Involvement in sport and social connectedness

Abstract

This paper explores the relationship between involvement in sport and non-sport community organisations and social connectedness. Data were collected on types of community involvement, selected demographic variables and social connectedness. The findings support the contention that involvement in sport organisations is associated with increased levels of social connectedness. Sport involvement was found to be a predictor of social connectedness, while involvement in non-sport community organisations was not. The study also found that the tenure and intensity of involvement in organisations were not significantly associated with social connectedness scores.

Keywords

participation social capital social connectedness sport involvement well-being

For more than 25 years social capital has been used as a framework to conceptualise the way in which social relations generate outcomes for individuals, organisations, communities and nations (cf. Bourdieu, 1986; Bourdieu and Wacquant, 1992; Coleman, 1988; Lin, 2001; Portes, 1998; Putnam, 1993, 1995a, 1995b, 2000; Woolcock, 1998). Within these debates there appears to be general agreement that social networks (together with relationships characterised by reciprocity and trust) are central to the generation and mechanisms of social capital. These debates have been based on two broad approaches, the first emphasising the notion that access to resources and investment of resources with an expected return are central to the concept of social capital (Bourdieu, 1986; Bourdieu and Wacquant, 1992; Lin, 2001; Portes, 1998). An often-cited example of this view is that held by Bourdieu (1986: 248), who stated that social capital is ‘the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalised relationships of mutual acquaintance and recognition’. The second approach has emphasised that the outcome of social capital is the ability of people to work together through enhanced communication, cooperation and positive collective action (Putnam, 2000; Woolcock, 1998). Indeed, as (Putnam, 2000: 18–19) stated ‘the core idea of social capital theory is that social networks have value’.

The subsequent take up and use of these social capital concepts in policy circles connected to sport (although not confined to it) is illustrated by the French Commission on the Measurement of Economic Performance and Social Progress (2009), which noted that sport plays a role in the creation of social capital and that the latter is ‘essentially’ about social connections, social networks and the benefits they confer. Indeed, the relationship between social capital and sport has attracted increasing research attention in recent years (Blackshaw and Long, 2005; Brown, 2006; Coalter, 2007a, 2007b; Collins, 2003, 2005; Dyreson, 2001; Harvey et al., 2007; Jarvie, 2003; Nicholson and Hoye, 2008; Seippel, 2006; Sharpe, 2003, 2006). Many of these efforts have focused on conceptualising the potential links between social capital and sport, sparked in no small way by the seminal work of Putnam (2000) in analysing in part the decline in bowling leagues in America.

At the same time, policymakers in westernised countries continue to claim that sport is somehow special, acting as a ‘social glue’ that binds communities (cf. Commonwealth of Australia, 2001, 2008; Sport Canada, 2002; Sport England, 2004). Community sport organisations are often credited with facilitating a range of social benefits and outcomes: community capacity building; reducing crime and youth delinquency; empowering disadvantaged groups; improving confidence and self-esteem; and increasing social integration and cooperation (Long and Sanderson, 2001). Strong support for the value of sport can be found in the European Union’s Commission on the European Communities white paper on sport, which argued sport is a social institution with a range of social benefits, and articulated the potential of sport to facilitate social inclusion, integration and equal opportunities (Commission on the European Communities, 2007).

As Hoye and Nicholson (2008, 2009) demonstrated, social capital and its related concepts have been used to explain, justify and legitimise instrumental sport policies, yet they also argued the link between sport and social capital is often simply assumed, without strong empirical evidence. Sport’s relatively unquestioned role in developing social capital, as articulated by organisations such as the European Union and various national governments, appears to be based on long-held assumptions and generally accepted affirmations of the value of sport; however, the mechanisms for how sport delivers such outcomes is largely unknown. Coalter (2007a) argued that the evidence that does exist is beset by conceptual and methodological weakness.

The concept of social capital has not escaped criticism, with authors such as Fine (2010: 5) stating that its rise to prominence has been as a result of an ‘intellectual malaise within academic life’ and that the idea of social capital has become ‘definitionally chaotic, as it is imbued with so many different variables, approaches and applications’ (Fine, 2010: 5). Despite these criticisms, the lens of social capital is still applied by researchers in the quest to understand how individuals, groups and organisations work collectively and the associated benefits of such collective action. Like previous authors (Hoye and Nicholson, 2012), we are drawing from the viewpoints of both Putnam and Bourdieu to explore how social networks are created within the context of sport organisations.

It is important to note here that in response to this lack of evidence, there is an increasing amount of empirical research now being reported that has attempted to test the relationship between sport and aspects of social capital (see, for example, Atherley, 2006; Bradbury and Kay, 2008; Brown, 2008; Hylton, 2008; Long, 2008; Numerato, 2008; Okayasu et al., 2010; Persson, 2008; Tonts, 2005). However, many of these studies have utilised aggregated social survey data and have tended to use participation data as a proxy for social capital and/or have not separated associations with sport as against non-sport involvement (Coalter, 2007b).

This paper seeks to explore possible differences between sport and non-sport involvement in relation to a measure of social connectedness that aims to quantify the degree of interpersonal closeness experienced between an individual and their social world, as well as the degree to which they have difficulty in maintaining this (Lee and Robbins, 1995). As such, social connectedness is taken to measure part of an important marker of the social capital nexus, namely social network connection. While the scope of the paper is unable to provide evidence on the extent to which social connectedness is a cause and/or an effect of social capital production, it aims to help clarify logically prior questions in relation to whether involvement is associated with increased social connectedness and the extent to which this varies between sport and non-sport participation.

Social connectedness

In a recent review of the concept of social connectedness, Townsend and McWhirter (2005) identified a substantial number of benefits individuals accrue from being more socially connected: increased sense of well-being; increased self-worth; and better health. Social connectedness is an important factor in fostering positive human development, whereas a lack of social connectedness can lead to self-alienation, loneliness and a lack of meaning or purpose.

