Abstract
There has been for some time a significant and growing body of research around the relationship between sport and social capital. Similarly, within sociology there has been a corpus of work that has acknowledged the emergence of the omnivore–univore relationship. Surprisingly, relatively few studies examining sport and social capital have taken the omnivore–univore framework as a basis for understanding the relationship between sport and social capital. This gap in the sociology of sport literature and knowledge is rectified by this study that takes not Putnam, Coleman or Bourdieu, but Lin’s social network approach to social capital. The implications of this article are that researchers investigating sport and social capital need to understand more about how social networks and places for sport work to create social capital and, in particular, influence participating in sporting activities. The results indicate that social networks both facilitate and constrain sports participation; whilst family and friendship networks are central in active lifestyles, those who are less active have limited networks.
Social capital has been conceptualized and operationalized in a number of ways (Bourdieu, 1984; Coleman, 1988; Lin, 2001; Putnam, 2000). There are many reasons why social capital has attracted so much attention over the past 30 years or so: the assumption that civil society and democracy depend upon it; that some of the important features that contribute to social life, such as networks, norms and trust, have been diminishing; that civic engagement, volunteering, community networks and public service have all been under threat; and that the means or the resources for developing shared human objectives and capabilities have been challenged as a result of changing individual and societal priorities.
As scholarly interest in the concept of social capital has flourished, the relationship between sport and social capital has gained greater prominence, primarily through the seminal contributions of Jarvie (2003), Delaney and Keaney (2005), Coalter (2010), Nicholson and Hoye (2008) and, more recently, Widdop and Cutts (2012) and Nichols et al. (2013). This body of work has at its core some or all of the following themes: the extent to which sport contributes to the bridging (social ties that connect people together with others across a cleavage that normally divides society, e.g. race or class) and/or bonding (the social ties that link people together with others who are similar to them along some key dimension) aspects of social capital; the relationship between sports participation and social capital; the role of sport in fostering different aspects of communitarianism; and the role of sport in the development of not just human capabilities but community life, as suggested by a former United Nations (UN) Secretary General for Europe who asserted that “The hidden face of sport is also the tens of thousands of enthusiasts who find in their football, rowing, athletics and rock climbing clubs a place for meeting and exchange but above all the training ground for community life” (Jarvie and Thornton, 2012: 255).
Yet, despite the saliency of sport in the scholarly work of Bourdieu (1984) and calls by cultural sociologist Richard Peterson (2005) to examine omnivore–univore habits across the cultural spectrum, including sport, such studies investigating these relationships with the sporting field remain rare. Similarly, little attention has been paid to how resources embedded in social networks (Lin, 2001) – as measured by an individual’s position in the social structure, diversity and homophily of an individual’s network and the strength of ties in the network – influence sport participation.
It is precisely these gaps in the sport literature that this article seeks to fill. More specifically, the article is premised upon testing certain propositions in relation to sports participation and social capital. Firstly, an exploration of the key theories of participation enables us to examine whether sport participation patterns are segmented into different typologies or classes that closely match with the omnivore–univore thesis. Secondly, we examine how social capital, through resources embedded in networks (Lin, 2001), is a vital mechanism within these theories of participation. Our findings provide a first and original insight into how sport participation is structured in England and details how salient larger diverse networks and the strength of these ties account for omnivorous as opposed to other behaviours.
While our specific research questions associated with the evidence base in this article are outlined later, the key questions that this article addresses are:
How is sport participation structured in England – does it adhere to the omnivore–univore framework?
How does this framework contribute to the existing body of work on sport and social capital?
How can this be examined using a quantitative analysis of sports participation data from one country?
The omnivore–univore framework
To put this into context we must first discuss the omnivore thesis, as this is the position from which network effects will be explored. Up until the early 1990s, the path-breaking work of Pierre Bourdieu’s “Distinction” represented the most comprehensive theoretical understanding and explanation of the apparent interrelationship between cultural and social hierarchies, through cultural capital (forms of knowledge, skills, education and advantages that a person has, which give them a higher status in society) (Bourdieu, 1984). The existence of a homology in cultural stratification, that people belonging to the dominant classes had more levels of cultural capital, affirmed through their higher social status through the consuming of highbrow culture while those with lower social status preferred and consumed lowbrow culture, became the orthodoxy for 20 years or more. However, by the last decade of the 20th century, scholars began to question whether Bourdieu’s theory still reflected contemporary social reality (Lamont and Lareau, 1988; Stichele and Laermans, 2006; Van Eijck, 1999). In a number of important articles, Peterson and his colleagues reformulated the relationship between status hierarchy and cultural taste (Peterson, 1992; Peterson and Kern, 1996; Peterson and Simkus, 1992). Put simply, high-status groups had a broader cultural repertoire, appreciating more middlebrow and lowbrow activities than the orthodoxy suggested. These were labelled “omnivores”. The lower status groups were restricted in their consuming patterns to only the mainstream or popular culture, and were, therefore, coined “univores” (Peterson, 2005; Peterson and Kern, 1996; Peterson and Simkus, 1992). Following this ground-breaking work, numerous scholars have sought to classify cultural preferences in a broadly similar way, with many supportive, although not exclusively so (Bryson, 1996; Chan and Goldthorpe, 2005; Sintas and Alvarez, 2002; Van Eijck, 2000, 2001; Van Rees et al., 1999).
