Background Engaging in sport has been suggested to be mitigate against psychological distress. Using a prospective-longitudinal design, we examine whether sport participation is associated with reduced psychological difficulties in adolescents and whether associations differ by sport and personal characteristics.
Methods 17 550 adolescents from grade 7 to 10 (mean age=14.4 years old) self-reported on sport participation, depressive symptoms, social anxiety symptoms and loneliness in Spring 2007 and 1 year later. Analyses tested the association between sport participation and subsequent psychological difficulties using linear regression, while adjusting for pre-existing psychological and sociodemographic characteristics. Analyses also verified whether associations differed by sport frequency (weekly/more than weekly) and type (team/individual) using Wald tests of parameter constraints, as well as sex, age and pre-existing psychological difficulties using interaction terms.
Results After adjusting for potential confounders, sport participation predicted lower social anxiety symptoms (beta=−0.02 (−0.04, −0.01)) and loneliness (beta=−0.04 (−0.05, −0.02)), but not depressive symptoms. Moderation analyses showed that sport participation predicted greater reductions in depressive symptoms, social anxiety symptoms and loneliness in adolescents with higher baseline scores on each outcome. Associations did not differ by sport frequency and type, sex and age. Post hoc analyses revealed bidirectional associations between sport participation and social anxiety symptoms and loneliness.
Conclusions Sport participation predicts small reductions in adolescent psychological distress. Reductions seem most important in youth experiencing greater pre-existing psychological difficulty. Our findings suggest that sport participation represents a protective factor that supports psychological resilience in at-risk youth.
- psychological adjustment
- social anxiety
Statistics from Altmetric.com
Getting out there and moving around in the context of playing a sport may provide developmental benefits to children and adolescents.1 Youthful participation is associated with better physical outcomes in adulthood, such as improved fitness and an active lifestyle.2 Some literature suggests potential benefits of sport participation on educational and occupational achievement,3 positive social development4 and lower risky sexual behaviour and drug use.5 Less positive outcomes have also been reported (eg, alcohol use in some athletes).6
It could be argued that sport participation represents a protective factor against psychological distress in adolescence.7 Such a protective influence could involve two potential mechanisms. A first possibility is that sport participation might reduce depressive and anxious symptoms by means of the antidepressant and anxiolytic neurobiological effects of physical activity (eg, improved hypothalamic-pituitary-adrenal (HPA) axis functioning, increased monoamine neurotransmission).8 A second possibility is that sport participation might reduce distress through psychological and social mechanisms. These mechanisms have been discussed in the positive youth development (PYD) model.9 10 This model posits that sport helps youth actualise their individual strengths and potential through opportunities to develop life skills (eg, initiative, teamwork, self-control) and supportive relationships with and exposure to prosocial peers and adults. Positive relational experiences with others are believed to decrease the risk of symptoms associated with psychological distress, such as sadness, and social anxiety and isolation.1
Consistent with the expectation that sport diminishes emotional impairment, some literature suggests associations between adolescent sport participation and reduced symptoms of depression,11–13 social anxiety14 15 and loneliness.3 16 Nevertheless, much of it has been cross-sectional.1 11 13 17 Longitudinal studies vary in their effort to control for pre-existing factors.3 12 14 15 Consequently, sport participation is often not adequately isolated as a predictor. This makes it difficult to determine whether positive outcomes reflect a continuation of a pre-existing bias in more physically active youth or a net outcome of engaging in sport. Additional longitudinal research is necessary to demonstrate that sport participation independently predicts psychological adjustment over time.