Conceptually, social connectedness has been approached in a number of different ways. It is generally considered a psychological construct and understood as playing a crucial role in identity development, mental health and well-being (Townsend and McWhirter, 2005: 193). Timpone (1998: 59) defined social connectedness as ‘the level of an individual’s integration into his or her social milieu and the fullness of the resulting associative networks’. Connectedness is also considered to manifest in different forms: ‘connectedness to the self; connectedness to others, including the social network of family, friends, colleagues, and other social groups; and connectedness to a larger meaning or purpose in life’ (Townsend and McWhirter, 2005: 193). Importantly, research into connectedness has focused on exploring questions around both the quantity and quality of relationships.

Recent efforts to develop a robust measure of social connectedness have been conducted by Lee and others (Lee et al., 2002; Lee and Robbins, 1995, 1998, 2000). Their work is based on Kohut’s (1984) theory of self-psychology, which posits that individuals have three needs: grandiosity (ambitions and the desire for accomplishment); idealisation (having ideals and goals); and belongingness (seeking to confirm a sense of belonging to something in order to avoid loneliness or alienation). The notion of belongingness is argued to comprise three aspects: companionship, affiliation and connectedness (Lee and Robbins, 1995). The efforts of Lee and others have led to the development of a social connectedness scale, an eight-item measure that captures ‘one’s opinion of self in relation to other people … in particular, the scale focuses on the emotional distance or connectedness between the self and other people, both friends and society’ (Lee and Robbins, 1995: 239). In this sense, the social connectedness scale consists of items that capture all three elements of belongingness: connectedness; affiliation; and companionship. Lee and Robbins (1995) argued that this scale captures those aspects of belongingness that Kohut (1984: 200) described as a sense of security and being ‘human among humans’. The social connectedness scale has been utilised in a number of studies (Williams and Galliher, 2006; Yeh and Inose, 2003; Yoon et al., 2008). Social connectedness has been found to be an important ‘psychological resource in personal adjustment and well-being’ (Lee and Robbins, 2000: 489) and that social connectedness ‘is equally salient in both women’s and men’s lives’ (Lee and Robbins, 2000: 487).

Efforts to identify the drivers of social connectedness have, however, proved to be more elusive. It is unclear in the literature whether particular voluntary associations contribute to making their members more or less socially connected (Stolle, 1998). While Timpone (1998) found that formal group membership was related to greater levels of social connectedness, he also identified that education, income and age were all also related to greater social connectedness. Stolle and Rochon (1998: 48) were more explicit, stating that ‘associational memberships have become the indicator of choice for examining the formation and destruction of social capital … [and that] … it seems plausible that some voluntary associations produce strong member-oriented bonds and therefore higher levels of personal civicness’. They further reported in their review of six surveys in three countries that ‘members of associations were frequently found to be significantly higher than non-members on all indicators of social capital’ (Stolle and Rochon, 1998: 54). Importantly for the purposes of this paper, they highlighted the need for more detailed research that can further specify the possible effects of different associational membership types on particular measures of (disaggregated) social capital (Stolle and Rochon, 1998: 57).

The potential for aspects of social connectedness to variously limit or support involvement in voluntary organisations has also been considered within leisure theory through the concept of constraints on leisure and the effects of leisure participation on group and social network formation. Crawford and Godbey (1987) conceptualised intrapersonal, interpersonal and structural constraints on individual leisure activity, and in their later work developed an integrated hierarchical model comprising these barriers (Crawford et al., 1991). Low levels of social connectedness may here be seen as constituting part of an interpersonal leisure barrier for those individuals lacking in sets of relationships necessary for a successful transition to involvement and participation. Wood and Danylchuk (2011) showed how continued involvement (for women members of a golf club) relied in part on the formation and sustenance of ‘group culture’ and Stokowski (1990) used a social networks perspective to argue for the need to situate ‘social group’ cultures within broader networks of social connections that overlay and influence each other.

Extending the idea of the importance of a focus on sets of social relations rather than individuals, Wollebeak and Strømsnes (2008: 250) argued from a Political Science perspective that the primary contribution of voluntary organisations ‘lies not in socializing individual active members but in institutionalizing social capital’. This provides a theoretical case for the possibility that no clear patterns would emerge between types of organisational involvement in terms of social capital outcomes, as it would be more important that voluntary organisations exist as a focal point for collective action than for any possible micro-level interactions they may help foster between individuals.

In relation to involvement in sport organisations, Walseth (2008: 14) concluded that sport club environments are important for the creation of social capital, specifically as ‘team sports seem to value collectivity more highly than individuality’ – a statement in support of Putnam’s (2000) claims that sport clubs are ideal vehicles for bridging social capital. Some more recent work by Hoye and Nicholson (2012), Rosso (2010) and Tonts and Atherley (2010) has reinforced these claims. Tonts and Atherely (2010: 394) concluded that for many residents of rural Australia, ‘sport was one of the central arenas of local socialization, drawing together social networks and underpinning a sense of place’.However, sport organisations were also found to be exclusionary, particularly in relation to indigenous people and women, a point they highlight as being in ‘stark contrast to the notions of inclusivity and egalitarianism’ (Tonts and Atherely, 2010: 394) often associated with sport. Rosso (2010: 189) concluded that the role of social capital was central in the development of women’s soccer in Adelaide, South Australia, as evidenced by the establishment of social networks that facilitated the transfer of ‘coaching knowledge and skills, player development, delivery of programmes and management of organisations’ between national, state and regional soccer organisations. Finally, Hoye and Nicholson’s (2012) study of country race clubs in Victoria found that the social networks created through race clubs facilitated collective action within local communities. Following on from these works we hypothesise that:

community organisational involvement is associated with higher levels of social connectedness;

sport-based community involvement is associated with higher levels of social connectedness than non-sport involvement.