Most scholars in this field have observed the existence of an omnivore group and claim that greater socio-cultural heterogeneity reflects the rise in social mobility over recent decades (Peterson, 2005). Scholars also stress the growth of the mass media, advancement in online technology, the development of the leisure industry and easier access to higher education as other important drivers (Peterson, 2005; Stichele and Laermans, 2006; Widdop and Cutts, 2012). Nonetheless, numerous empirical findings suggest that the omnivore group is relatively small in number and that its socio-economic make-up does not purely reflect the relationship between economic class and patterns of participation (Katz-Gerro, 2006; Sullivan and Katz-Gerro, 2007). Generally, studies have found that higher education, higher income and higher occupational status are strongly associated with omnivorous cultural preferences (Sintas and Alvarez, 2002; Van Eijck, 2001). However, the effects of gender and age are contested. Whether gender is strongly associated with omnivorism depends upon the domain of activity selected for the analysis, as shown by the differentiated gender effects found in a number of studies (Sintas and Alvarez, 2002; Van Eijck, 2001; Warde and Gayo-Cal, 2009). Similarly, age effects have been contested, with some scholars suggesting that younger age cohorts are more inclined to be omnivores (Stichele and Laermans, 2006; Van Eijck, 2000; Widdop and Cutts, 2012), whilst others disagree (Warde et al., 2007; Warde and Gayo-Cal, 2009). However, apart from the odd study (Widdop and Cutts, 2012), the existence of omnivores and other participation patterns in the sporting field have been under-researched. Moreover, if they resemble other cultural fields, what determines their existence remains a source of considerable debate.
The omnivorousness literature now spans much of Europe, Australia and North America, and even some countries in South America (Alderson et al., 2007; Torche, 2007; Van Eijck, 2000; 2001; Van Rees et al., 1999). These studies have established that omnivorousness is related to high status (Chan and Goldthorpe, 2005, 2007), class and education (Bryson, 1996; Chan and Goldthorpe, 2007; Erickson, 1996; Peterson and Kern, 1996; Peterson and Simkus, 1992; Sintas and Alvarez, 2002, 2004; Tampubolon, 2008; Van Eijck, 1999), gender and age (Erickson, 1996; Sintas and Alvarez, 2002; Stichele and Laermans, 2006; Van Eijck, 2000, 2001; Warde and Gayo-Cal, 2009; Widdop and Cutts, 2013) and place (Widdop and Cutts, 2012). One area that is under-researched is the relationship between participation, networks and social capital in the rise of omnivorism. Both Erickson (1996) and Van Eijck (1999) claim that omnivores benefit from a broader and more diverse social network where they can display knowledge gained from interaction with individuals in different social circles, which reinforces social approval within these circles. As Erickson (1996) claims, “the most powerful teacher of cultural variety is contact with people in many different locations”. Omnivores, therefore, benefit from resources embedded in networks as a basis for forging social capital. However, such claims have yet to be discussed or tested within the sport and social capital literature.
Sport participation, social capital and social networks
As we highlighted earlier, since the late 1980s the development of social capital has been viewed as a way of renewing democracy. In this sense it is referred to as the network of social groups and relationships that fosters co-operative working and community well-being. We suggested two reasons why social capital attracted so much attention (Jarvie, 2003; Jarvie and Thornton, 2012). On the one hand, civil society and communities depend upon it (Auld, 2008; Jarvie, 2008; Jowell, 2005). On the other hand, democracy depends upon social capital. This is true in one very obvious sense, that democracy depends upon everyone trusting that everyone else will operate the system constructively. When that trust breaks down – for example, as a reaction to certain screening practices aimed at the control of drugs in athletes or the failure to deliver sustainable sporting and economic benefits for deprived inner-city urban ghettos, or the adequate funding arrangements for sport in universities or local authorities – the result is cynicism about democracy in general. However, the systematic critique of the concept or the potential role of social capital as a basis for understanding social networks and much more remains profound.
The reviews of existing UK evidence on the relationship between sport and social capital are to some extent best represented by Delaney and Keaney’s 2005 review of the British statistical evidence on sport and social capital (Delaney and Keaney, 2005). The former compare the level of social capital and sporting participation in Britain with the rest of the European Union (EU) and examine the links between different types of sporting participation and individual measures of social capital. While it is ill-advised to make causal inferences from this type of analysis, the findings do provide valuable benchmarking information as a background to more experimental studies. The 2005 results demonstrate substantial correlations between measures of social capital and measures of sporting participation, both at the national level and, within Britain, at the individual level. Further analysis, controlling for several different types of individual characteristics, yields a more complex picture, with sports club membership positively affecting well-being and sociability but having little effect on political participation and personal trust.
