Abstract
The relationship between social capital and sport has been an increasing focus of scholarly literature in recent decades. However, very few of these studies consider social capital alongside theories of cultural consumption. Even fewer seek to assess the place of social capital in sports spectatorship. Taking primarily a Bourdieusian and neo-Bourdieusian theoretical approach, this study seeks to rectify these gaps by analysing three key components of social capital – social network size, social network prestige and social network variety – and how they relate to patterns of sports spectatorship and participation. Results indicate that the type of social capital that is most predictive seems to rely heavily upon the nature of the cleavages between cultural patterns of sports engagement. While the size of social network seems most universally applicable to predict sports engagement generally, network variety also seems to be highly applicable to the most omnivorous engagement profiles. Finally, network prestige appears applicable to some highbrow profiles of sports engagement.
Introduction
Arguments for the importance of social relationships go back to the founders of the sociological discipline. However, since Pierre Bourdieu’s seminal theorizations of social capital (1984, 1986, 1990), there has been an increase in the formal development of a social capital concept. Prominent in these endeavours are the theories of social capital developed by Putnam (1993, 2000), Coleman (1988, 1990) and Lin (2001). All are generally concerned with how elements of the ‘social’, that is human interrelations, shape societies. The social shapes societies in every sphere, from politics, trust and civic engagement to social stratification and the allocation of economic resources.
Of these foundational ways of conceptualizing social capital, this paper adopts a Bourdieusian approach, one that is also informed by, and modelled after, similar approaches of recent scholars (e.g. Savage et al., 2013, 2015). While Bourdieu did not neglect sport in his writings (e.g. 1978), it was not until recently that scholars started to investigate how social capital interacts with the domain of sport. There has thus been increased interest in the previous two decades in how dynamics of social capital manifest themselves and influence the realm of sport (Jarvie, 2003; Jarvie and Thornton, 2012), which has resulted in studies into how different aspects of sport generate specific elements of social capital (Coalter, 2007, 2010; Delaney and Keaney, 2005; Forsell et al., 2020; Graham and Glover, 2014; Nichols et al., 2013; Nicholson and Hoye, 2008; Widdop et al., 2016). While some of these studies have assessed the relationship between sports engagement and social capital, many of these papers have used other theories of social capital than this paper employs, particularly as they relate to the developing and strengthening community bonds and developing cross-social ties. This paper is more concerned with how social capital intersects with inequalities and may manifest itself in different modes of sports engagement.
This paper seeks to contribute to the debates on social capital and sport in a number of ways. First, while direct sports participation is frequently studied (Lera-López et al., 2012; Thrane, 2001), sports spectatorship, such as in person, on television (Perényi, 2010; Taks and Scheerder, 2006; Thrane, 2001) or via another contemporary spectating medium (Gemar, 2019a), has received significantly less scholarly attention (Gemar, 2019b, 2020a; Lera-López et al., 2012). Those that consider both at once are even rarer. This paper considers both sports spectatorship and participation at once. Second, while there is indeed one significant paper to this effect (Widdop et al., 2016), studies assessing elements of social capital as they related to cultural consumption approaches to sports participation remain sparse (Widdop et al., 2016). Along with Bourdieu’s (1984) theories, a primary theory of cultural consumption considered in this paper is that of the cultural ‘omnivore’ (Peterson, 1992). This theory describes those who partake in a larger variety of culture, regardless of relative prestige, but who have been found to be generally associated with higher social position (e.g. Chan and Goldthorpe, 2010; Gemar, 2019c; Peterson and Kern, 1996). It is this kind of cultural variety that Erickson (1996) argues contributes to wider and more varied social networks.
This paper thus uses a different theoretical perspective of social capital (that of Bourdieu) to investigate similar questions of the relationship between social capital and sports participation, while empirically expanding it to include sports spectatorship. I therefore ask the following guiding research questions:
How are sports spectatorship and participation organized in contemporary UK culture?
Is social capital a primary driver of this organization?
If social capital is indeed a primary driver, which elements of social capital are most prominently so?
While prior research into the relationship between social capital and sports participation has shown different elements of social capital associated with increased direct sports participation (Delaney and Keaney, 2005; Widdop et al., 2016), which types of social capital, or a mixture thereof, that influence sports participation are still relatively unknown (Coalter, 2010; Widdop et al., 2016). This paper therefore seeks to contribute to knowledge about the mix of social capital associated with direct sports participation, while also providing novel investigations of social capital for sports spectatorship.