Methodology

Participants

The sample consisted of 5655 individuals randomly selected from the white pages telephone directories of the state of Victoria, Australia (sampling frame). Participants were contacted by mail and invited to complete and return the questionnaire if they were aged 18 or older. One thousand, eight hundred and thirty-three useable returns were collected. A measure of the representativeness of the sample can be achieved by comparisons to known population parameters. The proportion of the sample that was female was 48% compared to the population statistic of 50.9% (ABS, 2007); 73.4% of the sample was born in Australia compared to 71.1% for Victoria (ABS, 2002); the sample proportion that was married was 50.9% compared to the population (Victoria) of 50.1% (ABS, 2007). Differences between the sample and the population were also noted. The sample (median age = 55.0 years) was older than the population (Victoria median age = 37.0 years at June 2009 [ABS, 2009]). Twenty-two and a half percent of the sample was aged 65 years or over compared to 13.7% for Victoria in 2006 (ABS, 2007). Forty-three percent (42.8%) of the sample was aged over 55 years compared to 24.5% for Victoria in 2006 (ABS, 2007). While patterned unit non-response by younger age groups is a well-known effect in survey research (Groves and Couper, 1998), which can be exacerbated in the context of a mail-based survey using a sampling frame taken from telephone directories (Dillman, 2000), it nevertheless needs to be taken into account in assessing the results reported here. The sample also contained proportionately fewer people working full time (44.3%) compared to the population (Victoria = 60.1% [ABS, 2007]), although the median individual annual income was higher (sample median = A$30–60K compared to Victoria median = A$23,712 [ABS, 2007]). The sample had proportionately more Victorians who lived outside Melbourne (51.6%) than the population of Victoria in 2004 (27.6%) (ABS, 2005).

Measures

Social connectedness

Social connectedness was measured with the Social Connectedness Scale (Lee and Robbins, 1995), which measures the degree of interpersonal closeness experienced between an individual and their social world, as well as the degree to which they have difficulty in maintaining this. The scale consists of eight items rated on a six-point Likert Scale (1= ‘strongly agree’ to 6= ‘strongly disagree’). Example items are: ‘I don’t feel related to anyone’ and ‘I feel disconnected from the world around me’. Higher scores represent a stronger sense of social connectedness. The eight items have been shown previously to have a scale reliability score of alpha = .91 (Lee and Robbins, 1998). In this study, Cronbach’s alpha was .96.

Community involvement

Overall community participation was measured with a single item that asked: ‘Are you currently a member of any formal organised community group?’ Respondents who answered ‘yes’ were classified as members and respondents answering ‘no’ were classified as non-members. Members comprised 54.8% (849) of the sample and non-members 45.2% (699). Members were asked subsequent questions that directed them to give the number of groups to which they currently held membership and to choose a type for each organisation from the following list: sport; social services; culture/arts; education; health; environment; housing; law; philanthropy; international; or religion.

Data were also collected on levels of involvement in community organisations. Measures of involvement were length of time involved with the organisation/s and the number of hours per week spent engaged with the organisation/s. Demographic variables included age, gender, employment status, education level, marital status, whether the respondent was living alone or with a partner, gross annual income, housing arrangements and country of birth.

Procedure

The sample was surveyed by means of self-administration questionnaires via direct mail with two reminders, following the standard ‘tailored design method’ for mail surveys recommended by Dillman (2000). The survey was administered between February and March 2009. A total of 1833 responses were received giving a response rate of 32.4%, which is at the high end of the response range for similar surveys (Kaplowitz et al., 2004: 98).

Analysis

The data were analysed with SPSS (PASW Statistics 17) software, using both univariate and multivariate procedures. The design involved the use of a cross-sectional survey that gave access to a large and representative sample for which the key measures of social connectedness and community involvement could be measured and generalised to the population group (adult Victorians). As the key dependent variable (social connectedness) was measured at the ratio level, the analysis design was predicated on a two-stage procedure: firstly, the search for bivariate associations between potential predictor variables and the dependent variable; and secondly, assuming that significant results had been obtained in the first stage, the search for patterns of primacy amongst groups of predictors. The first stage was achieved through the application of simple measures of association having regard to the fact that several independent variables were measured at nominal (for example, gender, birthplace and partner arrangement) or ordinal (for example, educational achievement and income) levels. The second stage was completed through Ordinary Least Squares (OLS) multiple regression. This allowed the use of categorical explanatory variables (including the key involvement variable) in the models by means of dummy variable creation.

Results

Initial analysis of missing values showed that the item measuring income had a 5.2% non-response, the item on community involvement, 4.3%, living with a partner, 3.6%, and all other items had missing values in the range 0.1–2.2%. The income, community involvement and partner items were then assessed for any differences in missing values between categories of the other demographic variables and it was concluded that the data were missing completely at random (MCAR). Given the relatively large size of the sample and the MCAR status, complete case analysis was undertaken using listwise deletion to exclude any case with one or more missing values. Data screening resulted in the deletion of 285 cases, leaving 1548 cases available for analysis. The loss of some statistical power was deemed acceptable given the assumption that MCAR status would lead to unbiased parameter estimates (Howell, 2008; Pigott, 2001).

Thirty-two percent of respondents were involved in a formal sport organisation. Of these, 67% had been involved with one such organisation in the previous 12 months. Twenty-three percent were members of two organisations, 6% three organisations and the remainder (4%) were members of four or more. For the single or primary organisation (in the case of respondents who were members of more than one organisation), the median number of years involved with the sport organisation was 8 years and the median number of hours involved per week was 4 hours.

Thirty-one percent of respondents were involved in another or other (non-sport) types of formal community-based organisations (‘hobby clubs, Country Fire Authority, State Emergency Service or other group’). Of these, 56% had been involved with one such organisation in the previous 12 months. Twenty-five percent were members of two organisations, 12% three organisations and the remainder (7%) were members of four or more. For the single or primary organisation (in the case of respondents who were members of more than one organisation), the type of organisation was ‘social services’ (22%) followed by ‘religion’ (21%), ‘education’ (12%), ‘culture/arts’ (12%), ‘philanthropy (12%), ‘environment’ (10%), ‘health’ (7%), ‘housing’ (1%) ‘law’ (1%) and ‘international’ (1%). The median number of years involved with the single or primary organisation was 8 years and the median number of hours involved with the organisation per week was 3 hours.

