Gendered Relationships Between Sports Participation and Spectatorship

Abstract

This study investigates the relationship between sports participation and spectatorship, two dimensions of sports engagement that have significant policy implications, by analyzing data from a 2016 nationwide UK survey (N = 1105). Scholarly research into this connection is relatively limited, especially when examined through a gendered lens. Discussing theoretical considerations of gender and the interrelations between direct sports participation and spectatorship, we utilize statistical techniques, including latent-class analysis (LCA), which enable us to uncover patterns in spectatorship and participation in the UK (two classes for women; three classes for men). We further operationalize capital by drawing on Bourdieu’s framework—encompassing economic, cultural, social, and symbolic forms of capital—to assess the resources that shape individuals’ (gendered) access to and engagement with sport. We find strong evidence for links between sport participation and spectatorship (r = 0.194 for women; r = 0.360 for men). While men exhibit a diverse range of engagement profiles, our findings indicate that women must overcome additional barriers, requiring significantly higher levels of capital and a higher degree of personal engagement to participate, suggesting that women are still disproportionately challenged in accessing this socially salient cultural form.

1. Introduction

The traditional cleavage in categorizations of sports engagement is often between those participating in sports and those engaging as spectators [1,2]. Research into the former, however, has been significantly greater than research into the latter [1,3]. While a significant body of work on football (soccer) spectatorship exists, comparative work on spectators is generally lacking [4]. Likewise, while large-scale quantitative studies comparing sports participation patterns are more common, studies that consider both simultaneously are currently lacking from the existing literature on sports engagement. Quantitative studies of gender within the sociology of sport are also relatively sparse [4,5,6], an area contributed to through this research.

Considering sports spectatorship (through all mediums) and direct sports participation together can help with understanding if increasing engagement in one area can lead to increased engagement in others, which has numerous policy implications, from encouraging mass physical activity to public health implications to tourism and economic considerations. For instance, a public policymaker may want to know if increased spectatorship numbers (e.g., through the hosting or free broadcast of sporting events) lead to higher physical activity levels. Following England’s victory at the UEFA Women’s Euro 2022, there were predictions of knock-on effects for participation and spectatorship. Yet evidence of sustained changes remains limited. Such evidence may inform resource allocation to elite sport development or to the infrastructure needed to fully develop and realize these knock-on effects. Similarly, a large professional sports club may want to know if investment in women’s and girls’ community sport has the potential to grow its local fanbase.

These gaps in the literature and limitations in existing studies are only magnified when considering gendered comparisons between direct sports participation and spectatorship of sport. Starting from an early age, participation in sports is highly gendered, and these differences manifest in levels of engagement, choice of activity, and intensity of such involvement [7]. This is partly a product of the historically gendered development of physical education curricula in which sport is constructed as predominantly ‘male’ and subsequently socially valued [8,9]. Furthermore, the gendered nature of sport has been reinforced in how media cover sports events and athletes [10]. Sporting pressures at an early age can determine which sports girls and boys play in adolescence, influence which sports they participate in as adults, or even discourage them from participating in physical activity [7]. In recent years, there have been some examples of events and movements where women’s sport has gained traction and media interest (e.g., England women’s national and club football, among others) [11]. These events are often described in the moment as catalysts for change or ushering in a ‘new era’ [12]. However, evidence of sustained impact and progressive change is still required, as is evidence that this has translated into changed and sustained physical activity or spectatorship practices.

While the results of some studies suggest that gendered differences in overall sports participation do not apply in every national context, it is the case in the UK, with men more likely to participate in sport, both historically and throughout potential changes brought on by the COVID-19 pandemic [13]. The world of spectator sports and sports fans has historically been less ambiguously a male preserve [14,15]. Women are increasingly, and conspicuously, following sport and forming an ever-growing proportion of its spectators and fans [16,17]; however, women are arguably seen as less legitimate participants in the social space of sports spectatorship, particularly of men’s sport. There is still a dearth of work that considers sports participation and spectatorship together, even more so work that systematically considers their relationships and gendered relational dimensions. The research that does exist has been mixed, with some finding little or no relationship between participation and spectatorship in some contexts [18,19], while stronger relationships are found in others (e.g., in relation to geographical location) [1,20]. This is true, even as policymakers often make claims of a positive relationship between the two [19].

Spectatorship, whether through physical attendance at live events or through consuming media coverage via screens (i.e., TV or streaming), can take place in gendered spaces and is imbued with messages regarding gender and sport [10,12,21,22]. Therefore, in this study, the term ‘spectators’ encompasses individuals who attend live events and those who watch sports via television or digital platforms, specifically focusing on gendered viewing patterns in varied settings such as homes and other potentially gendered social spaces (e.g., pubs).

The gendered nature of sport as a male preserve [23] is emphasized through dominant media narratives prioritizing men’s sports [10]. In a recent study, Metcalfe [24] found that young women felt that sport was not ‘for them’ due to the underrepresentation of women athletes. There is a suggestion here that broader social expectations that equate sport with masculinity and maleness could influence how men and women engage in sports spectatorship and participation, particularly in sports that may challenge a cultural sense of gender appropriateness.

While studies of different samples from the same or different national contexts may give some clues as to these answers, we argue that it is most efficacious to consider these together with one sample. We thus use such a sample to understand the gendered relationship between sports spectatorship and participation for the same group of people. We ask the following research questions to guide our analysis.

RQ1. 

Is there a relationship between sports participation and spectatorship? Are those who engage in one more likely to engage in the other?

RQ2. 

What are the patterns of sports participation and spectatorship in the UK, including their socio-demographic composition?

RQ3. 

Are the patterns and nature of the relationship between direct sports participation and sports spectatorship different for women and men? If so, how and why?

2. The (Gendered) Relationships Between Sports Participation and Spectatorship

In her foundational work, Hargreaves [25] documents the gendered nature of sports. Although campaigns and initiatives that have promoted women and girl’s sport and challenged previously taken-for-granted assumptions about the athletic woman and girl (e.g., Sport England’s This Girl Can), persistent discrimination and patriarchal structures continue to leave many women disengaged and feeling that the field of sport is ‘naturally’ for men, and therefore, ‘not for them’ [26]. Extensive UK-based data has shown that women participate in less sport than men, and these differences start in childhood and track into adulthood [27,28]. Formative experiences of school and physical education are typically considered important socializing events that stabilize expected behaviors for young men and women in relation to sports [29].