The types of social capital associated with sport are varied and yet the exact mix of social capital and the exact outcomes remain open to question (Coalter, 2010, 2013). At an individual level sport may provide a basis for an individual to form a friendship base, provide goals and foster well-being. Sport may also absorb pro-social motivations and utilize the talents of diverse individuals. At a local-community level sport may provide a basis for the building of local networks. Through interacting with children’s sports, parents’ networks may form in such a way as to have potentially beneficial effects. Sport may provide a basis for bringing different sections of communities together. At a national level sport may provide a basis for common shared norms and conversational points, as well as providing a basis for collective memory. It can act to transmit pro-social values such as fairness and rule following. Sport may also act as a vehicle for citizens to engage with other countries (Cha, 2009; Murray, 2013; Murray and Pigman, 2013).
Yet the relationship between sport participation and the development of social capital is not straightforward. Delaney and Keaney (2005) concluded that British people are more likely than the average European to belong to a sports club and participate in a sport and are about as likely as the average European to volunteer in sports, but there is still a long way to go before reaching Scandinavian levels of sports participation. Sport is the most popular type of group activity in Britain, and sports organizations do better than most other types of organizations in building and sustaining friendships and networks. However, the most popular activities are ones that are often carried out alone and so are less likely to generate certain types of social capital.
A diverse number of positions are adopted within the existing research literature that examines the relationship between sport and social capital. The premise behind the notion of social capital is rather simple and straightforward in that it is premised upon the notion of investment in social relations with expected returns. The sport and social capital literature tends to draw upon three main theories of social capital, as reflected in the work of Bourdieu (1980, 1986, 1990) Coleman (1988, 1990) and Putnam (1993, 1995a, 1995b, 2000). While acknowledging the contribution that social capital has made to forms of social intervention, development and public policy, the term itself has recently been brought into question. Renard (2006) has highlighted in relation to international aid efforts that there are potential cracks in the new aid paradigm of social capital. The scrutiny is now very much upon the complexity of social capital, forms of social capital, the extent to which different forms of bridging or bonding or both work, the different contexts and the differential outcomes that inform or realize how, why and if social capital works. Perhaps surprisingly, the sport and social capital literature has said less about the contribution made by Lin (1999, 2001) and social network theory. This tends to explain social capital in terms of access to the use of resources that are embedded in social networks.
Social networks and sports participation
Although limited, there have been studies that have shown that participation benefits from a broader and more diverse social network. Here individuals can display knowledge gained from interaction with others in different social circles, which in turn reinforces social approval within these circles (Erickson, 1996; Kane, 2004; Lizardo, 2006; Relish, 1997; Warde and Tampubolon, 2008). These studies all observe that the network structure and an individual’s position within that structure have a key impact upon the resources available to them for participation behaviour. Put simply, it is somewhat of a mediating factor in their construction and socialization of cultural preferences and participation patterns. However, do the two concepts of the network structure, the diversity of network and type of ties, impact upon sports participation?
Diversity in the network
Network homophily and heterophily are important concepts in the network structure. Homophily works on the premise that people like people who are similar to themselves: birds of a feather flock together (Borgatti et al., 2013). Therefore, a homophilous network consists of individuals who are similar in characteristics, such as social class, age, etc. By contrast, network heterophily is indicative of a socially diverse mix of individuals in a network. Naturally these two network concepts impact upon sport participation, but both concepts have been used to explain behaviour. For Mark (1998), examining musical preference patterns, preferences are transmitted through homophilous network ties; similar people interact with each other and develop similar musical tastes. However, Erickson’s (1996) study of cultural preferences in the workplace noted that people with varied connections (heterophily) know more about different types of culture and sport and, as a consequence, developed omnivorous tastes that allow them to respond in different social settings. For Erickson (1996), the most widely useful cultural resource was cultural variety and this is closely linked to network variety. The greater the diversity of the network, the greater the exposure is to different forms of culture, to which the individual must respond, stimulating omnivorous behaviour.
For Erickson (1996) personal networks are a major source of cultural resources and a more powerful source than class itself. High-status people will certainly have a greater level of cultural capital, but this reflects not only their individual class position, but also the extent to which they are embedded in diverse class-based networks. Furthermore, Kane (2004) notes that omnivorous behaviours and diverse networks may indicate an underlying desire for cosmopolitanism. This is compounded by the fact that in all studies of this nature high levels of cultural participation and diverse networks are associated with high status. To that end one would expect to find that analogous to omnivorous behaviour, a low-status group would be characterized with low participation rates and restricted networks.
Types of ties
Whilst diverse networks might be the key to unlocking the growing omnivorous patterns found in different cultural fields across Western Europe and American, who is in this network might also be crucial. In this paper, sport participation is seen as a social act. An individual may consume or play sport on their own but inevitably they interact, communicate and consume physical forms of sport with family, friends and acquaintances. Therefore, as well as diverse networks, who you share sports with socially will be important; the types of ties in your social networks will mediate participation behaviours. For example, sharing time with a diverse friendship network might be very different to having a diverse family network.
