Only my group will do: Evidence that social support protects athletes from burnout when they identify with those who provide it

Perceived availability of social support can reduce symptoms of burnout in athletes. As such, it is important to understand the circumstances under which perceived social support is most effective. Social influences such as strength of social identification with a particular group or identity are believed to play an important role in the provision and effectiveness of social support. Across two studies, we investigate whether social identification in a sport can strengthen the protective association between perceived social support and burnout. In Study 1, athletes completed questionnaires assessing perceptions of social support availability, social identification, and burnout. In Study 2, participants completed the same measures at systematic time points across a six-month timespan. In both studies, participants ’ levels of social identification moderated the association between perceived social support and burnout, whereby there was a stronger negative relationship between social support and burnout when participants reported higher levels of social identification in their sport. These results indicate that social identity may play an integral role in improving the efficacy of social support on burnout.

Sport participation provides a range of benefits throughout adulthood; however, many athletes drop out of sport during their lifespan (Jenkin et al., 2017;Murray et al., 2021).Burnout is one reason that contributes to sport dropout (Isoard-Gautheur et al., 2016).In the context of sport, burnout is a syndrome typically consisting of a reduced sense of accomplishment, a devaluation towards sport, and physical and psychological exhaustion (Raedeke & Smith, 2001).Athletes often experience burnout stemming from high levels of stress associated with the sport context (Goodger et al., 2007).Beyond stress levels, social influences are instrumental in conditioning the extent to which athletes experience burnout.Specifically, high levels of social support have been negatively associated with burnout (Eklund & DeFreese, 2015;Gustafsson et al., 2017;Hartley & Coffee, 2019) and it has been theorized that the potency of social support's impact on burnout will be influenced by the extent to which the provider and recipient of support share a common identity (Hartley et al., 2020).Underpinned by social identity theory, the purpose of this study was to examine whether the association between social support and burnout is conditioned by social identification.
Perceived social support (herein referred to as support) refers to the perception of support available from one's social network such as friends, family, team-mates, and coaches who may provide assistance if needed (Rees & Freeman, 2010).Consistent with the work of Rees and colleagues (Rees & Freeman, 2010), perceptions of support consist of four dimensions: emotional (e.g., making an athlete know they are loved and cared for), esteem (e.g., bolstering and reassuring an athlete's sense of competence), informational (e.g., providing guidance and instruction to an athlete), and tangible support (e.g., providing instrumental assistance, such as the resources and conditions necessary for athletic functioning).These dimensions are often highly intercorrelated and can be aggregated to provide a global measure of perceived social support (Raedeke & Smith, 2004;Shang & Yang, 2021).
Support is often associated with lower levels of burnout in sport (Cresswell, 2009), and may be considered an effective resource for reducing the risk for developing burnout (Eklund & DeFreese, 2015;Gustafsson et al., 2017).This is because support may inhibit athletes' reduced sense of accomplishment and their tendency to devalue sport participation (Hartley & Coffee, 2019).However, the social nature of sport means that the effects of support are likely to vary significantly depending on a range of social influences, such as if the support is associated with a trusted group member, and/or support is provided within normative social conventions for the group (e.g., a hand-tap after points in doubles tennis is commonplace and normative to the setting).The social identity approach (Haslam et al., 2020) offers a theoretical framework to examine how social support may result in positive, negative, or indifferent outcomes in sport (Hartley et al., 2020).
Social identification focuses on the cognitive and evaluative aspects of identification with a group (Doosje et al., 1995).This means that individuals who identify with a group think of themselves as part of the group, and gain self-esteem through group membership (Ellemers et al., 2002).Central to this theory, is that our perceptions of the world are structured by the groups we belong to, to the extent that our social identity with a group can shape how we perceive and appraise events.In the sport context, higher levels of social identification are associated with a range of beneficial outcomes pertinent to burnout, including higher levels of motivation (Murray et al., 2022), commitment (Martin et al., 2018), and lower dropout (Murray & Sabiston, 2022).Beyond sport, there is evidence that social identification with a group negatively relates to burnout (Steffens et al., 2014).As such, it is possible that athletes who identify with their team experience lower levels of burnout compared to athletes who do not identify with their team.However, beyond this main effect of social identification on burnout, recent research indicates that levels of social identification in sport may not directly protect against burnout, instead Glandorf and colleagues posit that social identity may act as a moderator of the relationship between support and burnout (Glandorf et al., 2022;Hartley et al., 2022).
Ones' social identification in sport can impact the way athletes appraise events (Coffee et al., 2020), and can underpin the provision and receipt of social support (Levine et al., 2005;Rees et al., 2015), such that an athlete is more likely to value and trust the support from a fellow in-group member (Hartley et al., 2022).Specifically, social identification can impact individuals' appraisal of social support by offering the support provider and recipient a common framework that can facilitate a more harmonious support exchange process (Haslam et al., 2012).For example, before completing challenging mathematical exercises, individuals are more likely to feel comforted by support when the information comes from an individual with whom they identify (Haslam et al., 2004a).This is because the support recipient and provider share the same understanding of the world, and thus the provider is believed to have the individual's interests and concerns at heart (Nicholson et al., 2011).Evidence that social identity can impact an individual's appraisal within a sport setting was observed by Murray et al. (2020) whereby, an athlete's appraisal of their attributions was significantly more impactful on their collective efficacy when they reported higher levels of identification with their team.This can be exemplified as a support-social identification interaction, whereby an athlete who strongly identifies as a runner will likely value support from a fellow runner more than support from someone who competes in a different sport (e.g., fencing, golf, football, etc.).That is, the underlying social identity as 'runners' may allow the athletes to experience the adaptive consequences of social support.In line with social identity theorizing, social identification may provide individuals with a common ground for effective social support to occur (Hartley et al., 2022;Haslam et al., 2012), whereby if an athlete identifies with his coach, technical advice on his serve (i.e., informational support) may be better received and interpreted compared to if he does not share an identity with the coach.However, to the best of our knowledge, the research on this topic in sport, and psychology more generally, has been largely exploratory and cross-sectional.As such, there is a need to examine the nature of these relationships more systematically and rigorously.
Across both cross-sectional and longitudinal studies, our research aimed to examine the nature of the relationship between support and burnout at high and low levels of social identification in sport.After controlling for the effects of stress on burnout, it was hypothesized that (1) support would be negatively correlated with burnout, (2) social identification would be negatively correlated with burnout and that (3) social identification would moderate the relationship between perceived support and burnout, whereby, the negative relationship between support and burnout would be stronger at higher levels of social identification compared to lower levels of social identification.In Study 1 we tested these hypotheses cross-sectionally, and in Study 2 we tested these hypotheses at three points over the course of a 6-month timespan.