Light and Kirk [30], in their analysis of Rugby Union’s position in the formation of hegemonic masculinity in elite private schools, emphasize that ‘[t]hey nostalgically recounted rugby matches watched by the entire school that had since assumed legendary status’. The position of male sport as central to how young people construct their understandings of gender-appropriate sports is critical to how future participation and spectator ‘rules’ may be formed. In the example of rugby union in elite private schools, this is also inextricably linked to social class and, by extension, often geography. Using a theoretical framework based on Bourdieu’s sociological concepts of habitus, field, and capital [31,32], the development of a gendered habitus [33] formed through previous socialization and experiences provides a framework through which an individual can develop tastes and beliefs underpinned by tacit knowledge of what constitutes appropriate behaviors and actions for one’s assumed gendered identity.

Bourdieu’s [32] (p. 290) sociological habitus refers to a ‘system of acquired, permanent and generative dispositions’ that underpins all thoughts, behaviors, and embodied realities. Scholars have applied the concept of habitus to that of gender, identifying that individuals also have a gendered habitus which influences which behaviors and actions are appropriate for one’s sense of self as a male (and thereby masculine) or female (and feminine) [34,35,36]. In seeking to explore how sports participation and spectatorship patterns are gendered, using the sociological habitus offers an opportunity to explore how socially constructed expectations of gender may influence the relationship (or lack thereof) between participation and spectatorship choices.

The issue of choice and agency is considered through Bourdieu’s habitus. For instance, Wacquant [37] writes, ‘[w]e can always say that individuals make choices, as long as we do not forget that they do not choose the principle of these choices.’ Thus, the choice of sport to participate in/spectate may be constrained by one’s habitus and its characteristics linked to one’s gender, class, or other defining formative experience or status. This indicates that broader social notions of acceptable gender norms or behaviors may influence how individuals act—reflecting the cultural shifts in meanings. Furthermore, studies have demonstrated that these cultural shifts are not static but dynamically change and evolve, leading to a more fluid understanding of gender roles that further impacts sports participation and spectatorship [38,39]. These ongoing social transformations emphasize the dynamic relationship between both durable and flexible habitus [40] and emerging social norms, highlighting that the constraints imposed by established gender expectations can and are gradually being reshaped by broader societal changes in the understanding of gender. In exploring how gendered expectations about sport spectatorship may have changed, [39] suggest that changing gendered expectations and a relaxing of masculinity may open more opportunities for young women to spectate sports and simultaneously may allow young men to explore other cultural activities such as arts, literature or music.

Capital denotes ‘the different goods, resources, and values around which power relations in a particular field crystallize’ [36] (p. 106). This can take the form of economic, cultural, social, or symbolic forms of capital, among field-specific forms of capital, which are usually particular configurations of one of these. Sociologically, different forms and capital levels have been found to be associated with specific sports participation and spectatorship patterns. For instance, increased levels of economic capital are predictive of both direct sports participation (e.g., refs [20,41,42,43]) and in-person sports spectatorship [e.g., refs [1,44,45]. Similarly, higher levels of cultural capital, in both the form of formal education and non-sport cultural engagement, have also been found to be predictive of direct participation (e.g., refs [46,47,48]), and professional sports following [49].

Increased social capital is also linked to sports participation [50,51] and spectatorship [52]. However, while sport is often praised for being a site for fostering community cohesion and cross-status connections [53], social capital can also serve an exclusionary and stratifying function arising from close, homogenous ties based on class position [54]. Bourdieu [54] argues that social capital is defined as the sum of resources from a network of mutually recognized relationships, essentially reflecting membership in a symbolic status group. This membership provides a collective ‘credential’ that legitimizes one’s claims to various forms of credit within and outside the group. Such credentials are often tied to higher education, specific professions, or cultural and sporting activities that display cultural capital and class position [50,55,56]. To this effect, Gemar [50] found that the prestige of one’s social network is an important predictor for participating in certain more ‘highbrow’ sports, thus potentially evidencing Bourdieu’s arguments around symbolically circumscribed group membership, credentialization, and mutual recognition.

Different capital is thus allocated according to different patterns of participation, and also according to gender. For instance, Metcalfe and Lindsey [7] demonstrated that young men and women can accrue different amounts of capital depending on their participation in sport: young men’s participation in competitive team sport is valued, rewarded, and revered; whereas in contrast, young women are ostracized when their participation in sport threaten and challenge socially constructed images of femininity. The traditional linkage of masculinity to specific sports (e.g., Rugby Union, Football, Boxing) means that men are socially rewarded, through social and symbolic capital, for their engagement [30,33]. Referring to Lagaert et al.’s [39] work, the softening of masculinity that is developing in current times [57], the once-assumed link between male participation in sport and/or spectatorship may be less prevalent, and as such there are other fields and spheres of life in which men can accrue capital. The reverse may also be true for women, who can increasingly access capital from alternative fields to those traditionally associated with femininity.

While the sustained impact of the most recent events (e.g., 2022 UEFA Euros) necessitates ongoing study, it is often an assumption underpinning the hosting of major national events that spectatorship and consumption of elite sport positively influence participation. This is regularly used as justification for hosting major events (e.g., Olympic Games), yet the evidence to suggest that there is a consistent and meaningful link between spectatorship and future participation is sparse [19] or found not to be the case in many instances [18]. This assumed connection links to Bourdieu’s concept of the field, in which each field (in this case, the field of sports participation and the field of sports spectatorship) has its own specific rules, struggles for legitimacy, and structuring principles that govern individual’s behaviors, within such fields [58]. Underpinning the assumption that Shamir and Ruskin [59] present in relation to the implicit connection between participation and spectatorship is that the rules that govern the two fields are symmetrical. This assumption has persisted in the intervening decades, which further emphasizes the relative consistency of habitus and non-conscious assumptions about the rules of different fields and what is acceptable (or not) within them.