PYD theory emphasises that the outcomes of sport participation depend on the characteristics of the participation itself.1 10 18 Specifically, benefits may be enhanced when sport participation is more frequent19 because developing cognitive and interpersonal skills related to the sport requires time and effort. Team sport participation may also produce greater psychological gains than individual sport because teams offer richer opportunities to practise social and relationship building skills.1 The current evidence does suggest benefits of more frequent participation13 17 20 and team-oriented14 20–23 activities. However, longitudinal findings remain limited and mixed. For example, some studies suggest no benefits of sport frequency,14 no benefits team versus individual sport,24 or reductions in social anxiety in response to individual sport compared with team sport.15
Finally, the outcomes of sport participation may possibly vary as a function of personal characteristics.25 Sociodemographic characteristics could play a moderating role. For instance, sport participation may be especially protective for girls by countering their greater reactivity to stress or tendency to use more passive emotion regulation strategies.17 Protective effects could also be stronger in younger adolescents who face stress related to the contingency of multiple transitions (puberty, moving to middle school, and so on).26 Considering the well-established therapeutic benefits of physical activity interventions on mental health and behaviour,27 sport participation could provide special benefits in adolescents who have pre-existing psychological difficulties. Very few longitudinal studies explored the potential moderating role of personal characteristics other than sex, which has not emerged as an important moderator in previous studies.11 13 17
Using a large longitudinal sample of adolescents, this study examines the psychological outcomes of sport participation over time. Our first objective is to determine whether sport participation independently predicts lower depressive symptoms, social anxiety symptoms and loneliness the following year. These outcomes have been highlighted in previous research1 3 13 14 and represent some of the most common psychological difficulties in adolescents. Our hypothesis is that sport participation will be associated with reduced psychological difficulties on all three outcomes the following year. Our second objective is to examine whether associations vary as a function of sport frequency and type (individual/team), as well as adolescent characteristics (sex, age, pre-existing psychological difficulties). We expect that frequent participation in sport and team sport will be associated with greater reductions in psychological difficulties the following year. We offer no directional hypothesis regarding the moderating role of personal characteristics given limited or mixed findings in previous research.
This study relies on secondary data from high school students who took part in the evaluation of the New Approaches New Solutions (NANS) dropout prevention programme.28 Participants were sampled from 71 schools in the province of Quebec (Canada). Three quarters of these schools were exposed to NANS and were selected to be representative of all schools located in disadvantaged areas of Quebec in terms of size, language (ie, English and French) and geographical location using a stratified random sampling strategy. Disadvantaged areas were defined as lowest three deciles on a provincial index based on the prevalence of low maternal educational attainment and parent unemployment in school neighbourhoods.29 The remaining schools were comparison schools that were not exposed to the programme and were located in areas of slightly higher socioeconomic status (ie, decile 5, 6 or 7). Written informed consent was obtained from all study participants (77% of eligible participants).
Participants in this study include 17 550 adolescents in grades 7–10 (mean age=14.4, SD=1.3; female=54%) in the 56 NANS French-speaking school who provided information on sport participation. Sociodemographic characteristics as reported by participants are presented in table 1. In comparison, the general Quebec population has lower rates of visible minority status (9%),30 parental unemployment (men=8.5%, women=6.6%)31 and low parental education (13%),32 but higher rates of single-parent families (25%).33 Participants completed self-reported questionnaires in Spring 2007 (Wave 1) and were followed up 1 year later in Spring 2008 (Wave 2). Questionnaires were administered in class by trained research assistants. Sixty-two per cent of participants provided information at follow-up. Participants who dropped out of the study tended to have higher pre-existing psychological difficulties than participants who did not (online supplementary appendix A). Missing data on potential confounders ranged from 0% to 18%.
Supplementary file 1
Predictor: sport participation (Wave 1)
Participants were asked, ‘Do you regularly take part in an organized sport?’ with the following answer choices: no, I don’t participate in any sport; yes, once a week; yes, twice a week; yes, three times a week; yes, four or more times a week. We dichotomised answers into any sport participation (=1) and no sport participation (=0). This assessment did not delineate sport from physical activity and exercise. We coded sport frequency by dividing sport participation into two dummy variables, weekly and more-than-weekly sport participation, that were compared with no sport participation. Students who reported sport participation were asked whether they played in a team (involving two or more people) or individual sport. These categories were used as dummy variables and compared with no sport participation.