Granovetter’s (1973) seminal strength of weak ties theory noted that input of new information into a network was more likely to occur in more heterogeneous networks, where weak ties are more preferable to strong binding ties. Whilst he was looking at the employment market, the same rationale can be applied to sport participation. The network structure of weak ties allows individuals to tap into a greater variety of sport genres, and act as conduits for these sport sources otherwise removed from the individual (Granovetter, 1973; Kane, 2004). Therefore, under this framework, individuals with omnivorous behaviour are more likely to have looser, less dense networks made up of more bridging types of contacts where new information about sport is more readily available. As a consequence, we would expect that omnivorous groups would be more reliant on diverse friendship and acquaintances networks, measured against less physically active groups who have more bonding ties characteristic of family ties.
Thus, Nan Lin’s (2001, 2009) approach to social capital as a network resources approach stresses how resources embedded in social networks are the crucial element of social capital. In the Nan Lin tradition (1999, 2001), emphasis is placed on an individual’s position in the social structure, diversity and homophily of an individual’s network and strength of ties in the said network. However, is this the case in the sport? Is it more important for omnivores than other sport participation classes? Here we test several social capital perspectives in addition to Lin’s (such as trust, social participation and belonging) to determine their importance on sport participation after controlling for other established influences. As a consequence, the empirical contribution to this article sets out to (i) establish if there are well-defined omnivorous patterns in sport in England; (ii) whether omnivorous patterns are socially stratified (education and class), and influenced by other socio-demographic factors; (iii) assess the impact of cognitive/subjective social capital (shared norms, belonging and trust) on sport participation patterns; and (iv) examine the importance of family and friendship networks on participation in sport, specifically, whether omnivorous behaviours are more or less likely to be associated with larger diverse networks and the strength of these ties.
Empirical research questions
In this article, we set out to test several social capital perspectives to determine their importance with regard to sport participation in England after controlling for other established individual-level influences. One of the key innovations is to determine the extent to which omnivorous behaviours are more or less likely to be associated with larger diverse networks and strong ties. Thus, within the more empirical part of this article we seek to address the following questions:
RQ1: To what extent are the lifestyle patterns found in the sporting field distinctive? Do they concur with the framework laid out in the omnivore–univore thesis?
RQ2: To what extent do key socio-economic variables explain differences in sporting lifestyle membership?
RQ3: How important are the effects of social participation, trust and belonging on sporting lifestyle membership, after accounting for other established factors?
RQ4: Are certain sporting lifestyle groups more likely to have heterophilous than homophilous networks?
RQ5: How important are social participation and the different types of ties in diverse/similar networks when differentiating between the two omnivore groups?
Data and methods
Data
This study uses Wave 3 of the Taking Part Survey (TPS) to examine our key research questions and the intersection between sports participation and social networks and social capital as formulated by Lin. The TPS surveyed adults via face-to-face interviews, about their participation in sport and cultural activities, between July 2007 and June 2008. Households were drawn from the United Kingdom national postcode address file, and interviews were conducted with a randomly selected member of each household aged 16 or over. As part of the questionnaire design, questions on social capital and participation are only asked of a randomly taken sample of respondents. This sample consists of 12,991 respondents.
Sport participation
To assess sport participation, respondents were asked a series of questions relating to their sport activities in the last 12 months (1 = Yes, 0 = No). For the purpose of this study we define sport participation as engaging in the following sports, that cross-cut the perceived symbolic boundaries of sport. A total of 25 sports from the data were used to represent eight sporting indicators: swimming (indoor and outdoor), health and fitness (keep fit and aerobics, health, fitness, gym or conditioning activities), cycling, football (5-a-side and 11-a-side), golf, water sports (rowing, yachting or dingy sailing, canoeing, windsurfing or boardsailing, water skiing and other water sport), racket sports (tennis, badminton and squash) and recreational sports (including orienteering, rambling, hill trekking or backpacking and climbing/mountaineering) (see Table 1). Some indicators are more attached to the masses (football 9%), while others are far more exclusive (racket sports 11%, water sports 4.3% and recreational sports 9.6%).
Participation in sport.
Individual socio-economic characteristics
In the TPS, education is coded to the six official National Vocational Qualifications levels (England), ranging from degree level to no qualifications (see the Appendix). It follows a near-linear distribution so we treat it as a continuous variable. Following the National Statistics Socio-economic Classification, we distinguished between the salariat class (managerial and professional occupations), intermediate class (intermediate, lower supervisory and technical occupations, and small employers and own account workers) and working class ((semi-)routine occupations, long-term unemployed and people who have never worked). Along with social class and educational attainment, other variables included gender (female dummy variable), age (continuous) and age squared to mediate the curved relationship of age (which allows us to model the effect at differing ages, rather than assuming the effect is linear for all ages).