Participants and procedure
Power analysis conducted using the pwr package in R indicated a sample of at least 434 would be sufficient to detect a small effect size of 0.04 with 6 parameters, power set to 0.90 and a significance threshold of 0.05.After obtaining ethical approval from the General University Ethics Panel at the University of Stirling, participants were recruited opportunistically and through snow-ball sampling with the use of recruitment posters and online announcements.The sample consisted of 444 athletes (278 male) ranging in age from 14 to 68 years (M age = 25.89years, SD = 10.32),partaking in a range of 45 different sports, including 25 individual sports.Predominant sports were tennis (n = 23), rugby (n = 28), and golf (n = 71).The competitive levels of participants ranged from recreational (n = 66), club (n = 126), regional (n = 97), national (n = 106), to international standard (n = 49).After informed consent had been provided, participants were either provided with an email link to the questionnaire or completed a paper-and-pencil version.

Measures
Perceived support.The 16-item Perceived Available Support in Sport Questionnaire (the PASS-Q; Freeman et al., 2011) was used to assess perceived support.The PASS-Q has demonstrated good reliability and validity indices across independent samples (Boat & Taylor, 2015;Freeman et al., 2011).The stem for the PASS-Q is: "Please indicate to what extent you have these types of support available to you".Participants were required to respond on a 5-point Likert scale, ranging from 0 (not at all) to 4 (extremely).The PASS-Q has been used as a global and dimensional (emotional, esteem, informational, and tangible) measure of perceived social support (Hartley & Coffee, 2019;Shang & Yang, 2021).Because there were no specific hypotheses pertinent to dimensional associations, we examined perceived social support as a global measure.In the current study the PASSQ demonstrated strong internal consistency (Cronbach's α = .93).

