ≤ .20; Mickey & Greenland, 1989). If necessary, redundancy among predictors was avoided by removing any variable that correlated very highly with another (>.90). The variables were entered simultaneously.
Checks for multicollinearity were made by examining the tolerance and variance inflation factors; values <0.1 and >10 were taken to indicate multicollinearity (Pallant, 2013). The influence of outliers was assessed using Cook’s distance and leverage statistics (Pallant, 2013). Standardized residuals were examined, and a cutoff of ±3.3 was used to define significant outliers (Pallant, 2013). The number of outliers was small, so no data were removed from the analyses (Pallant, 2013). In the interest of parsimony, secondary regression analyses were performed that included only those variables that made significant independent contributions to the full models (p < .05).
Tests of normality suggested social diversity and skill-based diversity data were not normally distributed. A reflection/square root equation was used to transform the former, and a log transformation was used for the latter. However, the regression analyses using the transformed variables did not improve model fit; therefore, untransformed variables are reported, because these findings are more readily interpreted.
Post hoc analyses investigated the determinants of activity preference. Stepwise imputation was performed with activity preference as the dependent variable. The independent variables during these post hoc analyses included those previously described except for activity diversity. All data analyses were conducted using IBM SPSS Statistics (Version 22.0; IBM Corp., Armonk, NY).
Results
Of 9,337 potential participants targeted through school advertising, 512 agreed to participate (5.5% consent rate) and 422 returned completed questionnaires (82% response rate; see Table 1). Our sample size targets were met for each age group except the 15- to 18-yr-olds (Imms, Froude, Adair, & Shields, 2016). The study sample was representative for gender and language spoken at home compared with the Victorian population (Department of Education and Early Childhood Development, 2014). The SEIFA scores were higher than the Victoria population average, mean = 1,049.8, standard deviation = 87.2, p < .001 (see Table 2), indicating the sample had slightly higher socioeconomic resources than the state average.
Determinants of Participation Diversity
Preliminary bivariate analyses revealed relationships between diversity scores of at least one activity type and five of the eight proposed predictor variables: age, gender, activity preference, SEIFA, and school type. In each case, there was a significant small to moderate relationship, r = .15–.62, between the proposed predictor variable and the given activity type, p < .005. The correlation between predictor variables ranged from negligible to moderate, rs = .01–.62; therefore, no predictor variable was removed because of multicollinearity. The resultant regression models explained a large amount of variance in the diversity of recreational activities (44%) and moderate amounts of variance (between 19% and 25%) for the other four activity types (Table 3).
Full Regression Model, Including Independent Variables, for Diversity of Participation in Each Activity Domain
Note. For the full model, a R 2 is the overall predictive value. CI = confidence interval; PAC = Preferences for Activity of Children; SEIFA = Socio-Economic Indexes for Areas.
Of the five predictors, and consistent with our hypothesis, preference made a significant independent contribution to participation diversity for all activity types (see Table 3): The higher a child’s preference for an activity type, the greater their diversity of participation in that type of activity. Preference contributed unique variances of between 10% and 18% to the total variance of the full regression models.
Age made a significant independent contribution to diversity of participation in recreational activities only (Table 4): Younger children participated more diversely in recreational activities than did older children and adolescents. Gender made a significant independent contribution to diversity of participation in social and self-improvement activities: Girls were more likely to participate in these types of activities than boys. Socioeconomic status predicted diversity of participation in self-improvement activities, indicating that children from communities with more socioeconomic resources (i.e., higher SEIFA scores) were more likely to have a higher diversity of participation in these activities (see Table 3). Those who attended a private or Catholic school (where fees are paid) were likely to participate with less diversity in social and self-improvement activities than those who attended a public or government school.
Final Regression Model of Predictors of Participation Diversity and Preference
Note. — = no variables entered; CI = confidence interval; SEIFA = Socio-Economic Indexes for Areas. Preference for Participation: Variables are listed in the order of which best predicted preference scores for the activity types. All models were found to be a good fit for the data with p < .001.
Variables Influencing Preferences for Participation
Of the seven proposed predictor variables, five demonstrated a bivariate relationship, r > .1, p ≤ .2, with preference for at least one activity domain and were included in the analyses: age, gender, SEIFA, school type, and parent country of birth. These models explained moderate amounts of variance (between 14% and 35%) in the preference for the activity types (see Table 4, Preference for Participation).
The stepwise regression ceased after the inclusion of three variables (age, gender, and school style) for recreational, skill-based, and self-improvement preferences (see Table 4). For social preferences, the stepwise regression ceased after the inclusion of a single variable (gender). There were no variables entered into the equation for active physical preferences because no predictor variable explained a sufficient amount of variance.