Social capital variables
We used a number of variables to measure the different aspects of social capital: neighbourhood trust, belonging, social participation (socializing with friends and family) and network resources. We dichotomized the neighbourhood trust variable (generally speaking, would you say that most people in your neighbourhoods can be trusted) into a “distrust” category, including the answers “you can’t be too careful” and “it depends”, and a “trust” category. Social participation was measured by asking how often respondents meet up with friends (1) and with relatives outside the household (2). Response categories were “never”, “less often than once a month”, “once or twice a month”, “once or twice a week” and “most days”. Participation with friends and family were dichotomized into low participation (never or less often than once a month) and high participation (once or twice a month or more). We also controlled for an individual’s sense of belonging to an area. Here we conceptualize belonging as a socially constructed, embedded process where individuals subjectively gauge the suitability of their locale in light of their social trajectory. In the sporting context, such an attachment may have an important bearing on their sporting participation patterns.
Social network resources were measured using the position generator, a measure of social capital developed by Nan Lin (Lin, 2001). This instrument asks people about their network members’ occupational positions and considers these positions as good indicators of the network resources (Verhaeghe and Tampubolon, 2012). In this study, respondents were asked whether they know friends, relatives or acquaintances who have any of the jobs from a list of 11 occupations. All 11 occupations are salient in British society and range from factory worker to university/college lecturer (Verhaeghe and Tampubolon, 2012). In this paper, the position generator is used to calculate the volume of network resources by counting the number of different occupations accessed by respondents. This measure is related to the network size (Van der Gaag, 2005). Furthermore, this is split into three variables: volume of network resources that are friends; volume of network resources that are family; volume of network resources that are acquaintances. We split this variable into three as there may be different underlying processes at work that influence whether individuals invest in the different types of network ties (Verhaeghe and Tampubolon, 2012).
Analytical approach: latent class modelling
Participation in one particular type of sport does not happen in a social vacuum: it is part of the wider cultural make-up of an individual. Rather than examine sport items as discrete components, individuals should be grouped on observed patterns of participation (Chan and Goldthorpe, 2005; Peterson and Kern, 1996; Sintas and Alvarez, 2004; Van Eijck, 1999). Here we assume that there are relatively well-defined types of sports participation groups in which individuals can be placed, based on their engagement in different sports. This is executed through a latent class analysis (LCA) modelling approach.
The LCA identifies typology groups or classes whose sporting behaviour will be different depending on membership of these classes. Individuals form sporting patterns based on participating in sporting indicators “u1 … u n” and can thus be assigned to different levels of a latent variable (class C in Figure 1). From this, it is possible to identify different types of sport participation groups. LCA usually assumes local independence and estimates two essential parameters, latent class probabilities (the probability of an individual being in a particular level or lifestyle group) and conditional probabilities. Conditional probabilities are akin to factor loadings and are the probabilities of an individual in class t of the latent variable C, being in a particular level of the observed variable (Magidson and Vermunt, 2004; Widdop and Cutts, 2012). The LCA is traditionally termed the measurement part of the model.
Path diagram – latent class analysis and Multiple Indicator Multiple Cause (MIMIC) model.
In a latent class model, the standard chi squared measurement (L2) can be unreliable because of the number of sparse cells in the model. We therefore use alternative measures to determine the goodness of fit, including the Bayes Information Criterion (BIC), the Akaike Information Criterion (AIC) and the Consistent Akaike Information Criterion (CAIC). These measures are used because they weight both model fit and parsimony and are useful to compare models. The most widely used and statistically robust is the BIC, where a model with a lower BIC value is preferred over a model with a higher BIC value (Asparouhov and Muthen, 2006; Widdop and Cutts, 2012).
To introduce explanatory variables into the model we use a Multiple Indicator Multiple Cause (MIMIC) approach, which is presented in Figure 1. As mentioned earlier, the subscript “u” defines a categorical variable of interest (i.e. football participation, swimming, etc.) and the circle encapsulating the “C” is an underlying latent class measure (can include 1, 2, 3 … n classes). Thus, the indicator variables are seen as arising from the unobserved latent class measure and are subject to measurement error. This is the measurement part of the model or LCA. The X variables influencing the latent class measure are independent explanatory variables (i.e. social class, education). This second component adds structure to the model and allows investigation into the relationship between latent class groups and a set of theoretically informed explanatory variables. In its simplest form, a MIMIC model is a simultaneous method of LCA and multinomial regression, or logistic regression when there are only two levels of the latent variable (two classes). We use the software Latent Gold for the models in this paper.
Results
Number of sporting classes
A LCA enables us to estimate the probabilities that an individual belongs to a certain class/typology, given their participation frequency patterns in the eight sporting variables. The initial aim is to determine the appropriate number of participation groups (classes) that exist in the population; in other words, the most parsimonious model that provides the best fit to the observed data. Table 2 identifies the model fit statistics for a 1–5 class solution. From our data a four-class solution is the best model. Each goodness-to-fit measure reached its optimal point at a four-class solution.
Model fit statistics.
