Abstract
Sports clubs may be an ideal setting for social integration for people from different backgrounds. Using heuristic multilevel models, prior studies linked the individual characteristics of the members with social structures at the club context (e.g. club goals) to explain social integration. However, due the organisation of sport activities in teams, another social context with distinct social structures (e.g. team culture) exists within clubs that is likely relevant for social integration as well. Based on data from 1415 members in 140 teams of 42 Swiss football clubs, this study analyses social integration in the dimension of identification in a three-level multilevel model that is the first to include the team context as a level of analysis. The results revealed that teams differ considerably in the social integration of their members. Besides individual factors (e.g. education level, membership duration), a team culture of social togetherness and especially a pronounced team sociability are relevant for identification. Cross-level interactions showed that these factors play a role for members independent of their migration background. Yet, additional positive effects exist for members new to the club. Based on these results, sports club researchers should consider including the team level in multilevel analyses.
Acknowledgement
We would like to thank the club members for their participation in the study Alexander Steiger, Sarah Piller, Sarah Vögtli and Delphine Reymond for their valuable support at various points in the project.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 The originally sample contained 1681 members in 145 teams and 42 clubs. After screening, 156 cases (9.28%) were excluded due to a high number of missing values for the social integration items (>30%) and inconsistent answering behaviour (e.g. long string answers; see Curran, Citation2016) leaving 1525 cases. People born abroad and people with less education were slightly overrepresented in the excluded cases, indicating some difficulties in answering the questionnaire. To reduce the loss of cases due to listwise deletion of missing data points, missing data points (2.1% of data points in items on identification were missing) were imputed using the expectation maximisation algorithm (Snijders & Bosker, 2012; Tabachnick & Fidell, Citation2019; see also Adler Zwahlen et al., Citation2018). Auxiliary variables associated with the missing values or highly correlated with the items were included for estimation purposes (e.g. age, membership duration, education level, volunteering; PMB were not associated with missing values). Considering the nested data, imputation was done by cluster Graham (Citation2009, p. 564). Remaining incomplete variables at the individual and team level led to the exclusion of 5 teams and 110 cases, leaving 1415 cases in 140 teams and 42 clubs.
2 Bivariate correlation results of conversations beyond football and identification are slightly not relevant (p = .101). However, controlling for youth teams, membership duration, or volunteering in partial correlations, conversations beyond football are relevant.