The mean score for the social connectedness scale for the sample reported here was M = 40.88 (SD = 7.03; n = 1548). This is comparable to Lee and Robbins’ (1995) reported mean score of M = 38.85 (SD = 8.09; n = 313) using their original eight-item scale in their study of college students. However, dimensionality testing of the social connectedness scale (Lee and Robbins, 1995), specifying a one-factor congeneric model with eight indicators, found that the model did not fit our data well: χ² (20) = 330.97, p = .000, root mean square error of approximation (RMSEA) = .105 (.096, .461), goodness-of-fit index (GFI) = .946, Tucker–Lewis index (TLI) = .955 and comparative fit index (CFI) = .968. Following the approach recommended by Jöreskog (1993), an examination of the standardised residual covariance matrix indicated that four of the items should be omitted. The one-factor congeneric model of social connectedness was re-specified as a latent variable with four reflective indicators. The model fitted the data: χ² (2) = 4.00, p = .135, RMSEA = .025 (.000, .062), GFI = .999, TLI = .999 and CFI = .999. The first of the four omitted items – ‘I don’t feel related to anyone’ – can be considered as ambiguous and could be interpreted by respondents as meaning familial relationships rather than a sense of social connectedness. Similarly, the second omitted item – ‘Even among my friends, there is no sense of brother/sisterhood’ – was also omitted. The third omitted item – ‘I catch myself losing all sense of connectedness with society’ – was considered to not necessarily infer a lack of connectedness with people and was omitted. The fourth omitted item – ‘I don’t feel I participate with anyone or any group’ – was omitted on the basis that it emphasised active participation in an activity rather than connections with others. The four items retained in the modified scale of social connectedness that were used in subsequent analyses were (figures in parenthesis show standardised regression weight in the four-item model from the latent variable to the item): ‘I feel disconnected from the world around me’ (β =.770); ‘Even around people I know, I don’t feel that I really belong’ (β =.849); ‘I feel so distant from people’ (β =.882); and ‘I have no sense of togetherness with my peers’ (β =.835). The mean score for the modified social connectedness scale (four items) for the sample reported here was M = 20.16 (SD = 3.72; n = 1548). Cronbach’s alpha for the modified scale in this study was .90.

Involvement data were used to construct a community involvement measure with four categories or types: sport involvement only (23.5%; n = 364); sport and other involvement (12.2%; n = 189); other involvement only (19.1%; n = 296) and no involvement (45.2%; n = 699).

Table 1 shows results by community involvement type for variables where the difference between groups was significant. There were no significant community involvement group effects for marital status, birthplace, income, place of residence, religious attendance or housing tenure.

Table 1.

Descriptive results for community organisation involvement groups.

Sport only			Other only		Sport and other		No involvement		Test
Measure	N		N		N		N
Age (yrs)	364	50.06 (M) 15.06 (SD)	189	58.06 (M) 15.07 (SD)	296	53.62 (M) 15.10 (SD)	699	52.55 (M) 15.42 (SD)	F(3, 1544) = 15.7**
Age groups									62.78**(x²)
18–25	12	3.3%	3	1.0%	5	2.6%	19	2.7%
26–35	44	12.1%	20	6.8%	21	11.1%	89	12.7%
36–45	101	27.7%	47	15.9%	28	14.8%	123	17.6%
46–55	84	23.1%	56	18.9%	50	26.5%	184	26.3%
56–65	62	17.0%	68	23.0%	43	22.8%	141	20.2%
66–75	36	9.9%	67	22.6%	25	13.2%	85	12.2%
76+	25	6.9%	36	11.8%	17	9.0%	58	8.3%
Total	364	100%	296	100%	189	100%	699	100%
Gender									26.66^**(x²)
Male	242	28.3%	146	17.1%	106	12.4%	362	42.3%
Female	122	17.6%	150	21.7%	83	12.0%	337	48.7%
Education									28.66**(x²)
Other	25	6.9%	29	9.8%	17	9.1%	88	12.7%
Secondary	178	48.9%	113	38.2%	67	35.8%	307	44.4%
Diploma	53	14.6%	48	16.2%	30	16.0%	115	16.6%
Bachelor degree	70	19.2%	62	20.9%	47	25.1%	111	16.0%
Postgrad. degree	38	10.4%	44	14.9%	26	13.9%	71	10.3%
Total	364	100%	296	100%	187	100%	692	100%
Employment									58.15**(x²)
Not in labour force	90	24.7%	133	44.9%	61	32.3%	228	32.6%
Unemployed	8	2.2%	3	1.0%	2	1.1%	25	3.6%
Part time	61	16.8%	63	21.3%	53	28.0%	135	19.3%
Full time	205	56.3%	97	32.8%	73	38.6%	311	44.5%
Total	364	100%	296	100%	189	100%	699	100%
Living arrangement									15.88**(x²)
Lives alone	77	21.2%	93	31.4%	53	28%	225	32.2%
Lives with partner	287	78.8%	203	68.6%	136	72.0%	474	67.8%
Total	364	100%	296	100%	53	100%	699	100%
# orgs in last year	364	1.68(M)1.06(SD)	296	1.84(M)1.50(SD)	189	3.17(M)1.83(SD)	699	0	F(2, 846) = 74.92**
Hours per week involved	364	8.55(M)7.23(SD)	296	7.52(M)8.80(SD)	189	12.54(M)9.47(SD)	699	0	F(2, 846) = 22.25**
Total years involved	364	17.34(M)16.40(SD)	296	20.40(M)21.05(SD)	189	33.60(M)26.41(SD)	699	0	F(2, 846) = 40.10**
SCS(8 items)	364	42.02(M)6.32(SD)	296	41.53(M)5.78(SD)	189	41.61(M)6.68(SD)	699	39.81(M)7.03(SD)	F(3, 1544) = 10.2**
Modified SCS(4 items)	364	20.69(M)3.39(SD)	296	20.32(M)3.17(SD)	189	20.49(M)3.51(SD)	699	19.73(M)4.09(SD)	F(3, 1544) = 6.4**

SCS: social connectedness scale.

The significant age effect was firstly that the group ‘other only’ had the highest mean age (58.06 years) followed by ‘sport and other’ (53.62 years), ‘no involvement’ (52.55 years) and ‘sport only’ (50.06 years). Secondly, the majority (50.8%) of ‘sport-only’ members were aged 36–55 years, whereas 57.4% of ‘other-only’ members were aged over 55 years. Males were more likely to be in the group ‘sport only’ (28.3% of males compared to 17.6% of females), while females were more likely to be in ‘no involvement’ (48.7% of females compared to 42.3% of males) and ‘other only’ (21.7% of females compared to 17.1% of males). Other significant differences between the groups at the descriptive level were possibly reflections of these age differences to a greater or lesser degree, which were accounted for at the final multivariate stage of analysis.