BIC: Bayes Information Criterion; AIC: Akaike Information Criterion; CAIC: Consistent Akaike Information Criterion.
Profile of sport clusters
In this section we address our first research question (RQ1). As Widdop and Cutts (2012) found using earlier data (Wave 1 of the TPS), there are well-defined underlying sport participation groups that share similar response patterns given membership of a given class. These four classes also emulate those found in other cultural fields (Chan and Goldthorpe, 2006, 2007a, 2007b, 2007c; Sintas and Alvarez, 2002, 2004; Stichele and Laermans, 2005; Tampubolon, 2008), whereby there is a large inactive group, a popular class group often referred to as “univores” and two omnivorous groups separated by attachment to high and popular culture (lowbrow omnivores and highbrow omnivores). Table 3 presents the estimated size of the latent class clusters and the estimated conditional probability of consuming each of the eight sport indicators given membership in a latent lifestyle cluster.
Latent class probabilities.
The lowest populated group (7%) is Latent Class 1, which we label the “highbrow omnivores”. They are highly distinguishable from the other classes for their extremely active participation and sheer insatiable appetite for all the sporting items. These “highbrow omnivores” not only have a high probability of consuming all of the sporting items, but of all the lifestyle groups, they are the most likely consumers of highbrow sports, which include water sports (28%), racket sports (46%) and recreational sports (53%). Interestingly, they are unlikely consumers of association football, that is, in the Bourdieusian tradition, they engage in legitimate culture but distance themselves (aesthetically) from more popularized activities (Bourdieu, 1984; Peterson and Kern, 1996).
Latent Class 2 comprises 10% of the survey population, and is noticeably an omnivorous group in the traditional sense (Bryson, 1996; Peterson and Kern, 1996), but with a caveat. Whilst this class group consume all types of sport measured here, they are light consumers of those sports traditionally deemed as status-defining sports (highbrow), namely water sports (6%) and outdoor recreational sports (10%). Members of this group have a 71% probability of being consumers of association football, a popular sport in Britain often associated with the working classes (although this has changed since 1993 and the introduction of the Premier League and opening up the game to the middle classes). Therefore, we label this group “lowbrow omnivores”, a group also found by Stichele and Laermans (2006).
The remaining two classes have more restricted participation patterns, but make up 83% of the survey population. Latent Class 3 is labelled the “fitness class”. Whilst they have relatively low participation of team and highbrow sports, when they do participate, it is in those sports most associated with health and body fitness. Latent Class 4 are the “inactives”; omnipresent in research of this nature, they are differentiated from the other classes through their disengagement with sport. Nonetheless, it is clear from our findings that there are different types of sporting lifestyle groups that broadly concur with the omnivore–univore framework found in other cultural fields (RQ1).
Sporting lifestyles: what are the key drivers of membership?
What is the individual socio-economic profile of each sporting class or lifestyle previously identified? Are certain social capital variables more important for membership of some sporting classes than others? Here we examine the individual profile of the latent classes and what influences membership. Table 4 presents the conditional probabilities of membership for each sport cluster by education, class, age, gender, size of networks, trust, social participation and belonging. The findings provide an insight into the socio-economic make-up of each sport cluster (specifically RQ2), and also illustrate the importance of social participation, trust and belonging in determining the latent class membership beyond stratification variables, such as class and education (RQ3). Finally, we examine whether certain sporting lifestyle groups are more likely to have heterophilous than homophilous networks (RQ4).
Fitness class as ref.
As mentioned previously, a MIMIC model is essentially a multinomial regression with a dependent variable that is latent or unknown. Like all multinomial regression models, the dependent variable, in our case the four latent classes, requires a reference category, which other categories of the said variable are measured against. In the models shown below we use the “fitness class” as the reference (see Table 4). The justification for this is twofold; firstly, it is comparable in size to the “inactives” class; secondly, it allows us to compare an actively engaged group against other engaged and non-engaged classes. As a consequence, the information derived is much more meaningful, as opposed to using “inactives” as the base (which other studies have done).
Firstly, we address our second research question (RQ2) by examining the key socio-economic variables. When measured against the “fitness class”, “highbrow omnivores” have a greater likelihood of being highly educated (+0.833), and significantly more likely to be from the salariat classes (+0.422). Clearly, education and class play a significant role in differentiating between these two sporting classes. This is not the case for “lowbrow omnivores”: when compared against the “fitness lifestyle group”, class and education are not significant, suggesting that there is little difference in the socio-economic profile of these two sporting lifestyle groups. As expected, the non-participant class (the “inactives”) are less likely to be educated (–0.493) and tend to be drawn from the lower working classes. There is, however, some distinction by age and gender. Both the “lowbrow omnivores” and the “inactives” tend to be from the younger age cohorts (-0.172 and -0.054, respectively), while there is evidence that gender is also a salient predictor of group memberships. The findings suggest that women are much more prevalent in the “fitness lifestyle group” than any of the other sporting lifestyle classes identified.