Regarding gender in these fields of sport, the gender-based construction of these fields often privileges the participation of male athletes and spectators. This likewise affords them greater power within these fields and can lower thresholds of social and symbolic capital (among others) to legitimately occupy and participate in these activities and increase the capital accrual potential for those participants. These power differentials, in turn, reproduce power dynamics within these fields. Exploring how the gender habitus may influence individuals’ engagement with sport (either as a spectator or a participant), this paper will contribute to ascertaining whether the ‘rules’ of these two fields are either symmetrical (i.e., the role of, and therefore may help to explain a positive pattern in behavior, or indeed, disparate, and support the assertion of previous research that suggests these two behaviors are uncorrelated.

This study covers a range of the most popular participatory and spectator sports in the UK, including those featuring spectator matches and direct participation regardless of gender. Alongside the dynamics we focus on in this paper, we also acknowledge that much prior research indicates that sporting governance structures can significantly shape participation and spectator behaviors. For instance, football (soccer) is administered by multiple governing bodies whose marketing strategies, such as those employed by the Football Association (FA) in promoting the Women’s Super League through family-focused and other initiatives, can significantly influence public engagement and viewership [11], as can the priorities and focus of media organizations [10]. Additionally, specific regional structures and disparities affecting availability to participate or view certain sports may also affect viewership or participation patterns.

This section has outlined the key research that explores the gendered nature of participation and spectatorship and the assumed link between the two. Using a sociological theoretical framework inspired by Bourdieu’s habitus, field, and capital concepts, we intend to explore and interpret our data to examine how gender, sports participation, and sports spectatorship are interconnected. This offers a new reading of the matrix of participation and spectatorship within the field of sport and will address the gaps identified in the literature above.

3. Materials and Methods

The data for this study come from an original UK-wide online survey (N= 1105) of adults conducted over four weeks in April–May 2016. The survey was designed for academic research on cultural participation in the UK and administered in cooperation with the survey firm SurveyMonkey, which distributed the instrument to its recruited population of nationally representative respondents. Outside the United States, SurveyMonkey outsources recruitment to companies specializing in this process within the target country. Quota sampling was applied on age and gender, with race and region also monitored for representativeness, resulting in a sample that closely approximates the broader UK population. We did not apply post-stratification weights (e.g., raking to the 2021 UK census), as the sample achieved already matched census benchmarks across these dimensions. The only notable divergence was a modest underrepresentation of older respondents, which we acknowledge as a limitation. Because our analyses focus on associations and patterns of participation and spectatorship, rather than producing nationally representative prevalence rates, unweighted data remains appropriate.

Therefore, this survey utilizes non-probabilistic sampling methods to achieve a sample that reflects the UK population. These methods, such as online population recruitment, may suffer from sampling biases and have certain limitations [60]. However, these biases (e.g., from internet access and age-related usage differentials) have substantially decreased over the past two decades. Online survey methods are increasingly common because of other strengths, such as lessened interview bias and increasingly high non-response rates and biases for traditional methods, which has seen more social researchers shifting towards employing these methods in their survey mixes [61].

The resulting sample closely mirrors the results of the UK census with respect to all four of these variables (age, gender, race, and region). Compared to the 2011 final UK census results (the latest census for which final results are known for all four UK nations at the time of submission), the only one of these four variables that has subcategories that differ to a notable degree (more than 0–3 percentage points) between sample and census data is for the age variable. While 18–44-year-olds are virtually identically represented between the sample and the UK population, there is a slight overrepresentation of 45–59-year-old respondents in the sample compared to the population (29.5% vs. 24.8%) and a slight under-representation of 60+ respondents (23.6% vs. 28.5%). This potentially could be an example of a limitation precipitated by the online nature of the survey. However, even with this one notable divergence in representation, we argue that the overall resemblance to census data in the sample illustrates its representativeness.

The analyses of direct sports participation and spectatorship in this paper use two survey questions asking respondents whether they engaged ‘sometimes or often’1 in direct sports participation or watching sports (online, on TV, or in person). Therefore, we unfortunately cannot distinguish between the mode or location of spectatorship, but rather only between more passive and active forms of participation. Similarly, while using similar language to other prominent survey instruments [62], we acknowledge that response options such as ‘sometimes or often’ are subjective and may be inconsistent between respondents. While it is true that survey respondents may not be able to remember more precise instances, future survey instruments may benefit from using more granular frequency scales to try and capture engagement more precisely.

For both direct sports participation and spectatorship, we chose ten sports. These ten were chosen for two primary reasons. First, one of our objectives is to assess the relationship between participation and spectatorship. The sports listed in the survey (and written in an additional space) slightly differ regarding participation and spectatorship. We have therefore chosen the ten sports included for both, which allows us to make direct comparisons across the two domains. The second reason concerns the quantitative methods we seek to deploy. Although we have a large sample, it is not large enough to efficaciously perform advanced statistical analyses on a list of sports much more extensive than ten, especially as we seek to assess women and men separately for a better comparison of gendered patterns of engagement. Finally, these ten sports represent popular participatory and spectator sports, capturing a mix of individual and team activities, sports with traditionally male associations (e.g., football, rugby, boxing) and those with higher female engagement (e.g., swimming, tennis), as well as activities that vary in accessibility and capital intensity (e.g., jogging; golf). We acknowledge that this selection necessarily omits some other popular sports, such as netball, hockey, or darts, which may understate certain forms of gendered or classed engagement. However, we believe that the ten sports selected capture a sufficiently broad spectrum of UK sporting culture while enabling robust relational analysis. Similarly, although understanding whether participation and spectatorship for each individual sport is related (i.e., do people play the same sports they watch), this was not pursued because several sports, particularly among women, had small subsample sizes, making sport-specific correlations statistically less reliable. Moreover, the objective of this analysis was to capture overall patterns of engagement across sports rather than to test sport-by-sport correspondences, which would require a substantially larger and purpose-built sample. However, we recognize the policy value of this understanding and propose future research seek understanding of these relationships.