Outcomes: psychological difficulties (Wave 2)
We assessed depressive symptoms using the French version of the Center for Epidemiological Studies-Depression (CES-D) questionnaire,34 which includes items investigating how a person felt during the past week. The scale includes items such as ‘I felt sad’ each answered on a 4-point scale (0=rarely or none of the time to 3=most or all of the time). Items were summed (min=0; max=60). Internal consistency in the sample was good (α=0.88). We assessed social anxiety symptoms by averaging six items from the social anxiety subscale from the Spence Children’s Anxiety Scale (SCAS) (α=0.82).35 The scale includes items such as ‘I feel afraid if I have to talk in front of my class’ answered on a 4-point scale (0=never to 3=always). We assessed loneliness using a 5-item scale from the Questionnaire on the Social and Personal Adjustment of Quebec Adolescents (MASPAQ) questionnaire36 including items such as ‘I feel lonely at school’ (1=not really true to 4=certainly true). Internal consistency was good (α=0.84). Items were averaged.
Baseline psychological difficulties (Wave 1)
To control for pre-existing psychological difficulties, we adjusted for depressive symptoms, social anxiety symptoms and loneliness measured at time 1 (T1) using the same scales (CES-D, SCAS, MASPAQ).
Other potential confounders (Wave 1)
We also controlled for the following sociodemographic indicators: sex, age, ethnicity (0=Canadian-born Caucasian; 1=minority) and cumulative index of family adversity (low maternal occupational prestige, low paternal occupational prestige, low family wealth, low home educational resources, parental separation, mother secondary school non-completion, father secondary school non-completion, sibling school non-completion, frequent home move-outs) (min=0; max=9).
We conducted analyses using Mplus V.7.1.37 We first examined the association between sport participation and subsequent psychological difficulties using unadjusted and adjusted regressions. We tested whether associations differed as a function of sport characteristics (frequency, type) using Wald tests of parameter constraints. We compared (1) models in which associations were constrained to be equal between two characteristics (eg, team vs individual sport participation) with (2) models in which associations were allowed to differ freely. We considered that a difference existed if models with free associations were statistically better than models with equal parameters. We probed moderation by sex, age and pre-existing psychological difficulties by testing the significance of interaction terms (eg, sport participation × sex) added to previous models. Assumptions of regression models were tested and met. We took missing data into account using full information maximum likelihood estimation.38 This strategy allowed to preserve the complete sample and correct the risk of attrition bias using available information at T1. Since participants were nested in schools as a consequence of the original study design, we corrected for the non-independence of observations using Mplus ‘Type=Complex’. Standardised regression coefficients were used as measures of effect size and interpreted using Cohen’s rules (0.1=small; 0.3=medium; 0.5=large).
Table 2 reports descriptive statistics for the main study variables. More than two-thirds of students in the sample reported participating in organised sport. Rates were lower in older than younger adolescents. Rates were significantly higher in boys (76%) than girls (63%) (p<0.001). Students predominantly participated in team-based and frequent (more-than-weekly) sport.
Associations between sport participation and subsequent psychological difficulties
We first examined unadjusted associations between sport participation and psychological difficulties the following year (table 3). Sport participation modestly predicted lower depressive symptoms (beta=−0.05), social anxiety symptoms (beta=−0.09) and loneliness (beta=−0.04). We then adjusted for the baseline value of each outcome. This reduced, but did not eliminate associations between sport participation and subsequent outcomes. Finally, we tested models adjusting for baseline outcome values as well as sociodemographic confounders. Sport participation remained predictive of lower social anxiety symptoms and loneliness, but not depressive symptoms. Effect sizes for associations with social anxiety symptoms (beta=−0.02) and loneliness (beta=−0.04) were small.