The (Bourdieusian) concept of social capital
The importance placed upon social capital stems from theories regarding its necessity in maintaining communities and civil society (Auld, 2008; Jowell, 2005; Putnam, 2000), and even democracy itself (Widdop et al., 2016). Putnam’s (2000) influential theory of social capital argues that collective social engagement forms social ties which then strengthen the collective conscious and encourage civic engagement and community wellbeing. However, even as there is seemingly strong democratic potential of social connections, social capital is even more so an exclusionary mechanism that benefits society’s already most advantaged persons and groups (Savage et al., 2015). For Bourdieu, social capital was primarily a manifestation and driver of social inequalities and their reproduction. Indeed, there is previous research (e.g. Erickson, 1996; Granovetter, 1995) which finds that those from more advantaged backgrounds and more prestigious occupations have more prestigious and fuller social networks to then get better jobs or be promoted. Regardless of specific theory, the concept of social capital writ large relies upon arguments around investment in the social for a certain reward, benefit or return (Widdop et al., 2016). In Bourdieu’s work, this investment includes the expansion of personal connections to include those who are most beneficial to be known by the individual. Bourdieu (1986: 250–251) explains it this way: Because the social capital accruing from a relationship is that much greater to the extent that the person who is the object of it is richly endowed with capital (mainly social, but also cultural and even economic capital), the possessors of an inherited social capital, symbolized by a great name, are able to transform all circumstantial relationships into lasting connections. They are sought after for their social capital and, because they are well known, are worthy of being known (‘I know [them] well’); they do not need to ‘make the acquaintance’ of all their ‘acquaintances’, they are known to more people than they know, and their work of sociability, when it is exerted, is highly productive.
This is true for both tangible and symbolic rewards. This can happen through close family and/or friend connections (Field, 2017), perhaps where parents pull strings with close friends of the family or hire their children more or less directly. However, social capital is perhaps even more likely to be generated through the ‘strength of weak ties’ (Granovetter, 1973). It is perhaps thus less about a malevolent cabal of established elites who are all very close as it is that those from more advantaged backgrounds and more prestigious occupations are linked widely by weak ties (Savage et al., 2015), or very limited degrees of separation. Granovetter (1973) argues that because those less close to us are less likely to share redundant information, experiences and social networks, these ties are more advantageous to ultimately expanding one’s own network of contacts, information and experiences. Making a related argument, Erickson (1996) argues that social network variety conveys more advantages in the workplace because social network variety can provide greater cultural variety by which one can, in turn, extend one’s social network and culturally participate more efficaciously to one’s advantage. Indeed, she argues that ‘social relationships at work are the fundamental site of class processes in their most direct form’ (Erickson, 1996: 217).
Bourdieu also emphasizes the critical importance of the institutionalized credentials that then circumscribe symbolic status groups, within which one’s network connections may belong.
Social capital is the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition – or in other words, to membership in a group – which provides each of its members with the backing of the collectivity-owned capital, a ‘credential’ which entitles them to credit, in the various sense of the word. (Bourdieu, 1986: 248–249)
A major component of these networks, then, is the tangible or symbolic credential of group membership. The groups within which one can have membership through a type of tangible and/or symbolic credentialization include, for example, those with higher education, those in certain occupations and those who engage in certain cultural activities. Savage et al. (2015: 133) make this kind of argument regarding the aristocracy in Britain in recent centuries when they argue that it has been an example par excellence of a large network of weak ties, within which ‘everyone would know many people with titles, if only by repute’.