Social identification.
Given there were no a-priori reasons to expect different effects on the individual components of social identification (self-investment, solidarity, satisfaction, and centrality: Doosje et al., 1995), social identification was assessed using the recommended Four-Item measure of Social Identification (FISI; Postmes et al., 2013).The FISI is an adaptation of the scale reported by Doosje et al. (1995), and shows good cross-sectional and longitudinal internal-reliability, and correlates highly with each of the components in Leach and colleagues' (2008) 10-item social identification scale (Reysen et al., 2013).Further, the FISI is aligned with the cognitive evaluative aspects of group identification (Doosje et al., 1995) and therefore aligns well with this theoretical framework.The reference group for social identification was specified on the FISI as: "others in my sport" (e.g."I identify with others in my sport"; item 1) and "a member of my sport" (e.g."I see myself as a member of my sport"; item 2).Participants were required to indicate their level of agreement to four statements by responding to a 7-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree).Item responses were averaged to create a single score of social identification.The FISI demonstrated strong internal consistency in the current sample (Cronbach's α = .90).
Burnout.Athlete burnout was assessed using the 15-item Athlete Burnout Questionnaire (ABQ; Raedeke & Smith, 2001), which has demonstrated good construct and structural validity in independent samples (Cresswell & Eklund, 2006;Raedeke & Smith, 2009).The stem for the ABQ is: "Please indicate the extent to which you are currently R.M. Murray et al. experiencing each feeling".Participants were required to respond on a 5-point Likert scale, ranging from 1 (almost never) to 5 (almost always).In line with previous research (DeFreese & Smith, 2013) athlete burnout was conceptualized as the composite score of three subdimensions (reduced sense of accomplishment, a devaluation towards sport, and physical and psychological exhaustion).The ABQ demonstrated strong internal consistency in the current sample (Cronbach's α = .90).

Covariates.
Participants were asked to report their age and their sex before completing the study questionnaires.Participants were also asked to indicate the degree of stress experienced, measured using four sources of stress commonly drawn upon within the sport literature (Freeman & Rees, 2010;Hartley & Coffee, 2019).The stem for each item was: "Please indicate how stressed you felt as a result of the following situations over the past two weeks".Participants were asked to respond on a 5-point Likert scale ranging from 1 (not at all) to 5 (a lot).Item responses were summed to create a total score of stress.Cronbach's alpha for the stress scale demonstrated good internal reliability (α = .81).These variables were included as covariates in the analyses.

Analyses
R version 4.2.0 was used for all analyses.Descriptive statistics were examined through means, standard deviations, and Pearson correlation coefficients (Table 1).Three linear models were conducted to examine whether burnout relates to covariates, social support, social identification, and the interaction between social support and social identification.The first model included covariates (age, sex, and stress) as predictors of burnout.The second model included the covariates and the main effects of social support and social identification.Finally, the third model included the predictors in the first two models and the interaction between social support and social identification.To understand the nature of significant interactions, follow up simple slope analyses were conducted (Robinson et al., 2013)

Descriptive results
Descriptive statistics are reported in Table 1.Overall, older participants typically reported higher levels of stress and lower levels of social support.Females reported higher levels of stress compared to males, but no sex differences between support, identification, or burnout were observed.
Social support and social identification explained an addition 20% of the variance in burnout R 2 = .30.In the final model, levels of social identification significantly moderated the relationship between support and burnout (B = -.08,SE = .03,p = .02),whereby at lower levels of social identification there was no significant association between support and burnout (B = -.07,SE = .05,p = .15).However, at higher levels of social identification, there was a significant negative association between support and burnout (B = -.25,SE = .06,p < .01).The interaction term explained an additional 1% of the variance in burnout R 2 = .31.Models are detailed in Table 2 and the interaction is illustrated in Fig. 1.A sensitivity analysis examining reversed associations indicated no significant moderation between support and burnout on social identity (B = .002,SE = .09,p = .98).