There was a significant difference between completed education groups, with greater proportions of ‘other-only’ and ‘sport and other’ members in more educated levels – 39.0% and 35.8%, respectively, with a degree or above compared to 29.6% for ‘sport only’ and 26.3% for ‘not involved’. ‘Sport-only’ members were significantly more likely to be in full-time employment (56.3%) than either ‘other only’ (32.8%) or ‘sport and other’ (38.6%) members.

‘Sport-only’ members were more likely to be living with a partner (78.8%) than members of the other groups, with ‘no involvement’ having the smallest proportion (67.8%). As hypothesised, there were significant differences between groups for their scores on social connectedness: On the four-item modified social connectedness scale, the ‘sport-only’ group had the highest mean score (M = 20.69; SD = 3.39), followed by ‘sport and other’ (M = 20.49; SD = 3.51) and ‘other only’ (M = 20.32; SD = 3.17). All three involvement groups scored higher than the ‘no involvement’ group (M = 19.73; SD = 4.09).

A correlation analysis between possible predictors of social connectedness is provided in Table 2.

Table 2.

Zero-order correlations between predictors of social connectedness.

	Age	Gender	Born	Education	Partner	Income	Employ	Sport	Other	SC8	SC4
Age	1
Gender	–.152^**	1
Born	.071^**	–.052^*	1
Education	–.176^**	.007	.035	1
Partner	–.070^**	–.185^**	.015	.061^*	1
Income	–.180^**	–.225^**	–.052^*	.312^**	.146^**	1
Employ	–.431^**	–.248^**	–.020	.159^**	.102^**	.509^**	1
Sport	–.090^**	–.114^**	–.110^**	.039	.089^**	.093^**	.089^**	1
Other	.139^**	.045	–.080^**	.097^**	–.017	–.021	–.126^**	.046	1
SC8	.003	.072^**	–.038	.074^**	.127^**	.093^**	.034	.106^**	.065^*	1
SC4	.011	.057^*	–.026	.069^**	.121^**	.093^**	.029	.093^**	.042	.961^**	1

Correlation is significant at the 0.01 level (two-tailed).

Correlation is significant at the 0.05 level (two-tailed).

n = 1548.

Key to variables:

Age (in years).

Gender (0 = Male; 1 = Female).

Born (0 = Born overseas; 1 = Born in Australia).

Education (0 = Less than Degree; 1 = Degree and higher).

Partner (0 = Do not live with partner; 1 = Do live with partner).

Income (0 = below A$60K per annum; 1 = above $60K per annum).

Employ (0 = not in full-time work; 1 = in full-time work).

Sport (0 = Not a member; 1 = Member of community sport organisation).

Other (0 = Not a member; 1=Member of [non-sport] community organisation).

SC8: Social Connectedness eight-item scale (low to high); SC4: Social Connectedness four-item scale (low to high).

Correlation levels were acceptable for the multivariate procedures (shown below) and were below 0.25 in the majority of cases, except for the expected higher correlations between employment and income and employment and age. At the bivariate level, the strongest associations with the modified social connectedness scale were living with a partner (r = .121; p ≤ .001); income (r = .093; p ≤ .001) and sport involvement (r = .093; p ≤ .001). Results in Tables 3 and 4 show that none of the involvement measures (tenure, intensity or number of organisations) in either the sport involvement or other involvement type categories were significantly correlated with either the original or the modified social connectedness scales.

Table 3.

Zero-order correlations between involvement measures and social connectedness scales: sport members.

		Number org membership	Hours/week involvement	Total years involvement
SC8	Pearson correlation	–.007	.018	–.006
	Sig. (2-tailed)	.887	.731	.914
	N	364	364	364
SC4	Pearson correlation	–.010	–.002	–.022
	Sig. (2-tailed)	.846	.966	.672
	N	364	364	364

SC8: Social Connectedness eight-item scale (low to high); SC4 (Social Connectedness four-item scale (low to high).

Table 4.

Zero-order correlations between involvement measures and social connectedness scales: non-sport members.

		Number org membership	Hours/week involvement	Total years involvement
SC8	Pearson correlation	.030	.055	.090
	Sig. (2-tailed)	.609	.344	.121
	N	296	296	296
SC4	Pearson correlation	.029	.070	.095
	Sig. (2-tailed)	.621	.227	.102
	N	296	296	296

Notes: R² = .052 (p < .001), ** p < .001, * p < .05.

n = 1782.

SC8: Social Connectedness eight-item scale (low to high); SC4: Social Connectedness four-item scale (low to high).

A multiple regression analysis (OLS) with the modified social connectedness scale as the dependent variable was conducted, with the resultant models shown in Table 5. There was no multicollinearity, as indicated by the correlations (all under 0.5) in Table 2, and visual checks of the modelled variable distributions indicated that the normality assumption was met. Linearity and Independence of Residuals were checked through Scatter Plots and Normal P-P plots of Regression Standardised Residuals, and it was concluded that the heterogeneity assumption was met. No outliers were detected in this process. The procedure involved preparing variables as continuous and categorical with two categories or categorical dummy variables. As the key independent variable (involvement type) was dummy coded, a hierarchical method was used, together with the calculation of the R-square score change. The involvement predictor was loaded independently in a separate step so that its independent predictive effect could be measured. The result was the three-step model shown in Table 5. Model one entered ascribed and achieved predictor variables based on past research; model two added the involvement type dummy variable and model three added involvement measures. In common with the limitations of the multiple regression method, the analysis presented below can only indicate possible causality based on known association.

Table 5.

Multiple regression of involvement types and demographic variables on the modified (four-item) social connectedness scale.