Along the lines of this type of argument, sport is an element of culture that is certainly more orientated towards the group and membership within groups. This includes groups of players and groups of fans, along with wider collective solidarity groups of city and country. Some argue that sporting events are meaningful generators of social capital for a locality (Misener and Mason, 2006). Indeed, Widdop et al. (2016: 600) argue that ‘sport is the most popular type of group activity in Britain, and sports organizations do better than other types of organizations in building and sustaining friendships and networks’. However, many sports are more individualized. These included many of those that Bourdieu (1984) insists are associated with elevated social position. Some argue that these individualized sports therefore are much less likely to develop types of social capital (Widdop et al., 2016). However, according to Bourdieu’s more symbolic understanding of group membership, belonging to the symbolic group of ‘runners’, ‘golfers’, ‘tennis players’, or ‘cyclists’, to name a few, may confer upon one the necessary social capital to form weak ties with those partaking in similar sports, even as these sports are generally not done in large groups. The benefit of belonging in certain groups, such as these, is only enhanced if there is an institutionalization to them, such as country clubs for golf, or certain kinds of sports (and non-sports) members’ clubs (Forsell et al., 2020), where many in the membership may likewise partake in such activities. This is also true of spectatorship, where certain pubs in certain areas of town can be known for certain groups of fans, certain areas of stadiums likewise, and where people can host watch parties for those within their neighbourhood or for their work colleagues.
Social capital, patterns of cultural consumption and sports engagement
Bourdieu argued that different forms of capital are inextricably linked with each other. For instance, the relevant rewards that accrue from one’s social network cannot be separated from the economic and cultural capital resources of the individuals in that network (Bourdieu, 1986). Indeed, it is arguably only because of the cultural or economic capital that those connections confer symbolic or material benefits. It is also Bourdieu’s (1984, 1986) theories of cultural capital and how one’s cultural participation, education and fluencies contribute to one’s societal position and play an integral role in intergenerational class stability. This is because the type of culture that one participates in, consumes and has knowledge of illuminates one’s social position and has the power to entrench and reproduce this position. The more socially distinguished the element of culture, the more cultural capital is conferred and accumulated upon the person who can adroitly participate and discuss that culture. It is because of this dynamic that Bourdieu generally argued that it was those elements of traditionally highbrow culture that were most advantageous for conferring and accumulating this capital. This capital can then be transferred into social capital through interactions with those who likewise adroitly participate in these socially elevated forms of culture – not coincidentally generally those of elevated social position – and through these connections accrue a ‘multiplier effect on the capital’ that they possess individually (Bourdieu, 1986: 249). Therefore, the amount of one’s social capital can be assessed by understanding the amount of cultural, economic and symbolic capital aggregated within one’s social network.
However, like social networks, many have observed the advantages of cultural variety. The most influential of these scholars was Richard Peterson. Through empirical evidence in the United States, he argued that the paradigm of cultural capital had shifted, from circumscribed consumption of class-appropriate culture to a more varied and eclectic cultural diet and repertoire (Peterson, 1992; Peterson and Kern, 1996). This expanded cultural engagement and increased repertoire reconfigures cultural capital away from ‘snobbery’ and towards a cultural capital that necessarily includes this eclecticism (Peterson and Kern, 1996). The increased cultural knowledge of those with these patterns of cultural engagement allows a broader range of social connections. This is especially so within business environs, which values highbrow culture less than others (Erickson, 1991), through which they can generate valuable social capital network connections (Erickson, 1996).
Peterson (2005) argued for the necessity of studying these dynamics in every domain of culture, moving beyond the field of music that did and still has dominated much of the scholarly work in these areas. Specifically regarding sport, Erickson (1996) finds sport to be the non-work topic of conversation, and thus the primary area of leisure culture, that work colleagues have in common with each other. In this way sport may play a distinct role in the domain of culture in forming types of social capital networks at work, an area of our lives that constitutes social class (Erickson, 1996). These networks in turn help fill structural holes (Burt, 2000), and help one get a career boost (Erickson, 1996). Likewise, it works in reverse, with Erickson (1996: 277) arguing that ‘(social) network variety is important . . . because contact with different types of people includes contact with different types of culture’. It is in these ways that Bourdieu’s conception of social capital overlaps with his (and others’, including Peterson’s) theories of cultural capital and the importance of cultural activity and cultural knowledge in manifesting, forming and reproducing both social capital and social position.
Hypotheses
I may hypothesize, then, that variety in social networks is associated with increased variety in sports engagement. If this is the case, variables of social network variety would likely see increased predictive results for any omnivorous patterning of spectatorship, participation or both. Conversely, if there is closure within social networks on the basis of prestige and high status (or not), then I would expect to see the predictive force of the social capital variables that impact on sports engagement concentrated in the relative prestige of social connections. If it is simply the size of one’s network that most contributes to the predictive value of social capital in sports spectatorship and participation, then I would expect regression results that support this.