Discussion
Participation can be understood as the extent to which children attend activities and their degree of involvement while attending (Imms, Adair, et al., 2016). This study explored predictors of diversity over a 4-mo period for five activity types (Recreational, Skill-Based, Social, Active Physical, Self-Improvement) outside of mandated school hours.
Age was only a predictor of diversity for recreational participation; younger children took part in more recreational activities than older children. Consistent with other research, this finding appears to be related to the type of activities included in the recreational domain of the CAPE questionnaire (e.g., doing puzzles, drawing or coloring, pretend or imaginary play, playing on equipment). In contrast with prior research, there was little evidence that gender was associated with participation diversity in active physical activities. Previously reported gender differences were of small magnitude, with girls reported to participate less frequently in the activities they do undertake (Australian Government, 2012). Girls in this study took part in more social and self-improvement activities than boys. This finding is consistent with data from Canada (King, Law, et al., 2010) and Australia (Australian Bureau of Statistics, 2006a), although it is not possible to compare findings directly with the latter because of differing measures.
Based on the regression analyses, there was little evidence that socioeconomic status was important, with only a small positive influence of increased socioeconomic resources on participation diversity in self-improvement activities. Interestingly, whether a child attended a private or public school was a predictor of participation diversity in social and self-improvement activities: Children who attended public schools participated more diversely than those who attended private or Catholic schools. This is an interesting finding given the emphasis on sport and extracurricular activity within the Australian private school sector (Bloomfield, 2003). One explanation might be that the expectation to participate in organized sport in the private school sector resulted in these children and adolescents having less time to engage in social and self-improvement activities, although this premise is speculative. The finding is also encouraging because it suggests that in this sample, socioeconomic resources and home location tended not to restrict participation. In fact, a recent small study suggested that children living in rural settings participated more in outside school activities than children in urban areas (Brown, O’Keefe, & Stagnitti, 2011).
Previous studies suggested that a reciprocal relationship exists between being overweight and restrictions on the diversity and frequency of physical activities undertaken (Cairney, Rigoli, & Piek, 2013). Surprisingly, there was no relationship between BMI and participation diversity in active physical activities in our study. Statistically, some restrictions in the range of values for diversity may have influenced our findings. In addition, the Participation Diversity scale of the CAPE questionnaire describes only the range of activities a person does—it does not provide information about the frequency of participation or the vigor with which he or she participates, which might be influenced by BMI. Reduced physical activity in children is associated with poor motor coordination and physiological function on a range of fitness metrics (Cairney et al., 2013), so structural modeling and longitudinal work are recommended to better understand the causal relations among these variables.
Preference is related to, but not synonymous with, participation (Imms, Adair, et al., 2016). People’s preferences influence their activity choices, typically toward activities that hold meaning to them. Consistent with evidence from the disability sector (Imms et al., 2009; King, Law, Petrenchik, & Hurley, 2013; King, Petrenchik, et al., 2010), preference for activity types was associated with participation diversity. Given this finding, exploratory analyses were completed to further understand the association between child and environmental characteristics on preferences for participation.
Two child-related variables (gender and age) and one environmental variable (school type) were found to influence preference: Being female and younger predicted holding stronger preferences than being male and older. Interestingly, the strongest association with age was for recreational preferences. One explanation might be that younger children have less experience with a range of pursuits (e.g., doing a paid job, going to a live event, doing volunteer work, participating in community organizations), and they associate preferences with “being older” rather than a desire borne of experience. The finding that children and adolescents who attended public schools had higher preference for recreational, skill-based, and self-improvement activities is difficult to explain. One explanation might be that students who attend schools that mandate their participation in extracurricular activities experience a reduced preference for participation, perhaps related to perceptions of choice.
Last, none of the predictors were associated with preference for activities in the Active Physical domain. This might indicate that both boys and girls maintain a steady preference for active physical activities, even into the teenage years, when the range of activities (including screen-based activities) that compete for attention expands greatly. In addition, note that preferences may not necessarily translate into participation, as indicated by recent data suggesting high drop-out rates in sport and physical activities during adolescence (Basterfield et al., 2015).
Study Strengths and Limitations
One strength of this study is the relatively large sample size, but a limitation is that participants self-selected into the study, and it is possible those with higher participation diversity may have been more likely to take part. Moreover, compared with the number of children invited to participate, the response rate was low, potentially limiting generalizability because although these children were demographically similar to other children, it is not clear why they decided to participate while others did not. Another strength is the use of outcome measures with strong psychometric properties, although these measures do rely on retrospective recall over the previous 4 mo.