To analyze this data and answer the guiding research questions of this paper, we have primarily used statistical measures of association and models of latent-class analysis (LCA). Measures of association were used to analyze if there is a relationship between participatory sport and spectator sport, along with the strength of this relationship. To assess the overall relationship between participation and spectatorship, we created two count variables: the total number of sports in which each respondent reported participating, and the total number of sports each reported spectating. We then examined the correlation between these measures as an initial test of association, prior to the latent-class analyses. This method is used to answer our first research question. It has implications for theories and policies that are premised upon (or dismiss) such a relationship, such as increasing spectatorship to get people more directly active in participatory sport. Latent-class analysis allows for the assessment of patterns in the data and is therefore used here to identify patterns in sports participation and spectatorship to answer our second research question. This is important for identifying groups of sports participation and spectatorship patterns, which we use to answer our second research question. This approach will help us understand, for instance, if certain groups participate in only certain sports or if there are groups that have broader or narrower breadth in their sports engagement, both in volume or specific combinations. LCA could also potentially reveal patterns of accessibility to specific sports.

Because LCA takes a modeling-based approach to identifying typological groupings in the data, it has the benefit of allowing goodness-of-fit assessments, contrary to other methods of cluster analyses. LCA also avoids model assumption biases of other methods, which are often violated in practice and represents an updated replacement for cluster analysis [63]. From the patterns found through our latent-class analysis, each typological group of participation and spectatorship can be assessed for relationships. We also report cross-tabulated probabilities by class, which provide the basis for interpreting the distinct profiles represented by each latent class. Finally, after estimating latent class models, we examined the sociodemographic make-up of each class using descriptive cross-tabulations. These breakdowns are presented to provide context for interpreting class profiles and help answer our second research question. All the statistical tests of this paper were completed separately for men and women to answer our third research question regarding whether the relationships and patterns of direct sports participation and sports spectatorship differ for women and men.

4. Results

4.1. Relationship Between Sports Participation and Spectatorship (RQ1)

Table 1. Survey frequencies for direct participation and spectatorship variables.

Participation	Women (W)	Men (M)	Difference
Swimming	36.8%	27.9%	W + 8.9%
Cycling	23.1%	28.9%	M + 5.8%
Jogging	22.1%	23.6%	M + 1.5%
Tennis	6.7%	10.4%	M + 3.7%
Golf	3.9%	11.1%	M + 7.2%
Football (soccer)	3.0%	19.1%	M + 16.1%
Boxing	1.9%	4.7%	M + 3.8%
Basketball	1.8%	5.7%	M + 3.9%
Cricket	1.2%	8.3%	M + 6.1%
Rugby	0.4%	7.0%	M + 6.6%
Average (all sports)	2.43	2.54
Spectatorship	Women	Men	Difference
Tennis	33.1%	32.1%	W + 1.0%
Football	30.8%	52.1%	M + 21.3%
Athletics	27.0%	26.6%	W + 0.4%
Rugby	26.4%	44.0%	M + 17.6%
Swimming	13.8%	10.9%	W + 2.9%
Cycling	12.1%	17.0%	M + 4.9%
Cricket	12.1%	28.1%	M + 16.0%
Golf	8.9%	18.9%	M + 10.0%
Boxing	6.1%	24.5%	M + 18.4%
Basketball	4.0%	8.9%	M + 4.9%
Average (all sports)	2.14	3.48

presents the relative frequencies for sports participation and spectatorship for women and men in our sample. As it shows, while there are some sports with more minor differences in the participation rates for women and men, all but one of the ten sports are participated in at a higher rate by men. Only for swimming is the participation rate for women higher than for men. Conversely, football has the highest gender difference overall, where the participation rate of men is more than sixteen percentage points higher than for women. The results also show three sports where women spectate more than men—tennis, athletics, and swimming. However, these gaps are small compared to the spectating gaps for football, boxing, rugby, cricket, and golf, where men have much higher rates of spectating than women.

In terms of associations between direct participation and spectatorship, both women and men show a similar statistically significant relationship of moderate strength between the number of sports one plays and the number of sports one watches among the ten sports included in the analysis. However, the relationship for men is stronger. Specifically, the Pearson’s R correlation coefficient for the association between direct sports participation and sports spectatorship for women is 0.194 (p < 0.001), and for men is 0.360 (p < 0.001). Therefore, the more sports one plays, the more likely one may watch more sports, and vice versa. This is an important finding regarding existing literature, which has often found little association between these two [19]. This argument has been used as evidence to push against public policies that may be premised on such an association (e.g., the Olympic Games).

However, an analysis of association cannot tell the whole story of these relationships. For instance, it cannot tell us if a group or groups participate in all sports, with the rest participating in none, or if it is less of an all-or-nothing proposition. Understanding this relationship between participation and spectatorship will allow us to better understand the impact of one’s gendered sporting habitus, as suggested in the literature on gendered sports engagement. Therefore, we now turn to latent-class analyses for women and men’s direct sports participation and spectatorship to understand these variables better and answer our second and third research questions.

4.2. Women’s Sports Participation and Spectatorship Latent-Class Analyses (RQ2)

The model fit statistics for the LCA model show a two-latent-class model as the optimal solution for women’s direct sports participation (

Table A1. Model fit statistics for latent-class analysis of women’s direct sports participation.

		BIC (LL)	AIC (LL)	CAIC (LL)
Model1	1-Cluster	2965.0843	2921.6104	2975.0843
Model2	2-Cluster	2891.7064	2805.4112	2918.7064
Model3	3-Cluster	2938.0320	2798.9156	2978.0320
Model4	4-Cluster	2988.7898	2801.8521	3031.7898
Model5	5-Cluster	3032.2562	2797.4972	3086.2562

). We reach this conclusion because the Bayesian Information Criterion 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, as sometimes occurs in the third common information criterion, the Akaike Information Criterion (AIC).

Table 2. LCA profiles of women’s direct sports participation and spectatorship.