In these ways, this paper tests the different elements of social capital on different forms of sports engagement in the UK. If, as Bourdieu claims, it is about multiplier effects of one’s network, then all three of the social capital variables tested in this study would contribute to such multiplier effects. Therefore, I also might expect all three social capital measures to be significant in some way in identifiable patterns of sports engagement, perhaps to differing degrees.
Data and methods
Survey data
This study uses data from a national online survey (n=1105), which was designed by the author and conducted in conjunction with the prominent survey company Survey Monkey. The survey was distributed to a large representative sample of the UK. The sample was gleaned using quota sampling methods, being balanced (by Survey Monkey using quota oversampling techniques) explicitly by age and gender, with geography and race naturally balancing in a representative manner. The sample’s representativeness can be judged against the broader UK population in Table 1.
Comparison of key demographics for the original survey (OS) and 2011 UK census.*
Percentages rounded to the nearest tenth and may not add up to 100%.
Benefits of online survey techniques include providing increased dynamism to social research. They are able to quickly and easily facilitate research into questions that may not be readily available from other secondary sources. In this way, they facilitate flexibility. For instance, I am able to capture social network resources in sports engagements, which traditional survey instruments do not generally consider. Because the survey companies have established relationships with their representative samples, response rates in these types of online surveys are also generally much higher than traditional methods. While there are arguably drawbacks with respect to the online and non-probabilistic nature of the survey, because the composition of original survey so closely mirrors the census data, and is of requisite size for the population, it suggests that it is a valid sample of the UK population.
Sports spectatorship and participation
To assess both sports spectatorship and participation, the survey used in this paper asked respondents whether they engaged ‘sometimes or often’ in watching sports (whether online, on TV or in person) and direct sports participation. This specific wording was modelled after previous surveys of UK cultural activity (Bennett et al., 2009), and the Great British Class Survey (GBCS) (Savage et al., 2013, 2015). Tables 2 and 3 show the relative frequencies of those responding that they sometimes or often participated in, or watched, the following sports.
Survey frequencies for sports watching variables.
Survey frequencies for direct participation variables.
Social capital measures
To measure social capital, this study uses survey information regarding the occupations of respondents’ social networks. Eighteen occupations were included in the survey, and respondents were asked whether or not they knew people in these occupations ‘socially’. The occupations and question wording included in the survey again mirrored those of the GBCS. All 18 occupations are highly salient for the contemporary UK and come from differing levels of societal prestige. Table 4 presents the 18 occupations. This table also shows their prestige rank in contemporary Britain, according to Savage et al. (2015), and their relative frequency in the survey data.
Survey frequencies for social capital network occupations.
Out of 37 possible rankings for British society, as ranked by Savage et al. (2015).
This study measures social capital in three ways. First, the size of social network is considered. Bourdieu (1986: 249) argues that ‘the volume of the social capital possessed by a given agent thus depends on the size of the network of connections [they] can effectively mobilize and on the volume of the capital (economics, cultural or symbolic) possessed in [their] own right by each of those to whom [they are] connected’. The average number of connections here is seven, and a range between one and 18. The number of connections are put into groups for the analysis of this paper, with 32.3% of respondents having between one and four connections, 35.6% between five and eight, and 32.1% having networks of nine or above.
Second, the average prestige level of the social network is considered according to rankings of British occupations (Savage et al., 2015). This measure shows how likely those social connections are to be ‘sought after’, and the product of this connection to accrue and convert other types of capital (Bourdieu, 1986). Indeed, previous studies have also emphasized that who is included in an individual’s social network is an important factor in generating social capital (Widdop et al., 2016). Those who averaged a connection that was in the top 10 of occupational ranks formed 8% of respondents, 75.8% had an average connection that was between 11 and 20, while 16.2% of respondents averaged a connection in the lowest of the three occupational prestige categories. This large majority in the middle could be the result of a substantial enough group of people having more varied social networks. Indeed, it is partially for this reason that this analysis requires a variable for network variety. This sizeable middle could also reflect the fact that the two most common occupations within people’s social networks, teachers and nurses, also fall within this range. This is especially likely to have a major impact on average scores for those with smaller networks. Because the third social capital measure in this paper, network variety, likely includes many in the middle prestige category, it is perhaps the highest and lowest average prestige groups that are of most interest within this measure. This is especially true when comparing more exclusionary sporting engagement profiles with more exclusionary social networks.