	Model	B	Beta	Sig
1	(Constant)	16.853		.0001
	Female	.815	.109	.0001
	Age	.018	.076	.010
	Education	.506	.063	.015
	Full-time employment	.485	.065	.030
	Partner	1.114	.136	.0001
2	(Constant)	16.355		.0001
	Female	.891	.119	.0001
	Age	.019	.079	.008
	Education	.464	.057	.026
	Full-time employment	.480	.064	.031
	Partner	1.057	.129	.0001
	No affiliation vs. sport only	.952	.109	.0001
	No affiliation vs. sport and other	.717	.063	.017
	No affiliation vs. other only	.473	.050	.066
3	(Constant)	16.349		.0001
	Female	.926	.124^**	.0001
	Age	.018	.074^*	.016
	Education	.492	.061^*	.018
	Full-time employment	.499	.067^*	.025
	Partner	1.066	.130^**	.0001
	No affiliation vs. sport only	.841	.096^**	.006
	No affiliation vs. sport and other	.612	.054	.172
	No affiliation vs. other only	.408	.043	.204
	Number org membership	–.114	–.047	.224
	Hours/week involvement	.028	.059	.071
	Total years involvement	.003	.017	.621

Notes: Model one R² = .030 (p < .05); model two R² = .041 (p < .05); model 3 R² = .044 (p < .05).

p < .001; * p < .05.

n = 1548.

The first model loaded gender, age, education, employment and partner (see Table 2 ‘key’ above for a description of these variables). The second model added involvement coded as dummy variables, with ‘no involvement’ as the baseline group. The effect of loading the involvement variable in model two was an increase in the R-square score of .11 (from .03 for model one to .041 for model two). The final model added the three involvement measures (tenure, intensity and number of organisations in the last 12 months). The R-square value increased from .041 in model two to .044 for model three. Six variables had a significant predictive effect for social connectedness in the final model: being female (β = –.124, p ≤ .0001); living with a partner (β = .130; p ≤ .0001); involvement in a community sport organisation (β = .096; p = .006); being older (β = .074; p = .016); having full-time employment (β = .067; p = .025); and having a degree or above (β = .061; p = .018). ‘Sport and other’ involvement and ‘other-only’ involvement were not significant predictors of social connectedness and neither were the involvement measures of tenure, intensity and number of organisations in the last 12 months. The final model explained 4.4% of the variance in the modified social connectedness scale: F (11, 1547) = 6.36; p < .0001.

Discussion and conclusion

Before proceeding to a discussion of the findings of the study in relation to how various forms of involvement in community organisations are predictive of social connectedness, we would like to highlight some issues regarding the composition of the social connectedness scale of Lee and Robbins (1995). The original eight items were reduced to four through the testing of the congeneric model. These remaining four items still represented the three underlying elements of belongingness as argued by Lee and Robbins (1995): connectedness (two items – ‘I feel disconnected from the world around me’ and ‘I feel so distant from people’); affiliation (one item – ‘I have no sense of togetherness with my peers’); and companionship (one item – ‘Even around people I know, I don’t feel that I really belong’). The mean scores we reported for our four-item scale were consistent with Lee and Robbins’ (1995) eight-item scale, suggesting that the shortened scale is an effective measure of social connectedness. Further testing of the utility of this shortened scale is required, but the results from this study suggest that this four-item scale can be effectively deployed as a measure of social connectedness.

Using this modified scale, the study found that social connectedness was significantly and positively associated with the following predictor variables in order of magnitude: being female; living with a partner; involvement in a community sport organisation; being older; having full-time employment; and having higher educational attainment. The effect levels were, however, small as indicated by the Beta scores. While the level of explained variance at 4.5% (measured by the R-square score) was low, Newman and Newman (2000) point out that low R-square values are not unexpected in social scientific research, particularly when designs are cross-sectional and focused on group-based (rather than individual-based) comparisons. In addition, the measurement of complex constructs associated with social connectedness likely entails inherent error.

In relation to the hypotheses, the study showed that while involvement in one or more community sport organisations was a significant but weak predictor of higher levels of social connectedness (hypothesis 2), involvement in non-sport community organisations was not significantly associated with social connectedness scores (hypothesis 1). Support for hypothesis 2 may strengthen claims that involvement in sport fosters social connections over and above involvement in non-sport organisations, as argued by Walseth (2008). However, in not supporting hypothesis 1, the findings are at odds with the contention by Stolle and Rochon (1998) that associational membership of any type is associated with higher measures of social capital. Thus, the results do not provide conclusive evidence that sport somehow is a significantly more important context in which social connections are formed relative to other forms of associational life. Tempering this finding is the fact that we measured social connectedness, which is only one indicator of the social capital concept that covers a broader spectrum.

The finding that neither the intensity nor the length of time of organisational involvement were associated with higher social connectedness scores was unexpected in the light of previous work (Stokowski, 2009; Wood and Danylchuk, 2011). There are at least three possible explanations for this: firstly, that heightened social connectedness precedes involvement in community sport; secondly, that social connectedness is heightened through involvement with community sport within a short period of time subsequent to joining and that this is not substantially increased further over time as a member; thirdly, that other or additional measures of involvement are required. Of these, the last point is perhaps the most likely explanation. Our measures targeted the number of organisations in which membership was held over the previous 12 months, the average number of hours of involvement and the number of years involved. We did not measure the nature or quality of involvement and it is possible that there are differences between involvement roles in relation to effects on social connectedness. Seippel (2005: 261) has made the important distinction that ‘membership in voluntary sport organisations in practice implies very different social relations for those involved’ and, thus, social capital should be considered to be differentially distributed within organisations, with those in more administrative roles (i.e. board members or senior volunteer roles) having more social capital than others. A limitation of this study is the fact that we did not explicitly seek to examine the existence of such differences in social connectedness scores on the basis of roles individuals had within community sport.

If the second possible explanation – that higher social connectedness prompts community sport involvement – holds, then we would have expected it to also prompt other types of community involvement, which this study did not find. A qualifier to this statement also relates to our lack of more detailed data on the type of involvement. As Crawford et al. (2009) point out, interpersonal barriers to leisure participation may be more important for collective activities (many sports) and less so for solo activities (which might characterise a range of non-sport community organisations). If that is indeed the case empirically, then social connectedness might be a greater interpersonal barrier to sport-related leisure than to non-sport-related leisure practices.