Variations by sport and personal characteristics
Next, we investigated whether associations between sport participation and subsequent psychological difficulties differed as a function of sport and personal characteristics. Wald tests of parameter constraints indicated no difference in associations between (1) weekly versus more-than-weekly sport and (2) team versus individual sport. Moderation tests using interaction terms also indicated no difference in associations as a function of sex and age. However, we found interactions between sport participation and baseline score for depressive symptoms (beta=−0.04, p<0.05), social anxiety symptoms (beta=−0.07, p<0.001) and loneliness (beta=−0.08, p<0.01). As shown in figure 1, the association between sport participation and subsequent reduced psychological difficulties was stronger in participants who had higher depressive symptoms, social anxiety symptoms and loneliness at baseline. Sport participation only predicted lower depressive symptoms in participants who had scores of 25 or higher on the CES-D at baseline, as well as lower social anxiety symptoms and loneliness in participants who had average or higher social anxiety symptoms and loneliness at baseline (respectively). Intriguingly, sport participation predicted increased social anxiety symptoms in participants who had low social anxiety symptoms (1 SD below the mean) at baseline.
In post hoc analyses, all main effect and interaction models have the frequency of sport participation as a continuous predictor rather than sport participation as a dichotomous predictor. Results from these analyses were similar, suggesting that study findings were not substantially affected by the coding of the sport participation measure. Given that sport participation is dynamic, we also examined the stability of sport participation between Wave 1 and Wave 2. A vast majority (79%) of adolescents who reported participating in sport at Wave 1 reported participation at Wave 2 and most adolescents (80%) who reported no participation at Wave 1 also reported no participation at Wave 2. Sporting behaviour was thus stable for most participants over time. We also examined a reverse model, with outcome variables at Wave 1 predicting sport participation at Wave 2. Social anxiety symptoms (OR (1 SD unit)=0.95, 95% CI 0.90 to 0.99) and loneliness (OR (1 SD unit)=0.94, 95% CI 0.90 to 0.99) predicted decreased odds of sport participation at Wave 2 beyond previous sport participation and controls. However, depressive symptoms (OR (1 SD unit)=0.96, 95% CI 0.90 to 1.03) did not.
In this study, we prospectively examined the potential protective psychological effect of sport participation in secondary school students. Adolescent participation in sport seemed to have benefits by lessening symptoms of social anxiety and loneliness the following year, even after accounting for pre-existing psychological adjustment and sociodemographic characteristics. Specifically, 1 year later, adolescents who participated in sport tended to report less fear of having to talk in front of classmates, less worry regarding what their classmates thought of them and less isolation in school. This provides additional prospective evidence that sport participation may have some positive contribution to the long-term psychological development of adolescents.1 3 12 14 Associations were small, which may be expected from a self-reported measure of any sport participation. Interestingly, we also found social anxiety symptoms and loneliness to predict subsequent sport participation. This suggests a ‘mutual benefits’ model, with lower social anxiety symptoms and loneliness facilitating involvement in sport, which then favours further reductions in social anxiety symptoms and loneliness over time.
Sport participation was predictive of lower depressive symptoms but in contrast to our hypothesis, this association did not remain significant after adjusting for confounders. We also found no evidence of reverse causality, as depressive symptoms did not predict sport participation in post hoc analyses. Taken together, these results suggest that the unadjusted association between sport participation and depressive symptoms likely reflected pre-existing adolescent characteristics (ie, third variables) that were associated with both lower sport participation and increased depressive symptoms. The level of control for confounders may be a reason why some studies found sport participation to predict depressive symptoms,12 while others did not.17
A key contribution of this study was to show that sport participation predicted greater reductions in psychological difficulties in students who had pre-existing psychological difficulties. Sport participation primarily predicted reduced depressive symptoms in students with high pre-existing depressive symptoms, social anxiety symptoms and loneliness at baseline. This consistent pattern of results suggests that sport participation may foster psychological resilience in students with pre-existing psychological vulnerabilities.22 39 This resilience effect may involve direct compensatory effects of physical activity on neurobiological risk mechanisms related to internalising symptoms, such as impaired HPA or monoaminergic functioning.8 The PYD model also suggests that sport participation may compensate for the social difficulties often experienced by adolescents who are depressed, socially anxious and/or lonely by providing opportunities to develop social skills and supportive relationships.9 10 Intriguingly, sport participation was associated with an increase in social anxiety symptoms in individuals with low symptoms at baseline. This may reflect a normative increase in performance anxiety or social desirability in relation to sport participation.