The third measure is social network variety. I here consider one’s network to have full variety if they know at least one person from each of the three tiers of the occupational hierarchy used for this paper. Those who do not know at least one person from each of these three groups do not have full network variety. With this criteria, 55.2% were deemed to have full variety, while 44.8% did not.
This type of measure mirrors the kind of cultural omnivorism discussed earlier. Indeed, Erickson (1996) argues that the variety within social networks is closely linked to cultural variety and cultural capital accumulation because culture (and the ability to discuss it) is a key input and output from these connections. This kind of measure is also useful in a contemporary environment where reflexive exclusionary mechanisms – cutting oneself off from specific types of people – is no longer acceptable and considered rude, or worse (Savage et al., 2015). Therefore, this expansion of one’s social network to include more variety not only forms the types of weak ties (Granovetter, 1973) and fills the types of ‘structural holes’ (Burt, 2000) that then forms social capital, but it also can be itself a type of ‘emerging’ cultural capital (Prieur and Savage, 2013; Savage et al., 2015), or at least avoids the loss of cultural capital that may arise from one’s snobbish exclusion of others.
Methods of analysis
This study first conducts a latent class analysis (LCA) in order to understand the different patterns of sports engagement. LCA represents an effective tool for this type of investigation as it identifies typographical groups. An LCA is done for both direct sports participation and for sports viewership. This type of analysis is also critical for identifying any patterns of omnivorousness in this sports engagement. Once the LCA identifies groups of similar sports engagement, I then perform multinomial logistic regression analyses. This is done to control for the various socio-economic and demographic variables that may affect membership in any of these groups. With the inclusion of social capital variables, I will be able to identify how strong a predictor of sports engagement social capital networks are for both direct sports participation and sports viewership in the UK. The social capital, socio-economic and demographic variables included in the regression model appear in Table 5. Finally, I use chi-square Phi-values to briefly assess the relationship between the spectating and participation LCA groups.
Survey frequencies of information included in the regression analyses of this paper.
Results
Latent class analysis of sports watching
The LCA model fit statistics (see Appendix Table AI) show a four latent class model as the optimal solution for sports watching. This is because both the Bayesian Information Criterion (BIC) and the Consistent Akaike Information Criterion (CAIC) reach their lowest (optimal) levels for this model solution. This is due to these criteria’s principles of maximizing likelihood, while avoiding overfitting of the model. Therefore, there is a type of complexity penalty as parameters increase, which in turn causes the criterion measure to increase. As the criterion measure decreases, the greater the likelihood of the model’s fit to the data. Therefore, this study uses a four class typology for sports watching in the UK.
The composition of each latent class group can be seen in Table 6. As the table shows, the four latent classes divide the sample into groups of 46%, 31%, 13% and 11% of the sample, respectively. The first of these groups is defined by respondents who have comparatively low probabilities of watching any of the sports. Therefore, this group represents the ‘non-sports watching’ portion of the population.
Latent class profile for sports watching variables.
Bold values represent probabilities that exceed the relative frequency of that variable in the sample.
The second latent class group shows probabilities of sport watching that exceed the relative frequencies of the sample for traditional sports of the UK, including sports that are generally considered more ‘lowbrow’ sports for this national context. The second latent class is therefore labelled the traditional/lowbrow sports group. The traditional aspect of this group seems to be the most salient, however.
Latent class three represents the ‘full omnivore’. This is because those in this group have high probabilities for watching all of the sports included. It is the second smallest cluster, with approximately 13% of the sample within this omnivorous group.
The final latent class group shows elevated probabilities to watch a very specific group of sports. This group especially watches athletics, cycling, swimming, tennis and, to a lesser degree, basketball. The sport watching patterns of this, therefore, revolve around specifically Olympic and highbrow sports. This group is thus labelled ‘Olympic/highbrow’.
Regression analysis of sport watching groups
The results of the regression model for sports watching (Table 7) show that the social capital variable for (larger) size of social network is a statistically significant predictor of membership in the traditional/lowbrow sports watching group, compared to the non-watching group. The variable corresponding to the average prestige of those connections is not statistically significant, even as there is non-statistically significant evidence that those with the lowest average network prestige are most likely to be in this group. The importance of these social capital variables, however, is outdone in importance by one variable, gender. Men are much more likely to be in this group compared to the non-watching group than are women. Moderately high incomes (£50–100k) are also similarly important to this group as the size of social network.