Participation	Inactive	Active	Overall
Cluster Size	0.7609	0.2391
Indicators
Basketball	0.0125	0.0334	0.0175
Boxing	0.0027	0.0719	0.0193
Cricket	0.0085	0.0241	0.0123
Cycling	0.0930	0.6711	0.2312
Football	0.0254	0.0437	0.0298
Golf	0.0307	0.0634	0.0385
Rugby	0.0013	0.0104	0.0035
Swimming	0.2513	0.7384	0.3678
Tennis	0.0208	0.2122	0.0665
Jogging	0.1113	0.5689	0.2207
Spectatorship	Non-spectator	Omnivore	Overall
Cluster Size	0.7301	0.2699
Indicators
Athletics	0.0827	0.7757	0.2697
Basketball	0.0233	0.0861	0.0403
Boxing	0.0418	0.1140	0.0613
Golf	0.0371	0.2306	0.0893
Cycling	0.0352	0.3524	0.1208
Football	0.1938	0.6177	0.3082
Rugby	0.1932	0.4573	0.2644
Swimming	0.0387	0.4080	0.1383
Tennis	0.1885	0.7166	0.3310

shows the composition of each latent class group of direct participation for the women in our sample. These groups probabilistically represent about 76% (0.7609) and 24% (0.2391), respectively. Table 2 shows the conditional probabilities of participation and spectatorship for each latent class. These probabilities indicate the likelihood that members of a given latent class are to engage in each sport, which allows us to substantively label classes. The first and biggest latent class group can be characterized by a lack of participation in any of these sports. This group, therefore, is primarily an ‘inactive’ group. Inversely, the second latent class group is characterized as an ‘active’ group by its widespread participation across all ten sports. This is primarily a measure of the breadth of participation. This is to say that for any particular sport, the level of commitment is not wholly known. However, the likelihood of broad participation in sports generally is how we arrive at this active label. Thus, women’s direct sports participation is characterized by an active and an inactive group. However, the most delineating sports in this classification are boxing, rugby, cycling, and tennis, while the least delineating is football.

In terms of the top-line social composition for these two groups (see

Table 3. Demographic and socioeconomic crosstabulations of latent class groups of participation.

	Inactive	Active
Age
18–29	18.4%	32.9%
30–44	21.5%	32.9%
45–59	33.1%	26.6%
60+	27.0%	7.7%
Income
£100k+	3.8%	5.6%
£50 to 100k	17.9%	28.0%
£25 to 50k	35.5%	40.6%
£10 to 25k	28.7%	14.7%
£0 to 10k	14.0%	11.2%
Education
Degree or higher	41.1%	57.1%
Less than degree	58.9%	42.9%
Race
Non-white	6.3%	4.2%
White	93.7%	95.8%
Regions
Scotland	9.3%	12.9%
North of England	21.6%	22.1%
Northern Ireland	2.0%	2.9%
Wales	3.6%	3.6%
Midlands & East of England	24.3%	20.0%
London & The South	39.1%	38.6%

), the inactive group is much older, slightly less likely to be white, and shows relative signs of lower average incomes and formation education than the active, more omnivorous group. The active group, by contrast, is much younger, more likely to be degree-level educated, and has relatively higher income levels.

Regarding women’s sports spectatorship,

Table A2. Model fit statistics for latent-class analysis of women’s sports spectatorship.

		BIC (LL)	AIC (LL)	CAIC (LL)
Model1	1-Cluster	5040.6646	4996.9810	5050.6646
Model2	2-Cluster	4687.4736	4605.7380	4708.4736
Model3	3-Cluster	4711.7859	4571.9985	4743.7859
Model4	4-Cluster	4732.7721	4544.9327	4775.7721
Model5	5-Cluster	4771.1930	4535.3016	4825.1930

again shows a two-latent-class solution as statistically the soundest for analysis, and for the same reasons as our women’s participation LCA. This LCA again shows unambiguously that the largest group comprises those who do not watch any of these sports, although this group is slightly smaller (73%) than the inactive participation group. We label this group ‘non-spectators’. Those in latent class two, comprising 27% of women in the sample, are most likely to watch every one of the sports. This group, therefore, represents an ‘omnivorous spectators’ group. It is again noteworthy that there are only two groups of women spectators. The specific sports seemingly the most delineating between these latent class groups are athletics, golf, cycling, swimming, and tennis. However, there appears to be a starker contrast across sports than for direct participation. This may be reflective of two separate but related dynamics by which women face higher barriers to access to participation in sporting activities based upon a community of shared interest, such as spectatorship, than for sports by which there are alternative acceptable motivations for inclusion, such as physical health and body maintenance.

In terms of composition for the spectatorship groups (

Table 4. Demographic and socioeconomic crosstabulations of latent class groups of women’s spectatorship (N = 1105).

	Non-Spectator	Omnivore
Age
18–29	23.2%	18.2%
30–44	25.7%	19.7%
45–59	30.1%	35.8%
60+	21.0%	26.3%
Income
£100k+	4.1%	5.3%
£50 to 100k	19.2%	24.8%
£25 to 50k	37.1%	35.3%
£10 to 25k	25.3%	24.8%
£0 to 10k	14.3%	9.8%
Education
Degree Plus	42.6%	50.9%
Less than degree	57.4%	49.1%
Race
Non-white	5.6%	6.6%
White	94.4%	93.4%
Regions
Scotland	11.2%	6.7%
North of England	20.7%	25.2%
Northern Ireland	2.2%	2.2%
Wales	4.0%	2.2%
Midlands & East of England	21.6%	28.9%
London & The South	40.2%	34.8%

), women in the first non-spectating group are slightly younger, with relatively lower incomes, relatively more likely to live in Scotland, the Midlands, London, and the south, and less likely to have a degree than those in the omnivorous spectator group.

4.3. Men’s Sports Participation and Spectatorship Latent-Class Analyses (RQ2)

Moving on to the latent-class analyses for men’s direct sports participation and sports spectatorship, the model fit statistics for the direct participation LCA model (

Table A3. Model fit statistics for latent-class analysis of men’s direct sports participation.

