2 Introduction: Social Network Analysis - A primer for sport scientists.
2.1 Relational data in sports
Wäsche et al. (2017) offer a typology of possible networks in sports. Their article is also an amazing resource to start your journey into SNA (in sports and beyond).
| Network.type | Description |
|---|---|
| Competition Networks | Results of games, rankings or structural patterns in sport (e.g. two-mode networks such as athletes at same competitions, players in same team). Relative performance or structural patterns are analyzed. |
| Interaction Networks (intra-event) | Relations are rule-based elements of the game (e.g. passes), often analyzed for effectiveness. |
| Inter-organizational Networks | Network structures between organizations (e.g. collaboration between sport providers, franchises, clubs, event organizers). |
| Intra-organizational Networks | Network structures within organizations (e.g. communication among team members or within sport associations). |
| Affiliation Networks | Affiliations of social actors with aggregations of social actors (e.g. membership of individuals in organizations, participation of actors at events). Data are usually collected as two-mode networks. |
| Social Environments | Representation of a sport-related social environment as a network in which individual sports actors (e.g. athletes, individuals, teams) are embedded. Analysis of support, influence etc. |
For a general introduction to SNA there are a number of good books, eg. Scott (2013), Hanneman and Riddle (2005) or Wasserman and Faust (1994) (in their respective current editions).
2.2 Network phenomena: Global and individual levels
We can broadly distinguish between network measures on the individual and global level. Individual level metrics tell us something about a member of the network. How many integrated is he/ she? How central? What role does he/ she play? Global metrics tell us something about the overall structure of the network. How centralized is it? How well connected? How many parts are there?
2.3 Networks as independent variables
What is the influence of structure?
2.3.1 Example: English soccer teams
Grund (2012) examine the relationship between the passing structure of English soccer teams (network characteristics on the global level as an independent variable) on team success.
2.4 Networks as dependent variables
How does structure come about?
2.4.1 Example:
Gyarmati, Kwak, and Rodriguez (2014) study different styles of soccer teams, again using passing data. This is maybe not an exact case of a network being the dependent variable, as they mostly seem to be interested in description. However, you could think about how these networks come about. Does it depend on the ligue? The age of the coach? The nationality of the players?…
2.5 Setting boundaries
Determining the boundaries of a social network (who belongs to it and who not/ who is included in the analysis) is a task fraught with ambiguity. There usually are no neat borders in the social world, especially if relations between social actors are considered. For example, studying friendship ties within a single sports team sounds straightforward at first. However, friendship is an obvious example where relations are usually not confined to a single group. Thus, member of a sports team will be embedded in a larger friendship network. Depending on the character of this embeddedness, this can have a great impact on the phenomenon under study. For example, two groups of team members might originate from friendship relations to polarizing characters outside the team, or former team member might be important actors who still have an influence on present team dynamics.
While recognizing this ambiguity, we should not despair. To ensure reproducability and transparency, it is most important to clearly state the strategy taken in determining network boundaries. The literature has developed a terminology for this.
On a very basic level, nominalist and realist strategies can be distinguished (Wäsche et al. 2017). In nominalist strategies, network boundaries are defined by the researcher. In realist strategies, this is done by the study participants.
When choosing a nominalist approach, the setting of boundaries by the researcher should be justified on theoretical ground. An argument should be supplied for why a chosen boundary setting is the most appropriate for answering the study questionn. Scott (2013) (p. 44) distinguishes different types of nominalist strategies. Using a positional approach, study subjects are chosen from formally defined categories (such as members of a sports team). If this is not possible (if, for example, such categories do not exist, or it is hard to identify them), a reputational approach is another possibility. In a reputational approach, study subjects are identified through nomination by others. Thus for example, to study an informal trend sport community that is not (yet) organized in leagues or teams, member of the community could be asked to name people they consider part of the community or the most important members of the community.
A somewhat different approach to boundary setting is snowball sampling, which can also be applied when a positional approach is not possible. In carrying out snowball sampling, member of a starting population are asked to nominate others they are related to regarding the type of relation under study. These are then included in the study population. This process can be repeated until a desired level of saturation is reached.
2.6 Conclusion: A checklist to inform your research design
- Am I studying a relational phenomenon? What are my ties/ links?
- What is the content of my ties?
- Do I look at network phenomena at the global or individual level?
- Are network phenomena my dependent or independent variable?
- How would I go about setting boundaries for my network?
References
Wäsche, Hagen, Geoff Dickson, Alexander Woll, and Ulrik Brandes. 2017. “Social network analysis in sport research: an emerging paradigm.” European Journal for Sport and Society 14 (2). Routledge: 138–65. doi:10.1080/16138171.2017.1318198.
Scott, John. 2013. Social Network Analysis. London ; Thousand Oaks, Calif.; New Delhi: Sage Publications.
Hanneman, Robert A., and Mark Riddle. 2005. Introduction to Social Network Methods. Riverside, CA: University of California, Riverside (published in digital form at http://faculty.ucr.edu/~hanneman/ ). doi:10.1016/j.cell.2011.03.009.
Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. Edited by Mark S. Granovetter. Cambridge, UK: Cambridge University Press. doi:10.1525/ae.1997.24.1.219.
Grund, Thomas U. 2012. “Network structure and team performance: The case of English Premier League soccer teams.” Social Networks 34 (4): 682–90. doi:https://doi.org/10.1016/j.socnet.2012.08.004.
Gyarmati, László, Haewoon Kwak, and Pablo Rodriguez. 2014. “Searching for a Unique Style in Soccer.” Proceedings of ACM KDD Workshop on Large-Scale Sports Analytics. http://arxiv.org/abs/1409.0308.