Turning to research question 3 (RQ3), it is clear that neighbourhood trust is a key predictor of sport participation across groups. When compared against the “fitness class”, those individuals with higher levels of trust are significantly more likely to be members of either the “highbrow omnivore” (0.673) or “lowbrow omnivore” (0.266) groups. As expected, non-participants tend to exhibit lower levels of neighbourhood trust. By contrast, having a sense of belonging is integral to membership of the “lowbrow omnivore” group (+0.431), but it is not important for membership of the other sporting classes.
Finally, in assessing research question 4 (RQ4), our findings suggest that networks also play a role in distinguishing between groups. Even when controlling for socio-demographic characteristics, both the “highbrow omnivores” and “lowbrow omnivores” are more likely to have a larger friendship network than the “fitness class”. The former are also less likely to socialize with family members. As expected, non-participants have fewer network ties, are significantly less likely to be trusting (–0.545) and less likely to report socializing with friends (–0.689) and family (–2.94), when measured against the “fitness class”.
Differentiating between highbrow and lowbrow omnivores
Is there a significant difference in the structure between the two omnivore groups? Put simply, to what extent can the two omnivore groups be distinguished by social participation, trust, belonging and the different types of network ties? We can address our research question (RQ5) by changing the reference group to “lowbrow omnivores” and re-running the models to determine how these groups are conceptually distinct. The findings are presented in Table 5.
Lowbrow omnivores as ref.
Once more, education and class are significantly different–“highbrow omnivores” are drawn from salariat classes (+0.419) and the highest educated (+0.967) in society. “Highbrow omnivores” are much more likely to be female, whilst age is not significantly different. However, the social capital variables do provide an interesting insight into the types of individuals who are members of the two omnivore groups. “Highbrow omnivores” have significantly higher levels of trust (+0.407), whilst “lowbrow omnivores” are more inclined to report belonging to an area and socializing with kin. This finding suggests that both are reliant on networks but alternative mechanisms of social capital are in place. However, there are no differences between the two groups in terms of networks by volume. Given the saliency of the friendship network for both omnivore groups (see Table 4), it is clear from these findings that this is not more important for membership of one group than the other. By contrast, “lowbrow omnivores” are reliant on their social network of their local area, with a strong sense of belonging to the area and socializing with family. “Highbrow omnivores” portray a socially mobile group with reliance on less dense and looser networks (less belonging, less socializing with family).
Discussion and conclusion
Sport is complex and governments across Europe and elsewhere continue to relate sports participation to broader social concerns. A significant number of sport studies have also used social capital as a basis for framing arguments and policy about the social value of sport. Critics have put forward a number of arguments and concerns about the value of social capital as a basis for framing and driving sporting interventions that testify to build a range of social benefits, including trust. These have included a greater emphasis being placed upon the dark side of social capital as well as a continuing concern and call for evidence of how social capital works and in what ways and when. In view of this, our paper is timely in that it directly addresses cultural consumption as evidenced by sports participation patterns.
In summary, this article addresses five key research questions. Although we anticipated that we would identify different types of sporting lifestyle groups, it was less evident whether they would be similar or distinct from other cultural fields and if they would concur with the now established omnivore–univore framework (Peterson, 2005). As regards RQ1, our findings largely supported recent scholarly research in these fields. Of the four active sports’ clusters, there is a “highbrow omnivore” lifestyle group whose members participate in all types of sports at a greater volume and range than any other cluster. We also identify a “lowbrow omnivore” group that participated in the more popular sports and had a marginal interest in more exclusive activities. The “fitness class” is not only unique to sport but also distinctive within the sporting field when compared against the two omnivore sport groups. Our findings suggest that a significant proportion of the population who are engaged in sporting activities simply take exercise purely for the purpose of fitness and body discipline. However, there are some differences between sport and the other cultural fields. Recent scholarly evidence (Peterson, 2005; Van Eijck, 2001) that there are distinctive highbrow and omnivorous patterns of cultural participation is not confirmed here. We found no evidence that a small number of individuals restrict their participation exclusively to highbrow sports and ignore popularized sporting activities.
Our second research question (RQ2) examines the extent to which the key socio-economic variables explain the differences in sporting group membership. Here we found that the socio-economic make-up of the lifestyle groups not only vary by class and education, but also age and gender. While the “highbrow omnivore” group is dominated by those from the higher social strata – highly educated; upper/middle class – there is little difference in the socio-economic profile of the two other active clusters – the “lowbrow omnivore” group and “fitness class”, except by gender. The “fitness class” is more distinctive from the other clusters because women are more likely to be members of this cluster. The participation of women in a few number of sporting activities may reflect a limited leisure time – care responsibilities, work opportunities (greater part-time work) – but also objectives and attitudes to fitness and general exercise that are different from men (Warde, 2006; Widdop and Cutts, 2013).