Odds ratios from the logistic regressions of each latent class group for sports watching (Non-sports watchers as reference).
p < .05; **p < .01; ***p < .001.
The same three variables of income, social capital (size), and gender are again the three most important for fully omnivorous sports watching compared to non-sport watching. However, this time the size of social network is tied for the most powerful predictor, with those with the highest number of connections many times more likely to be omnivorous sports watchers compared to those with the fewest connections. Omnivores are also more likely to have higher social network variety. Like the traditional/lowbrow group, there is non-statistically significant evidence that those with lower average network prestige are more likely to be in this group, even as this variable is again not statistically significant. Likewise, men appear to be most likely to be in this omnivorous sport watching group compared to women. Finally, the sports watching omnivores appear to have the highest income levels of any of the groups.
The final latent class group generally shows quite a different make-up from the other two sport watching groups. Age is the most important predictor for this group, at least between the oldest and youngest groups. While there is non-statistically significant evidence that this is a more English, whiter and social capital rich group, it is ultimately education (degree) and gender (female) that are the other two significant measures for membership in this final group. There is, however, non-statistically significant evidence that social network size may be positively associated with this group.
Latent class analysis of direct sport participation
While the model fit statistics for participation (Appendix Table AII) suggest that a three class latent class model may be statistically optimal (according to BIC), I choose the second most optimal four class solution for this paper for a number of reasons. First, a four class model significantly reduces the size of the latent class group that is modally considered as non-sports participants. The three class model solution produced an inactive class of 66%, as opposed to the 55% of the four class model. Although all using slightly different metrics, official governmental statistics from England (Audickas, 2017), Scotland (Scottish Government, 2017), Wales (Sport Wales, 2020), and Northern Ireland (Department for Communities, 2019) all report inactive rates below 50%. Likewise, Widdop and Cutts (2013) and Widdop et al.’s (2016) studies report an inactive latent class group of 41% and 42%, respectively. While the increased size of the inactive class in this analysis may reflect a slight decrease in participation in the UK since Widdop and Cutts (2013; using data from 2005–2006) and Widdop et al.’s (2016; using data from 2007–2008) studies, it more likely reflects the very low bar of survey data asking if people participated in sport sometime in the past 12 months used in these studies. It is plausible that respondents interpreted the ‘sometimes or often’ language of the survey used in this study as more frequent than this, thus leading to a larger inactive group. However, because the governmental data is more up to date, and reflects the time period of the survey used in this paper, I argue that the built-in penalty for complexity in the model fit statistics has caused an over simplified attribution of inactive respondents. Therefore, I argue that the four class group allows for a more accurate depiction of direct sports participation in the UK, and is the one chosen for analysis here.
The composition of each of these latent class groups can be seen in Table 8. The first of these groups is defined by respondents who have comparatively low probabilities of participating in any of the sports. Therefore, this group represents the ‘sports inactive’ group.
Latent class profile for sports participation variables.
Bolded values represent probabilities that exceed the relative frequency of that variable in the sample.
The second latent class shows elevated probabilities to participation in what may be termed fitness sports. This group is likely to participate in activities such as aerobics, jogging/running, swimming, walking and weightlifting. This group is also likely to participate in the highbrow sports of equestrian and downhill skiing. I thus label this the ‘fitness/highbrow’ group.
The third latent class group shows inclusion of traditional sports in the UK, including sports that are generally considered more ‘lowbrow’ in this national context. The second latent class is therefore labelled the traditional/lowbrow sports group. Like for sports watching, the traditional aspect of this group seems to be the most salient here.
Latent class four represents the ‘full omnivore’. This is because those in this group have high probabilities for participating in all of the sports included. While the smallest cluster, the fact of a fully omnivorous participation group, is important, as I can now test if this group is of the most elevated social position and increased social capital.