		BIC (LL)	AIC (LL)	CAIC (LL)
Model1	1-Cluster	4291.8000	4107.2592	4301.8000
Model2	2-Cluster	4098.1691	3849.7068	4118.1691
Model3	3-Cluster	4081.8687	3807.2787	4109.8687
Model4	4-Cluster	4114.2204	3795.5176	4157.2204
Model5	5-Cluster	4158.6191	3794.1759	4212.6191

) show a three latent class model as the optimal solution because of the optimal model fit statistics. This suggests groups of participation that comprise approximately 51%, 43%, and 6% of the sample, respectively (see

Table 5. LCA profiles of men’s direct sports participation and spectatorship.

Participation	Inactive	Moderately Active	Active Omnivore	Overall
Cluster Size	0.5116	0.4250	0.0634
Indicators
Tennis	0.0021	0.1428	0.6336	0.1019
Football	0.1129	0.1960	0.7660	0.1896
Jogging	0.0729	0.3966	0.5500	0.2407
Rugby	0.0018	0.0724	0.5816	0.0685
Swimming	0.0447	0.5050	0.7198	0.2831
Cycling	0.0732	0.4754	0.7673	0.2881
Cricket	0.0560	0.0394	0.5949	0.0831
Golf	0.0726	0.1024	0.4018	0.1061
Boxing	0.0062	0.0717	0.2105	0.0470
Basketball	0.0044	0.0748	0.3771	0.0579
Spectatorship	Non-spectator	Moderate Spectating	Omnivore	Overall
Cluster Size	0.5544	0.2729	0.1727
Indicators
Athletics	0.1297	0.1273	0.9232	0.2660
Basketball	0.0571	0.0297	0.2833	0.0887
Boxing	0.1015	0.2684	0.6704	0.2453
Cricket	0.0304	0.5648	0.6380	0.2811
Cycling	0.0505	0.1615	0.5660	0.1698
Football	0.2790	0.8077	0.8435	0.5208
Golf	0.0190	0.3478	0.4818	0.1887
Rugby	0.2318	0.6082	0.8404	0.4396
Swimming	0.0427	0.0402	0.4332	0.1094
Tennis	0.1502	0.3874	0.7631	0.3208

).

The first and biggest latent class group is again characterized by a lack of participation in any of these sports, with no exceptions. We, therefore, again label this latent class the ‘inactive’ group. Unlike women, there is a second largest group, which is not the most highly active group (at least in terms of breadth of participation). While the second largest participation group is likely to be active in many sports, this group is not predicted to be active in all sports, and the likelihood of participation is generally lower than the third latent class group. Therefore, this group represents a type of ‘moderately active’ participation. The final group is the ‘active omnivore’ group, which participates directly in sports for men. This group is by far the smallest group, and the active group for women in the sample was more than three times larger, with no moderately active group seen for women in the sample.

The most delineating sports between the moderately active and active omnivore groups are cricket, golf, rugby, tennis, football, and basketball. These six sports represent either those sports that are traditionally associated with middle and upper-class engagement (e.g., golf, tennis), team sports (e.g., football, basketball), or both (e.g., rugby, cricket). It is often those sports that are least accessible that are most critical in determining highly active omnivorous activity profiles, as those with elevated sources of capital resources are most able to include them in their lifestyle.

Jogging, swimming, and cycling are the specific sports that delineate the least between these groups. This, again, may reflect a couple of different dynamics. First, sports with a health motive may have more appeal because of a more explicit association with health. Second, these three individual fitness sports, especially jogging, are highly accessible compared to the sports of the omnivorously active group.

Regarding the top-line social composition for these three groups (

Table 6. Demographic and socioeconomic crosstabulations of latent class groups of participation. (N = 1105).

	Inactive	Moderately Active	Highly Active
Age
18–29	16.7%	20.4%	48.3%
30–44	21.4%	35.3%	24.1%
45–59	29.6%	25.7%	13.9%
60+	32.3%	18.6%	13.7%
Income
£100k+	9.7%	7.8%	6.9%
£50 to 100k	16.3%	24.4%	27.6%
£25 to 50k	40.3%	39.3%	37.9%
£10 to 25k	20.9%	21.1%	20.7%
£0 to 10k	12.8%	7.4%	6.9%
Education
Degree Plus	42.7%	56.1%	61.3%
Less than degree	57.3%	43.9%	38.7%
Race
Non-white	12.1%	15.6%	3.4%
White	87.9%	84.4%	96.6%
Regions
Scotland	11.3%	16.1%	3.4%
North of England	18.0%	19.1%	10.3%
Northern Ireland	2.0%	0.7%	3.4%
Wales	6.6%	1.9%	10.3%
Midlands & East of England	25.0%	19.5%	31.0%
London & The South	37.1%	42.7%	41.4%

), the inactive group is the oldest and least likely of the groups to be degree-level educated. The moderately active group possesses middle levels of education and age between the three groups, is more likely than other groups to come from non-white ethnic backgrounds, and is more likely to live in the North of England and Scotland. Finally, the group of the most omnivorously active men is the youngest group, most likely of the groups to be degree-level educated, most likely of them to come from white ethnic backgrounds, and most likely of the three groups to live in Wales, Midlands and East of England, and least likely to live in Scotland.

This final LCA again shows a three-class solution as most statistically defensible for men’s sports spectatorship, according to the model fit statistics (see

Table A4. Model fit statistics for latent-class analysis of men’s sport spectatorship.

		BIC (LL)	AIC (LL)	CAIC (LL)
Model1	1-Cluster	5964.6030	5921.3981	5974.6030
Model2	2-Cluster	5445.9111	5355.1807	5466.9111
Model3	3-Cluster	5410.7686	5245.7071	5442.7686
Model4	4-Cluster	5431.4885	5246.7483	5474.4885
Model5	5-Cluster	5480.0553	5282.5127	5534.0533

). As for direct participation, the substance and order of the three men’s spectating latent class groups again show three levels of spectating. There is a first group that is ‘non-spectators’, a second group that is more moderate spectators, and a third group that is ‘omnivorous’ in their patterns of spectatorship. Therefore, the primary differences between the men’s and women’s spectatorship latent-class analyses are the existence of the more moderate spectator and the fact that the omnivorous spectator group is smaller than for women.