Our data are limited because they only show an association between variables and cannot infer causality. A potential source of bias, for example, is that the SEIFA scores (a measure of advantage and disadvantage of the area where the participants lived) was approximately 0.5 standard deviation higher than would be expected in a normative sample. The average SEIFA score of the participants indicated that the areas where they lived had greater socioeconomic and resource advantage. Research has shown that socioeconomic status is an important determinant of participation in active physical and skill-based activities in children with disabilities (Shields, Synnot, & Kearns, 2015). Youth in disadvantaged communities may not have the same opportunities to participate in these types of activities because of limited financial resources, transportation, and family capacity (e.g., parents placing a lower priority on extracurricular leisure activities), all of which can result in less diverse participation. Therefore, our data need to be considered within this context.
Another limitation is that although an attempt was made to exclude children with disabilities from the cohort by identifying whether a child had an education support teacher or not, it is possible that some children with disabilities did not have this support and were included in the final sample. It is also important to acknowledge the potential for the different data collection methods to act as a confounder. Some children, younger children in particular, may require support from a parent or other family member to complete the CAPE/PAC. The extent to which a child’s responses may have been influenced by the presence of a parent is unknown. For example, the parent might have reminded the child about instances of participation that the child did not remember independently, or the child may have felt compelled to exaggerate his or her preference for an activity because a parent was present. However, 62% of participants completed the questionnaires by themselves without the support of a parent or other person; therefore, we expect the level of bias introduced to participation diversity and preference data is likely small.
A further limitation is that only a small number of predictive factors were included in the regression models and there may be other important personal (e.g., activity competence, self-efficacy) and environmental (e.g., qualities and experiences of the activity settings) factors that were not included (Skille & Øterås, 2011). In some cases, only a very small unique contribution to prediction of participation diversity across activity types was made (despite statistical significance). For example, less than 1% of the total variance of participation diversity in social activities was uniquely explained by gender, and less than 1% of the total variance in self-improvement activities was uniquely explained by socioeconomic and resource advantage. Therefore, we urge caution against overinterpreting the strength of relationships among variables.
It was beyond the scope of this study to determine every possible factor that might influence the participation of children in activities outside of school. Future studies might consider investigating the process of developing activity interests as a potential precursor to participation preference and diversity and how this process influences future participation. For example, does exploring what sparks a young person’s interest through thoughtful reflection, interest inventories, or actually trying out new activities help him or her discover a passion or preference for an activity, which might lead to participation? In addition, is reflection on the meaning of the activity after participation helpful to determine whether the young person would like to continue to participate in that activity? Future studies might consider using qualitative methods to explore the perceptions of youth on what influences their participation and better understand the process of how youth are educated about, exposed to, and develop sustained hobbies and interests outside school.
Implications for Occupational Therapy Practice
The results of this study have the following implications for occupational therapy practice:
Preference in leisure activity participation is important for all children, and the profession’s client-centered approach is supported by evidence in the literature (Townsend & Polatajko, 2013).
Supporting children to develop long-term preferences for balanced, healthy lifestyles to prevent development of long-term chronic health conditions is a critical public health need internationally, particularly in developed countries such as Australia where a 20% increase in the number of diabetes cases is expected by 2030 (Shaw, Sicree, & Zimmet, 2010). Health and education professionals, occupational therapy practitioners included, have a role to play in championing this approach.
The importance of child preference in determining participation in outside school activities is supported by study data.
It cannot be assumed the socioeconomic resources available to a family are a key consideration when aiming to support children and adolescents to develop healthy lifestyles, although this finding requires further investigation.
Conclusion
The CAPE and PAC questionnaires have proven to be valid and reliable tools for charting developmental change in participation and activity preferences. This study began to explore the influence of personal and demographic factors on these outcomes. We showed that activity preference maps predictively to the diversity of participation in activities outside school. Moreover, gender, age, and school type were found to influence preference, but their influence varied across activity types. These data could help shape health promotion policy on participation among children through targeting these specific attributes known to affect their activity choice and can also be used to compare with other cohorts of children, including children with disabilities.
The challenge now is to build a more complete causal model of participation, one where pathways between individual, environmental, and experiential factors are clearly mapped. The promise is that large cohort longitudinal and cross-sectional studies can be conducted, using these measures, to expand current theory. Further studies, including intervention trials, will also assist health professionals, educators, and policy makers to plan and implement programs that support the sustained involvement of children and adolescents in health-promoting activities.
Footnotes
Acknowledgments
This study received funding from a La Trobe University Faculty of Health Sciences grant. The authors have indicated that they have no conflicts of interest relevant to this article to disclose.