In terms of a third explanation, it remains possible that social connectedness increases upon first undertaking involvement and that levels do not change significantly after that event. Assessing change in social connectedness over time in individuals joining sport organisations and the types and forms of involvement they experience could help to throw light on this issue. Ultimately, because this study relies on cross-sectional data, causal inference from regression analyses should be made with caution and treated as suggestive findings that may prompt further enquiry.

Following Coalter (2007b), we emphasise that while our study has provided evidence about whether involvement in sport influences people’s sense of social connectedness, important unanswered questions remain: does heightened social connectedness facilitate one’s entry into sport involvement (rather than the other way around) and if there is indeed something special about being involved in sport that fosters people’s social connectedness, what is it? Other questions that lay outside the scope of this design also await clarification in relation to this research inquiry: what role, if any, does geographical location have within the mix of predictors and are there sub-sets of age and/or gender for whom the effects of involvement and social connectedness are relatively more or less strong? It is likely that only oversampling, together with longitudinal research complemented by qualitative approaches, can provide answers.

In conclusion, this paper makes two major contributions. Firstly, this study has demonstrated that there appears to be appropriate valid and reliable measure of social connectedness that can be applied in future studies to further explore the question of causality. Secondly, the study has shown that even a low degree of involvement in sport was associated with increased social connectedness for individuals. Yet, crucially, the measurable effect was small and can be regarded as much as a caution against some of the more extravagant claims made for sport’s contribution to social ‘goods’ than as a confirmation of them.

Footnotes

Funding

This work was supported by the Australian Research Council and the Victorian Health Promotion Foundation (VicHealth).

References

Atherley

(2006) Sport, localism and social capital in rural Western Australia. Geographical Research 44(4): 348–360.

Australian Bureau of Statistics (ABS) (2002) 2015.2-Census of Population and Housing: Selected Social and Housing Characteristics for Statistical Local Areas, Victoria, 2001. Canberra, ACT, Australia: Australian Bureau of Statistics.

Australian Bureau of Statistics (ABS) (2005) 2015 3235.2.55.001-Population by Age and Sex, Victoria, Jun 2004. Canberra, ACT, Australia: Australian Bureau of Statistics.

Australian Bureau of Statistics (ABS) (2007) 2006 Census of Population by Age and Sex, Victoria. Canberra, ACT, Australia: Australian Bureau of Statistics.

Australian Bureau of Statistics (ABS) (2009) 3201.0-Population by Age and Sex, Australian States and Territories, June 2009. Canberra, ACT, Australia: Australian Bureau of Statistics.

Blackshaw

Long

(2005) What’s the big idea? A critical exploration of the concept of social capital and its incorporation into leisure policy discourse. Leisure Studies 24: 239–258.

Bourdieu

(1986) The forms of capital. In: Richardson

(ed.) Handbook of Theory and Research for the Sociology of Education. New York: Greenwood, pp. 241–258.

Bourdieu

Wacquant

LJD

(1992) An Invitation to Reflexive Sociology. Chicago, IL: University of Chicago Press.

Bradbury

Kay

(2008) Stepping into community? The impact of youth sport volunteering on young people’s social capital. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 285–316.

10.

Brown

(2006) The position of Australian community sporting organisations in the third sector: Membership profiles, characteristics and attitudes. Third Sector Review 12(2): 17–39.

11.

Brown

(2008) Community sport/recreation members and social capital measures in Sweden and Australia. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 165–186.

12.

Coalter

(2007a) A Wider Social Role for Sport: Who’s Keeping Score? London: Routledge.

13.

Coalter

(2007b) Sports clubs, social capital and social regeneration: ‘Ill-defined interventions with hard to follow outcomes’? Sport in Society 10(4): 537–559.

14.

Coleman

(1988) Social capital in the creation of human capital. American Journal of Sociology 94: 95–120.

15.

Collins

(2003) Sport and Social Exclusion. London: Routledge.

16.

Collins

(2005) Voluntary sports clubs and social capital. In: Nichols

Collins

(eds) Volunteers in Sports Clubs. Eastbourne: LSA, pp. 105–118.

17.

Commission on the European Communities (2007) White Paper on Sport. Brussels: Commission on the European Communities.

18.

Commission on the Measurement of Economic Performance and Social Progress (2009) Report by the Commission on the Measurement of Economic Performance and Social Progress. Paris: CMEPSP.

19.

Commonwealth of Australia (2001) Backing Australia’s Sporting Ability. Canberra, ACT, Australia: Commonwealth of Australia.

20.

Commonwealth of Australia (2008) Australian Sport: Emerging Challenges, New Directions. Canberra, ACT, Australia: Commonwealth of Australia.

21.

Crawford

Godbey

(1987) Reconceptualizing barriers to family leisure. Leisure Sciences 9: 119–127.

22.

Crawford

Jackson

Godbey

(1991) A hierarchical model of leisure constraints. Leisure Sciences 13(4): 309–320.

23.

Dillman

(2000) Mail and Internet Surveys: The Tailored Design Method. New York: John Wiley.

24.

Dyreson

(2001) Maybe it’s better to bowl alone: Sport, community and democracy in American thought. Sport in Society 4(1): 19–30.

25.

Fine

(2010) Theories of Social Capital: Researchers Behaving Badly. London: Pluto Press.

26.

Groves

Couper

(1998) Non-Response in Household Interviews. New York: John Wiley.

27.

Harvey

Levesque

Donnelly

(2007) Sport volunteerism and social capital. Sociology of Sport Journal 24: 206–223.

28.

Howell

(2008) The analysis of missing data. In: Outhwaite

Turner

(eds) Handbook of Social Science Methodology. London: SAGE, pp. 208–224.

29.

Hoye

Nicholson

(2008) Locating social capital in sport policy. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 69–91.

30.

Hoye

Nicholson

(2009) Social capital and sport policies in Australia: Policy transfer in action. Public Management Review 11(4): 441–460.

31.

Hoye

Nicholson

(2012) Life at the track: Country race clubs and social capital. International Review for the Sociology of Sport 47(4): 461–474.

32.

Hylton

(2008) Race equality and sport networks: Social capital links. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 257–284.

33.

Jarvie

(2003) Communitarism, sport and social capital. International Review for the Sociology of Sport 38: 139–153.