The most delineating spectator sports for men were the Olympic sports of athletics, basketball, cycling, and swimming. At the same time, football and cricket were the least delineating, especially between the moderately and omnivorously active groups. It is notable that unlike for women, spectator sports for men show much more variable amounts of delineation between sports. However, this is slightly tautological as this is also what produced more latent class groupings than for women. This dynamic, however, is notable regarding the different gendered relationships towards accessing spectator sport and direct participation in sport.

In terms of composition for the male spectatorship groups (

Table 7. Demographic and socioeconomic crosstabulations of latent class groups of spectatorship.

Space	Non- Spectator	Moderate Spectating	Omnivore
Age
18–29	22.2%	19.8%	15.3%
30–44	31.0%	20.6%	29.6%
45–59	24.9%	28.2%	31.6%
60+	21.9%	31.3%	23.5%
Income
£100k+	7.3%	10.0%	10.3%
£50 to 100k	18.0%	25.4%	24.7%
£25 to 50k	40.2%	42.3%	34.0%
£10 to 25k	22.6%	16.9%	21.6%
£0 to 10k	11.9%	5.4%	9.3%
Education
Degree Plus	50.1%	51.5%	48.4%
Less than degree	49.9%	48.5%	51.6%
Race
Non-white	16.1%	10.8%	6.2%
White	83.9%	89.2%	93.8%
Regions
Scotland	15.3%	9.9%	10.3%
North of England	16.0%	25.2%	16.5%
Northern Ireland	1.8%	0.8%	1.0%
Wales	4.6%	3.1%	6.2%
Midlands & E of England	20.2%	25.2%	27.8%
London & The South	42.0%	35.9%	38.1%

), the first, non-spectating group is the most likely of the three groups to come from non-white ethnicities, most likely to live in Scotland, London and the South, and least likely to live in the Midlands and East of England. The moderate sports spectating group is most likely to live in the North of England and least likely to live in London and the south of England, has the highest proportion of the oldest age group but is otherwise not highly distinguished from the other two groups and falls between them in many categories. Additionally, age, income, and education are equally distributed amongst the three sports-watching groups for men, even though the non-spectating group appears to be slightly younger and have slightly lower incomes than the other two groups. The omnivore group is finally the most likely to come from white ethnic backgrounds, most likely to live in Wales, the Midlands, and the East of England.

4.4. Summary of Gender Differences (RQ3)

The gender-split results suggest several differences. Across both descriptive and latent-class analyses, men displayed a broader distribution of sporting engagement, spanning from inactive to highly omnivorous profiles, whereas women’s engagement tended to cluster more strongly in either low engagement or consistently active across both participation and spectatorship. The correlation between participation and spectatorship was moderate for men and weaker for women, suggesting that these domains are more closely linked within men’s sporting lives. Women’s participation and spectatorship patterns were also more concentrated in lower-cost and socially ‘feminine’ sports, consistent with the intersection of economic and symbolic constraints observed in the descriptive tables. In short, while men’s sport engagement appears more varied and flexible, women’s engagement may remain shaped by both structural and cultural barriers. These gendered contrasts directly address RQ3, highlighting that the patterns and nature of the participation–spectatorship relationship differ not only in magnitude but also in their underlying social composition.

5. Discussion and Conclusions

The results presented above support findings relevant to answering the guiding research questions of this research. RQ1 examined the overall relationship between participation and spectatorship, RQ2 explored the socio-demographic and class-based patterns shaping these behaviors, and RQ3 compared how these relationships differ for men and women. Together, the results illustrate how forms of economic, cultural, and social capital structure engagement across both fields of sport.

First, the results of this study show a significant positive correlation between direct sports participation and sports spectatorship in the sample. This finding contributes to the existing literature because few existing studies focus on finding a connection rather than studying these two dimensions of sports engagement separately. Additionally, they do not assess the gendered dynamics of this relationship. We find here that this relationship is stronger for men than for women. The results of this paper also show that the specific patterns of sports engagement are similar for sports participation and spectatorship but differ in this pattern for men and women. While some of these results may be able to be triangulated or presumed from the existing literature or other types of studies, the specific assessment of these dynamics from one representative sample and using methods of structural equation modeling (i.e., LCA) add additional data to our understandings of gender and sport.

Therefore, there appears to be a stronger relationship between sports participation and spectatorship than in previous studies, suggesting that the two fields are structured by comparable forms of capital accumulation, particularly cultural and economic capital, but that their accessibility is uneven across social groups. In other words, the rules of the field may be similar in that both require capital to participate and confer social distinction through engagement, but the capacity to mobilize that capital differs markedly by gender, age, and education. These differences reflect variations in habitus, which may make some forms of sport feel more ‘natural’ or legitimate for certain groups than others. This helps explain why, despite broadly similar field structures, men and women may experience and navigate them differently.

Likewise, it is important to recognize that the costs of participation mirror those of spectatorship and may be even more prohibitive in certain sports. Both watching and playing require varying levels of economic capital, from fees and equipment to travel and membership costs, and these financial demands are reflected in the class patterns apparent in the socio-economic composition of the different sports participation groups found in our analysis. These class gradients suggest that economic capital, alongside cultural and social capital in the form of gender norms and networks, jointly structure who can access different fields of sport engagement.

While our finding of a positive relationship between sports spectatorship and participation may suggest that increased levels of spectating may increase levels of direct participation, we are unable to say this for certain. Instead, these behaviors likely share common underlying factors, and further longitudinal or experimental research would be needed to determine whether increasing spectator engagement directly translates into higher levels of direct participation in sport. Moreover, the relative costs of watching various sports are an important complicating factor, because while attending a cycling road race is typically free, watching elite football (soccer) matches can be expensive due to high ticket prices and related costs. Such economic considerations may influence the decision to spectate and the subsequent translation of spectatorship into direct participation. In addition to in-person costs, differences in media accessibility also matter. Sports available via free-to-air terrestrial television (e.g., Six Nations rugby or most England football international matches) may allow for more casual or widespread engagement, whereas others, such as Premiership football, rugby, or boxing, often require costly paid-TV subscriptions. These variations in access not only shape who can spectate but may also affect the extent to which spectatorship fosters direct participation or requires a pre-existing sporting identity.