Regression analysis of sport participation groups
The results of the regression analysis of participation (Table 9) show that those with the largest social networks are much more likely to be in this fitness and highbrow sports participation group than those with the smallest social networks, compared to the inactive group. Unlike the sports watching groups, the variable capturing the average relative prestige of the social network is also a statistically significant predictor for membership in this group, with those having a higher average relative prestige of their network more likely to be fitness and highbrow sports participants. Having a university degree or higher is similarly predictive. The most predictive, however, is age, especially between the youngest and oldest age categories, with the former much more likely to be in this group. This result is common for studies of sports participation. There is notably no significant gender difference for this group. Representing a more noticeably middle class demographic, this group mirrors Bourdieu’s assertions (e.g. 1978, 1984, 1986), and others (e.g. Bennett et al., 2009; Gemar, 2020b; Smith-Maguire, 2008), regarding the affinities of these physical pursuits with middle class ideals and the cultural and physical capital that they can accrue.
Odds ratios from the logistic regressions of each latent class group for sports participation (sports inactive group as reference).
p < .05; **p < .01; ***p < .001.
There is a large gender divide, however, for the traditional/lowbrow sports participation group, with men being much more likely to partake in these traditional sports. Those in younger age groups are also more likely to participate in these sports than older age groups, relative to the inactive group. There are no statistically significant predictive parameters concerning either measure of social capital for this third group.
The omnivore group shows strong predictive parameters for social capital, age and gender. The strongest predictor is between the youngest and oldest age categories. However, aside from this age parameter, those with elevated levels of social capital, in the form of social network size, are many times more likely to participate omnivorously in sports than those with smaller networks. There are also non-statistically significant results that suggest the average relative prestige of one’s social network and network variety also play a role in omnivorous participation. Finally, there is a strong gender divide for this group, with men more likely to be in the omnivore group, relative to the inactive group, although the strength of this predictor lags behind social capital and age variables. Also, while there is non-statistically significant evidence that white ethnicities are more likely to be sporting omnivores, relative to the sport inactive group, it is interesting to note that there are no statistically significant predictors of income, race or geographic region for any of these groups.
Relationship between spectatorship and participation
Because I consider both spectatorship and participation in the UK at once, briefly investigating the relationship between the two provides its own opportunity to contribute to existing knowledge of this relationship. It is clear from statistically significant chi-square Phi-values (in parentheses) that there are indeed some relationships between the latent class groups. First, inactive spectators are negatively associated (–0.164) with omnivorous participation. Omnivorous spectators are negatively associated with inactive participation (–0.174), but are positively associated with traditional/lowbrow participation (0.185). Finally, Olympic/highbrow sporting spectators are negatively associated with inactive participation (–0.195), but are positively associated (0.208) with the fitness/highbrow participation profile.
Discussion and conclusion
From the results of the empirical analysis in this paper, it is clear that social capital plays a role in structuring both sports spectatorship and participation in the UK. Although using a different theoretical approach than others have employed, this overarching finding echoes prior studies on social capital in sport. Answers to our three specific research questions also emerge clearly from these results.
First, sports spectatorship and participation in the UK today are similarly patterned. Both are divided along lines of traditional versus less traditional, which sometimes also aligns with team sports verses individual sports and lowbrow versus highbrow sporting genres. There are also groups that do not engage with sport, and those that are omnivorous in their participation. Compared to a previous study of England and Wales using 2005–2006 data (Widdop and Cutts, 2013), my analysis of participation finds a larger group of non-participants, a similarly sized fitness group, a slightly larger traditional/lowbrow group, and a smaller omnivorous group of participants. However, it is ultimately difficult to compare results over time from different survey programs that ask differing questions regarding frequency of participation. In terms of spectatorship in recent decades, I could not find a direct comparative historical study for the UK. However, the relative frequencies of spectatorship found in this study closely mirror the percentages of an Ipsos MORI (2003) poll regarding sporting interest in Britain. Therefore, while mediums of spectatorship proliferated between 2003 and 2016, thus facilitating those with interest greater access, that interest seems to have stayed relatively steady over this time.
There are also notable differences found in this paper regarding the size of the LCA groups for spectators and participants. The non-sports and highbrow/individual sports groups are larger for sports participation, while traditional/lowbrow and omnivore groups are larger for spectatorship. These results make sense through the lens of lower barriers to entry and thus greater general access to sports spectatorship, and the mass spectacle effects for traditional popular sports whereby their appeal is greatest to the largest (that is lower) social classes of society (Bourdieu, 1978).