Study 1 discussion
The results of Study 1 indicate that social support and social identification relate to burnout in sport.These results are consistent with previous studies which suggest that support is an integral part of reducing/preventing burnout in sport (e.g., DeFreese & Smith, 2013).Further, the current results indicate social identification may play an important role in this relationship, whereby athletes who identified strongly with individuals from their sport were more likely to be protected from burnout when they perceived higher levels of social support.However, this study examined these associations at one point in time.Within Study 2, we examined whether these associations are stable across a six-month timespan.

Participants
Power analyses were conducted based on the effect size observed in Study 1 (R 2 = .31)using the sjstats package in R (Ludecke, 2019).With the power set to 0.9 and an alpha .05,analyses indicated 483 observations across three waves would detect a medium effect size of 0.31.The final sample consisted of 320 athletes (145 male) ranging in age from 18 to 68 years (M age = 25.78 years, SD = 10.67),partaking in a range of 39 different sports, including 25 individual sports.Predominant sports were rugby (n = 81), netball (n = 93), and basketball (n = 51).The competitive levels of participants ranged from recreational (n = 39), club (n = 118), regional (n = 60), national (n = 53), to international standard (n = 50).Participants completed measures at three measurement waves across a six-month timespan.Of the 320 athletes who participated in the study, 61 completed all three measurement waves, 74 completed two waves, and 185 completed one wave.In total, there were 516 observations, however, 13 individuals did not provide complete information, leaving an analytic sample of 499 observations across 307 individuals.

Measures
The measures used to examine stress, social support, social identification, and levels of burnout in Study 1 were also used in Study 2.

Procedure
After obtaining ethical approval from the General University Ethics Panel at the University of Stirling, participants were recruited opportunistically through advertisements on social media and through direct correspondence with local sport teams.After informed consent had been provided, participants were either provided with an email link to the questionnaire or completed a paper-and-pencil version containing the scales described above at three equally spaced measurement waves (each approximately two months apart).Questionnaires were matched using unique participant response numbers.

Analyses
Descriptive statistics and Pearson correlations are displayed in Table 3.To examine whether social identification moderated the association between social support and burnout across time, multilevel models were conducted examining associations at three waves across a six-month timespan.To account for athletes completing measures at multiple occasions, participant ID was included as a random effect with random intercepts and fixed slopes.Covariates and predictor variables were included as fixed effects.As in Study 1, our analyses in Study 2 consisted of three models, examining covariates, then main effects, then the interaction term in separate models.Models examining the main effects were tested with between person scores and within person scores.Within person scores were centered at the individual level, whereby individuals' scores were subtracted from their mean score.

Descriptive results
Pearson correlations at the between and within person level are displayed in Table 3. Overall, older athletes reported lower levels of Note.A significant b-weight indicates the semi-partial correlation is also significant.b represents unstandardized regression weights, CI represents 95% confidence interval Fig. 1.Study 1 interaction between support and social identity on burnout.Social identity was plotted at 1 SD (6.7) above the mean and 1 SD (4.6) below the mean.social support and less burnout compared to younger athletes.There were more males participating in a higher level compared to females, however, there were no other significant sex differences across the study variables.When comparing individuals who completed one, two or all three measurement occasions, an ANOVA indicated no significant differences between the groups on average levels of burnout across the study, F(2, 317) = .766,p = .47.A chi-square also indicated no differences in sex, x 2 (2) = 1.4,p = .50.However, groups significantly differed by age, F(2, 313) = 11.91,p < .01.A follow up post-hoc Tukey test indicated that, compared to individuals who completed only one measurement occasion, individuals who completed 2 or 3 occasions were more likely to be older.As such, the number of measurement occasions in the sample may be biased towards older athletes.However, multilevel modelling is robust towards missing data, and there are minimal differences in the outcome variable (burnout) between individuals who completed one, two, or three measurement occasions.