Concerning gender and sport, this is mediated through the gender-based power relations that have historically constructed and continue to predominate in sport. More broadly, gender-based power relations are also mediated through other forms of capital, such as economic capital resources. Both increased direct sports participation and sports spectatorship have previously been found to be predictive of elevated social, economic, and cultural capital resources (e.g., refs [20,47,48,50,51]). This paper’s results likewise find that those with elevated economic and cultural capital (education) resources are more engaged in sport. Thus, women must overcome the symbolic and social capital barriers to sporting entry and, more generally, economic barriers, which may be unequally distributed between men and women.

Therefore, it might be presumed that the integration of sport as a valuable and valid activity for women is less established. Thus, the reflection in participation rates is not as pronounced as for men. Indeed, the correlation between participation and spectatorship is greater for men, suggesting greater transmissibility or similarity in how resources and access accrue between these two domains of sport. These findings are thus disparate from much of the existing scholarly literature, which finds little connection between these two domains of sports engagement (e.g., refs [18,19]) or at least provides solid evidence for one side of this debate. They also support the assumptions of policymakers who argue for this connection regarding policies that are premised upon it.

From the results of the LCA, a greater proportion of women participate and spectate at a higher level than men, who have other profiles of engagement with sport available to them. In line with Bourdieu’s [31] work, more men broadly participate in sport and their participation is mediated through a broader range of capital-based cultural profiles, with certain capital-intensive sports not participated in by the middle and moderately active group. However, more women have a necessarily committed relationship to both sport participation and spectatorship broadly. This is potentially counter-intuitive if the cultural paradigm of women’s inactivity is assumed and unchallenged.

Using the concept of habitus to explore this notion is illuminating and provides a lens to explore how capital and choice are associated here in this field. The habitus often aligns with the field it is in, reproducing and stabilizing dominant cultural narratives—fearing the potential consequences of challenging the status quo. For young men, as demonstrated by Metcalfe and Lindsey [7], engagement with sport carries social significance and capital. In this regard, men do not have to actually be good at sport. However, engagement with sport, even in a spectating form, reinforces masculinity and forms part of their construction of a ‘successful’ masculine identity. This could be one reason, for instance, why the spectating disparities between men and women persist, even if, in theory, there is wide availability for actual consumption for all genders.

Furthermore, our data indicate that culturally constructed notions of gender acceptability critically shape women’s sports engagement. In line with Hargreaves [25] and Connell [64], women are more inclined to participate in or spectate sports that conform more closely to traditional feminine norms—such as swimming, tennis, and athletics—while sports like rugby, boxing, and football, which embody dominant masculine symbolism, remain less accessible. This selective engagement constrains women’s participation and reinforces a narrow range of acceptable activities. The lack of capital available for women engaging in sport means that those who choose to participate and/or spectate must have elevated levels of capital, as seen in the results of this analysis, or potentially do so because they really want to for the social benefits of doing so are negligent. Following culturally dominant norms that have historically equated sport with masculinity [23,25,64], we argue that women who engage in sport participation and/or spectatorship are, to an extent, challenging cultural expectations, and as such, are required to be confident that sport is integral to their sense of self, in order to present a non-typical and transgressive gendered self.

While not a guiding research question or absolutely claimed using this data, our results also suggest very low overall rates of sports participation in the UK that should be continuously updated, particularly in the post-COVID-19 environment. Across the sample for this paper, the largest group for both participation and spectatorship, for men and women, was for non-participation (76% for women and 51% for men) and non-spectatorship (73% for women and 55% for men). This is potentially worrying and reflects the broader social issue of an insufficient proportion of the UK population being active for health benefits. Sport England [13] reports that 36.6% of the UK population is inactive for health benefits. Thus, our findings paint an increasingly pessimistic view of how prominently sport features in the lives of adult men and women. Our findings show that for both women and men, those participating in sport are also more likely to watch sport and vice versa, to a statistically significant level of moderate strength.

A potential limitation of these findings is that our data were collected in 2016, prior to substantial shifts in the sport landscape since 2020, including the expansion of streaming platforms, the disruption of COVID-19, and record-breaking visibility for women’s sport. These developments have reshaped both media access and cultural narratives around gender in sport [65,66]. Accordingly, our results should be understood as evidence of structural relationships between participation and spectatorship, rather than precise estimates of current rates of engagement. Future research using post-2020 data will be essential to test whether these more recent transformations strengthen, weaken, or alter the gendered associations identified in this paper. We additionally note that the sociodemographic breakdowns of latent class groups are descriptive only. They provide useful context for interpreting class profiles, but we cannot make claims about statistical significance or generalizability of these differences. Future research with larger samples could formally test these associations.

Therefore, the results of this study generally support those policymakers’ assertions who argue that there is a strong link between sport participation and spectatorship in the UK, and depending on the specifics of such efforts, funds allocated towards related programs may not be wasted in their purpose. However, we ultimately cannot draw a definitive conclusion or causal claim for this relationship without further study in this area, including methods that include experimentation for this question. Additionally, this relationship is gendered and is not quite as strong for women, so any policy program based upon this premise may rebound to the exacerbation of gender imbalance in these areas. In analyzing how women in this study are confined to participation/non-participation sporting profiles, thus facing barriers by which they may have to demonstrate more commitment and breadth in their sport participation and spectatorship, the phenomenon that, even with elevated levels of other forms of capital, they may need to really want to could also apply to other areas of society. It could be argued that for other gender-typical activities, participation may similarly require little personal sacrifice or investment; therefore, participation may reflect the patterns of men in this study (i.e., broader general engagement). In contrast, activities that may challenge cultural norms require individuals to show higher levels of commitment to accrue requisite levels of symbolic capital. Therefore, this type of engagement depends on the threat of the activity to the socially constructed sense of cultural norms. It could be gender-, sexual orientation-, social class-, or race-dependent, and further study of such dynamics provides an important and promising area for future research into both sport and other areas of culture.