What about the role of social capital in this patterning? Our second research question asked if social capital was a primary predictor in these patterns of sports engagement. Our third asked which elements of social capital have this distinction, if any. For sports spectatorship, the size of one’s social network was the most powerful capital predictor for watching traditional and lowbrow sports, more predictive than either cultural (in the form of education) or economic (income) capital measures. There was also an inverse relationship between average social network prestige and the watching of these sports. It was again social network size, and network variety, that along with economic capital are significant capital predictors for omnivorous watching of sport in the UK. This result supports Granovetter’s (1973) foundational theory of weak ties, and Erickson’s (1996) arguments that cultural variety is both product and cause of social network variety. These results also may support Erickson’s (1996) assertions that it is particularly knowledge of sport that can easily be converted into these types of connections within work environments, and benefit one within these environments. It was cultural, rather than social, capital that was the primary predictor of Olympic/highbrow sports watching. There is finally some overall evidence from the results that average social network prestige is actually a negative predictor of sports spectatorship, again perhaps echoing Bourdieu’s (1978) assertions regarding mass TV sports events and their appeal to larger social classes.
It is, however, social network prestige, along with network size, that is a primary predictor of membership in the fitness/highbrow group of sports participation. Therefore, this type of circumscription of an elevated social network for this group mirrors the circumscription that Bourdieu (1984) argued for engagement in highbrow cultural variables. This result also suggests that even as there may be some ‘reflexive awareness’ that social network exclusivity is not the ‘done thing’ anymore, people’s actual networks do not reflect a similar democratization (Savage et al., 2015: 130). This is likely especially true where advantageous social capital networks intersect with certain groupings of cultural tastes and activities. Indeed, within sport, certain types of sports clubs strongly link these two and can operate in this manner (Forsell, 2020). The social capital variables used in this paper are approximately equally predictive as cultural capital for membership in this group, and much more predictive than economic capital. While age and gender are the most predictive variables for traditional/lowbrow and omnivore groups of sports participation, results suggest that all three social capital measures of network variety, prestige and especially size are the most important capital variable for omnivorous direct sports consumers. The results for omnivorous participation show how interrelated these elements of social capital can be for maximizing amplification of social capital benefits (Bourdieu, 1986), and expanding cultural knowledge (Erickson, 1996). These results mirror those of the omnivorous spectator, but to a much more acute level.
Although these results present some novel empirical contributions and applications of social capital theories, more needs to be done in understanding differences in modes of sports engagement, and the social class differences that they may manifest and reproduce. While this paper broadly captures sports spectatorship and participation, it is not able to account for modes beyond this. Likewise, the data used in this study is unable to parse social capital differences in modes of spectatorship. The results of the analysis of this paper do suggest, however, that further understanding of social capital network mixes (size, prestige, variety), and particularly gendered differences therein, provide promising and worthy nodes of further examination. Finally, more can qualitatively be done to understand the symbolic role that particularly social network prestige and variety play in social capital, social exclusion and the reproduction of social inequalities.
In conclusion, the results of this study show social capital as a primary predictor in the patterning of both sports spectatorship and participation. These patterns comprise discriminating cleavages of sport characteristics, as well as cleavages of inactive and omnivorous participation. However, the type of social capital that is most predictive seems to heavily rely upon the nature of these cleavages. While network size seems most universally applicable to predict sports engagement, omnivorism seems to rely more on social network variety than other engagement profiles, and network prestige is applicable to some highbrow sports profiles. These results make sense through prior theoretical work on social capital (e.g. Bourdieu, 1986; Burt, 2000; Granovetter, 1973), and how it operates within culture (e.g. Erickson, 1996) and sport (e.g. Widdop et al., 2016). However, more work needs to be done in understanding how different elements of social capital interact with each other in the domain of sport, and in taking more seriously how social capital impacts sports spectatorship and fandom.
Footnotes
Appendix 1
Latent class analysis summary report for direct sports participation. Model used in analysis in boldface type.
|
|
|
|
|---|---|---|
| 1-Class | 13091.2856 | 13108.2856 |
| 2-Class | 12431.7672 | 12466.7672 |
| 3-Class * | 12396.1379 | 12449.1379 |
|
|
|
|
| 5-Class | 12471.6996 | 12560.6996 |
| 6-Class | 12533.7217 | 12640.7217 |
Represents model of best fit according to BIC and CAIC fit statistics.
Acknowledgements
Special thanks to Professor Mike Savage of the LSE, under whose supervision the survey for this paper was designed and conducted. Thanks also to two anonymous reviewers for their valuable input on this paper.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