Main analysis
Analysis of residuals indicated no violations of linearity, homogeneity, or normality assumptions.There was evidence of significant within-athlete variation, with 66% of the variance in burnout occurring between individuals.Analysis of covariates indicated significant a difference in burnout by age (B = -.01,SE = .003,p = .01)and stress (B = .16,SE = .03,p < .01),but not sex (B = .03,SE = .07,p = .62).Covariates accounted for 6% of the variance in the model.Accounting for covariates, in the second model there was a negative association between social support and burnout (B = -.14, SE = .03,p < .01),and between social identification and burnout (B = -.15,SE = .03,p < .01).Support and identification accounted for an additional 14% of the variance in burnout.
Between person interaction: Social identity significantly moderated the association between support and burnout (B = -.04,SE = .02,p = .05,R 2 = .01),with support being more strongly negatively associated with burnout at high levels of social identification (B = -.19,SE = .04,p < .01),compared to at lower levels of social identification (B = -.09,SE = .04,p = .02).Overall, fixed effects in the model accounted for 21% of the variance in burnout across a season.A sensitivity analysis examining reversed associations indicated no significant moderation between support and burnout on social identification (B = .11,SE = .08,p = .14)(See Fig. 2).
Within person interaction effects.The same interaction between support and social identification was observed at the within person level (B = -.26,SE = .14,p = .06).The nature of the interaction was the same as the between person effects.The trend suggests that when individuals' social identification scores were higher than their mean score, higher levels of social support (in relation to their mean support level) related to lower levels of burnout.

Study 2 discussion
These results extend the results of Study 1 as the interaction between support and social identification on burnout was observed over a sixmonth timespan.Further, analysis of within person scores indicated that variation in one's mean score of social support and social identification over several months related to burnout.Overall, these results suggest that individuals looking to reduce burnout in athletes may target increasing social support and social identification strategies over time.