In contrast to our hypothesis, our study did not find the outcomes of sport participation to vary as a function of sociodemographic characteristics or sport frequency and type. Our results do not support increased benefits of more frequent13 17 20 and team14 20–23 sport. The quality of the sport context, as determined by PYD features such as physical and psychological safety, supportive relationships or positive social norms, is likely to be a more important determinant of psychological benefits than sport frequency and type.40 However, we acknowledge that our measures may not have provided enough contrast to capture meaningful distinctions, given that students were not asked to report non-regular participation (less than once weekly) and participation in multiple sports, including both individual and team sports.
Our investigation had several strengths including a prospective design, very large sample, examination of multiple psychological outcomes and adjustment for potential confounding. Several limitations should be noted. We did not control for some confounders, including fitness and body mass index. These data were self-reported. Moreover, we asked participants about the primary sport in which they participated and did not distinguish sport participation from other forms of physical activity and exercise. Isolating the contribution of sport remains a difficult task, giving the competing influence of having an active lifestyle (ie, active transportation, household chores, and so on). Given that most participants in our sample lived in a disadvantaged area, our results may not generalise to the general population. For instance, limited resources may reduce the potential of sport on positive development in disadvantaged areas, although some schools may use sport as a strategy to favour social inclusion and psychological adjustment.7 Data were also collected more than 10 years ago and may not be representative of today’s more digitalised and sedentary youth.
Future research that identifies the conditions associated with better psychological outcomes of sport participation versus other active forms of physical activity in adolescents is warranted. The best evidence will come from experimental or well-controlled longitudinal studies that consider multiple dimensions simultaneously, including general sport characteristics (intensity/frequency, type, and so on),1 PYD features (safety, structure, and so on)40 and adolescent characteristics (eg, level of motivation, pre-existing difficulties).25 Such examinations will have the best potential to clarify mixed findings from previous studies.
From a public health perspective, sport participation has the potential to be a positive, non-stigmatising and engaging approach to promote psychological well-being, especially if it is enhanced with strategic programming to support PYD.10 Our study further suggests that benefits of sport participation have the most clinically meaningful benefit for adolescents who have pre-existing psychological difficulties. Sport involvement may be promising as a strategy to promote psychological resilience in this subgroup of at-risk adolescents.39
What is already known on this subject
Sport participation is associated with better psychological adjustment in adolescents.
Relatively few prospective studies have tested the directionality of this association.
What this study adds
Our study shows that adolescent sport participation predicts small reductions in social anxiety and loneliness 1 year later beyond sociodemographic characteristics and pre-existing psychological difficulties.
Sport participation predicts larger reductions in depressive symptoms, social anxiety, and loneliness in adolescents who have high pre-existing psychological difficulties.
This suggests that sport participation may foster psychological resilience in these adolescents.
Contributors FNB conceived the study, analysed and interpreted the data, and wrote the first and final drafts of the manuscript. GYS, DGS and MJH helped conceive the study, analyse and interpret the data, and assisted in writing the manuscript. JM and LSP helped conceive the study, analyse and interpret the data, and assisted in revising the manuscript for critical content. MJ organised and managed the collection of data, helped conceive the study, analyse and interpret the data, and assisted in revising the manuscript for critical content.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval University of Montreal Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.