34.

Jöreskog

(1993) Testing structural equation models. In: Bollen

Long

(eds) Testing Structural Equation Models. Newbury Park, CA: SAGE, pp. 294–316.

35.

Kaplowitz

Hadlock

Levine

(2004) A comparison of web and mail survey response. Public Opinion Quarterly 68(1): 94–101.

36.

Kohut

(1984) How Does Analysis Cure? Chicago: University of Chicago Press.

37.

Lee

Robbins

(1995) Measuring belongingness: The Social Connectedness and the Social Assurance Scales. Journal of Counseling Psychology 42(2): 232–241.

38.

Lee

Robbins

(1998) The relationship between social connectedness and anxiety, self-esteem, and social identity. Journal of Counseling Psychology 45(3): 338–345.

39.

Lee

Robbins

(2000) Understanding social connectedness in college women and men. Journal of Counseling and Development 78(4): 484–491.

40.

Lee

Keough

Sexton

(2002) Social connectedness, social appraisal, and perceived stress in college women and men. Journal of Counseling and Development 80(3): 355–361.

41.

Lin

(2001) Social Capital: A Theory of Social Structure and Action. New York: Cambridge University Press.

42.

Long

(2008) Sport’s ambiguous relationship with social capital: The contribution of national governing bodies of sport. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 207–232.

43.

Long

Sanderson

(2001) The social benefits of sport: Where’s the proof? In: Gratton

Henry

(eds) Sport in the City. London: Routledge, pp. 187–203.

44.

Newman

(2000) A discussion of low r-squares: Concerns and uses. Educational Research Quarterly 24: 3–9.

45.

Nicholson

Hoye

(2008) Sport and social capital: An introduction. In: Nicholson

Hoye

(eds) Sport and Social Capital. Oxford: Butterworth-Heinemann, pp. 1–20.

46.

Numerato

(2008) Czech sport governing bodies and social capital. International Review for the Sociology of Sport 43: 21–34.

47.

Okayasu

Kawahara

Nogawa

(2010) The relationship between community sport clubs and social capital in Japan: A comparative study between the comprehensive community sport clubs and the traditional sports clubs. International Review for the Sociology of Sport 45: 163–186.

48.

Persson

(2008) Social capital and social responsibility in Denmark: More that gaining public trust. International Review for the Sociology of Sport 43: 35–52.

49.

Pigott

(2001) A review of methods for missing data. Educational Research and Evaluation 7: 353–383.

50.

Portes

(1998) Social capital: Its origins and applications in modern sociology. Annual Review of Sociology 24: 1–24.

51.

Putnam

(1993) Making Democracy Work. Princeton, NJ: Princeton University Press.

52.

Putnam

(1995a) Bowling alone: America’s declining social capital. Journal of Democracy 6: 65–78.

53.

Putnam

(1995b) Tuning in, tuning out: The strange disappearance of social capital in America. Political Science & Politics 28: 664–683.

54.

Putnam

(2000) Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster.

55.

Rosso

(2010) From informal recreation to a geography of achievement: Women’s soccer in South Australia. Geographical Research 48(2): 181–196.

56.

Seippel

(2005) Sport, civil society and social integration: The case of Norwegian voluntary sport organizations. Journal of Civil Society 1(3): 247–265.

57.

Seippel

(2006) Sport and social capital. Acta Sociologica 49(2): 169–183.

58.

Sharpe

(2003) ‘It’s not fun anymore:’ A case study of organizing a contemporary grassroots recreation association. Society and Leisure 26(2): 431–452.

59.

Sharpe

(2006) Resources at the grassroots of recreation: Organizational capacity and quality of experience in a community sport organization. Leisure Sciences 28: 385–401.

60.

Sport Canada (2002) The Canadian Sport Policy. Gatineau, QC, Canada: Sport Canada.

61.

Sport England (2004) National Framework for Sport. England: Sport England.

62.

Stokowski

(1990) Extending the social groups model: Social network analysis in recreation research. Leisure Sciences 12(3): 251–263.

63.

Stolle

(1998) Bowling together, bowling alone: The development of generalized trust in voluntary associations. Political Psychology 19(3): 497–525.

64.

Stolle

Rochon

(1998) Are all associations alike? Member diversity, associational type, and the creation of social capital. American Behavioral Scientist 42: 47–65.

65.

Timpone

(1998) Ties that bind: Measurement, demographics, and social connectedness. Political Behavior 20(1): 53–77.

66.

Tonts

(2005) Competitive sport and social capital in rural Australia. Journal of Rural Studies 21(2): 137–149.

67.

Tonts

Atherley

(2010) Rural restructuring and the changing geography of competitive sport. Australian Geographer 36: 125–144.

68.

Tonts

Atherley

(2010) Competitive sport and the construction of place identity in rural Australia. Sport in Society 13(3): 381–398.

69.

Townsend

McWhirter

(2005) Connectedness: A review of the literature with implications for counseling, assessment and research. Journal of Counseling and Development 83: 191–201.

70.

Walseth

(2008) Bridging and bonding social capital in sport – Experiences of young women with an immigrant background. Sport, Education and Society 13(1): 1–17.

71.

Williams

Galliher

(2006) Predicting depression and self-esteem from social connectedness, support and competence. Journal of Social and Clinical Psychology 25(8): 855–875.

72.

Wollebaek

Strømsnes

(2008) Voluntary associations, trust, and civic engagement: A multilevel approach. Nonprofit and Voluntary Sector Quarterly 37: 249–263.

73.

Wood

Danylchuk

(2011) Playing our way: Contributions of social groups to women’s continued participation in Golf. Leisure Sciences 33(5): 366–381.

74.

Woolcock

(1998) Social capital and economic development: Toward a theoretical synthesis and policy framework. Theory and Society 27: 151–208.

75.

Yeh

Inose

(2003) International students’ reported English fluency, social support satisfaction, and social connectedness as predictors of acculturative stress. Counselling Psychology Quarterly 16(1): 15–28.

76.

Yoon

Lee

Goh

(2008) Acculturation, social connectedness, and subjective well-being. Cultural Diversity & Ethnic Minority Psychology 14(3): 246–255.