General discussion
The purpose of this research was to examine whether social identification moderated the association between perceptions of social support and burnout in sport.Previous research indicates that support is often associated with levels of burnout (Eklund & DeFreese, 2015;Gustafsson et al., 2017;Hartley & Coffee, 2019).The current results are consistent with this research, whereby, higher levels of support correlated with lower levels of burnout.Moderation analyses indicated that this relationship was stronger when athletes reported higher levels of social identification with others in their sport.Specifically, compared to lower levels of social identification, at higher levels of social identification the negative relationship between support and burnout was stronger.Overall, these results indicate that social support in the sport context may be more effective in reducing athlete burnout if the athlete identifies with others from their sport.
The results of the current studies indicate that social support may reduce the likelihood of athletes experiencing higher levels of burnout.Consistent with previous research, the knowledge that social support is available, if needed, may reduce symptoms of burnout in athletes (DeFreese & Smith, 2013;2014;Hartley & Coffee, 2019).This may be due to support protecting individuals from feeling emotional and physical exhaustion (DeFreese & Smith, 2014) or support reducing the likelihood of experiencing devaluation and a reduced sense of accomplishment (DeFreese & Smith, 2013;Hartley & Coffee, 2019).Further, there was a main effect of social identity on levels of burnout, indicating that athletes who identify with others in their sport are less likely to experience symptoms of burnout.This association may be explained through perceptions of competence capabilities, whereby athletes who identify within their sport are more confident in their abilities (Murray et al., 2020).Furthermore, according to the revised theory of challenge and threat states (Meijen et al., 2020), social influences impact individuals' appraisals of situational demands (e.g., do I have the resources required to meet the demands of the situation?).A strong social identity might enhance perceptions of the availability of coping resources, which can facilitate the likelihood of adapting to stress related outcomes like burnout (Hartley et al., 2020;Rees et al., 2015;Smith, 1986).Beyond symptoms of burnout, these results are consistent with Haslam and colleagues' social cure theorizing, whereby, a sense of belonging within groups enables access to a range of physical and psychological resources which are adaptive for mental health (Haslam et al., 2012;Jetten et al., 2012).
These studies also advance our understanding of social support and burnout by indicating that social support may be more effective when it is coming from someone with whom one shares a social identity.This aligns with social identity theory and supports observations made in qualitative exploratory research (Hartley et al., 2022).Sharing a social identity with others lays a foundation whereby individuals can depersonalise and re-categorize themselves in terms of group membership Fig. 2. Study 2 interaction between support and social identity on burnout.Social identity was plotted at 1 SD (6.9) above the mean and 1 SD (4.8) below the mean.
and, as such, experience the potential adaptive consequences of social group membership (Turner, 1985) such as reduced drop out (Murray & Sabiston, 2022), more prosocial behaviours (Bruner et al., 2014), and better sport performance (Stephen et al., 2023).Specifically, social identification can provide ingroup members with a common point of reference that facilitates the communication and coordination of social support behaviours (Haslam et al., 2004b(Haslam et al., , 2012;;Postmes, 2003).Overall, our findings provide quantitative evidence that social identities provide the basis for effective social support, and thus facilitates adaptive consequences of support on burnout in athletes.
These results have meaningful implications on how coaches and sport leaders facilitate social support within their sport networks, to the extent that social identification may enhance its potential positive outcomes.Specifically, sport leaders might take a social identity approach to leadership by facilitating a sense of social identification in their groups (Stevens et al., 2021).In doing so, this can provide a foundation for supportive behaviours to exert a positive influence on athletes' symptoms of burnout.Existing research indicates that athletes who experience higher levels of social identification with their sport team are less likely to drop out of sport one year later (Murray & Sabiston, 2022), and the current results suggest this may be because the support athletes received reduced symptoms of burnout.As such, within the sport context, facilitating social identification may help improve athletes' sport experiences and limit potential drop out.
Caution is warranted, however, when building social identification within sport, as higher levels of social identification may also have the potential to exacerbate maladaptive consequences of social support.A strong social identification with one's sport can create a heightened importance on adhering to the norms of that sport (Benson et al., 2017;Rees et al., 2015), and the exchange of social support may violate established group norms (Ramaeker & Petrie, 2019).For example, Hartley et al. (2022) found that the exchange of mental health support within a sport team violated the established norms within sport.In this scenario, higher levels of social identification in the presence of social support may have maladaptive consequences, such as increased burnout.Specifically, the common framework that social identities can provide (Haslam et al., 2012) may be superseded by the violation of team norms attached to those identities.Further research is needed to understand how the norms within a sport and sport teams may influence the provision and appraisal of social support among athletes.
These studies used both cross-sectional and longitudinal designs, whereby data for Study 2 were collected across multiple time points, indicating the associations observed in Study 1 were stable across time.However, causality cannot be inferred from these approaches.Specifically constructs such as social identity (Murray et al., 2020) and burnout (Carson et al., 2017) are dynamic in nature with levels susceptible to change across days and weeks.As such, future studies could use research designs such as ecological momentary assessments and experimental designs to examine causality.In addition, because measurement occasions did not occur at each sport's start, mid, and endpoint, we could not reliably examine how associations change across a season.As such, further research is needed to examine how these processes evolve from the start to end of a season.Further, although these studies examined constructs across a variety of ages while including age as a covariate, the nature of these constructs likely change across the lifespan.For example, as people age, their motivations for social contact and development of social relationships change (Carstensen, 1992).Understanding how support and social identification interact within different demographics (e.g., race, age, gender) would further our understanding of burnout.Specifically, this knowledge would allow for more nuanced and targeted strategies aimed at keeping more adults in sport.Finally, the effect size for the interaction term was small, and as such, further research is needed to understand the magnitude of the effect that social identification has on the support-burnout process.
Overall, these studies indicate that the relationship between athletes' perceptions of social support and burnout is moderated by social identification.Not all support is equal.Instead, the providers of support may play an important role as to whether athletes experience adaptive consequences of perceived social support.It is worth noting, however, that in certain environments, social support may also have maladaptive consequences which can be exacerbated through social identity (Hartley et al., 2022;Kellezi & Reicher, 2012).As such, further research is needed to understand the circumstances under which these processes develop.Also, further research is needed to test the causality of these associations over time.If confirmed, strategies to decrease levels of burnout in athletes could be grounded within social identity theory, whereby athletes who identify with their teammates and feel social support from their teammates may be protected from the higher levels of burnout commonly experienced within sport.

Declaration of competing interest
Given their role as Editor in Chief Coffee P., had no involvement in the peer-review of this article and had no access to information regarding its peer-review.All other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Table 1
Study 1 means, standard deviations, and correlations.
* Note.M and SD are used to represent mean and standard deviation, respectively.* indicates p < .05. ** indicates p < .01.R.M. Murray et al.

Table 2
Study 1 and Study 2 Regression results using burnout as the criterion.

Table 3
Study 2 means and standard deviations at each time point, and correlations.