The sport and social capital literature has largely ignored the omnivore–univore argument as a basis for advancing not only Lin’s approach to social capital but moving the discussion of sport and social class from that of Bourdieu’s notion of distinction to Lin’s notion of network resources. Here, not only were we able to further validate the existence of omnivoral patterns – the presence of two omnivore groups – in the sporting field, but it was possible to assess the importance of aspects of social participation, trust and belonging (RQ3), and the significance of networks (RQ4) on sporting lifestyle membership.
Regarding RQ3, we found two key findings. Firstly, neighbourhood trust is a key driver of sport participation for all active lifestyle groups. Both “highbrow omnivores” and “lowbrow omnivores” have higher levels of trust than members of the “fitness class”, although trust is a key predictor of sport participation for this group when compared against non-participants. However, having a sense of belonging is less important, except for the “lowbrow omnivore” group, suggesting that place attachment is partly borne out by this group’s lack of social mobility. Our findings also suggest that networks play an integral role in differentiating between lifestyle clusters (RQ4), irrespective of class and education. Both omnivore groups have larger friendship networks than members of the “fitness class”, although the “highbrow omnivore” group seems to be more socially mobile, have less dense and looser networks (less socializing with family), made up of friends in different locations of the social structure. Apart from friendship networks, both the “fitness class” and “lowbrow omnivore” exhibit similar network structures. The non-participants have a much more restricted network, which reinforces their lack of partaking in sporting activities.
Our final research question (RQ5) examines the extent to which the two omnivore groups, in terms of trust, belonging and different types of ties in a network, are distinguishable from each other after controlling for established socio-economic influences. Here we found an important distinction that reinforces the findings above (for RQ4). Put simply, “highbrow omnivores” are far more socially mobile and less attached to place with far looser networks, less belonging and less socializing with family, while “lowbrow omnivores” are considerably more reliant on the social network of their locale, with a strong sense of belonging and attachment to place and significantly more likely to socialize with the family. In summary, this validates the approach taken in this paper. Not only are there distinct and unique lifestyle groups (i.e. the existence of a “fitness class”), but there are clear differences between omnivore sporting clusters both in terms of their socio-economic profile but also through their attachment to place and the strength and different types of ties in their networks.
Even though this article provides an original insight into how sport participation is structured and illustrates how salient larger diverse networks and the strength of these ties account for omnivorous as opposed to other behaviours, it is not without its limitations. For instance, in this paper we take a generic overview of sports participation and, given our modelling approach, we are limited in the number of sport activities we can include. We are unable to examine the subtle differences in different types of activities, such as team sports and individual sports, or competitive sports and recreational sports. Moreover, we use a proxy measure of personal networks that are not actual networks as in the social network analysis tradition. Whilst Nan Lin’s position generator measure is an excellent and well-cited measure of social capital, and is suitable for the work here, it does limit the kind of analysis that can be done. As Borgatti et al. (2013) notes, it is difficult to use position generators to examine mediating and constraining effects of structure on behaviour. Put simply, this requires a social network analysis approach.
To further unpack the power of networks on sports participation we need to explore actual networks both quantitatively and qualitatively. By taking a social network approach we can examine both macro and micro effects. At the macro level we can evaluate the social structure of the network and what factors influence the collective action in terms of sports participation, specifically how density, cohesion and formation of the network influence participation in sport. At the micro level we can look at the emerging properties and specifically at how an individual’s position in the network – who they are and who they are/not connected too – can influence whether they partaking in sporting activities or not. Moreover, we need to be able to imagine the global social world as a vast web of relations and interactions, on multiple scales, and involving a multitude of types of both relationships and actors (Crossley, 2011). Finally, there is also a need to embrace the importance of contextual effects on sport participation, and unpack the underlying mechanisms that occur in different contexts. These provide a vital connecting tie between individual socio-economic factors and taking part in sporting activities (Widdop and Cutts, 2013). The lack of available data did not permit such a wide-ranging study, but data-rich networks both nationally and cross-nationally over time are vital if we are going to enhance our understanding of sporting participation patterns in the future.
Nonetheless, our findings suggest that any critical politics concerning a contemporary discussion of any “age of austerity” might consider revisiting notions of social capital, social cohesion and social networks as part of a solution to what Klein and others have termed neoliberal disaster capitalism. The added advantage of Lin’s informed approach to both social capital and social network analysis lies in its ability to systematically map social relationships and in this study it is these social relationships and networks through patterns of sports participation that have enabled us to bring a more nuanced understanding back to the study of sport and social life.
Footnotes
Continuous variables – descriptive statistics
| N | Minimum | Maximum | Mean | Std. d | |
|---|---|---|---|---|---|
| Education | 12,991 | 1 | 8 | 4.47 | 2.696 |
| Age | 12,991 | 16 | 96 | 49.37 | 18.783 |
| Volume family members: number of accessed occupations through family member | 12,991 | 0 | 11 | 0.9 | 1.151 |
| Volume friends: number of accessed occupations through friends | 12,991 | 0 | 11 | 2.01 | 2.093 |
| Volume acquaintances: number of accessed occupations through acquaintances | 12,991 | 0 | 11 | 0.61 | 1.212 |
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.