This is an excerpt from Research Methods and Design in Sport Management 2nd Edition With Web Resource.
By Adam Love and Amy Chan Hyung Kim
In this section, we introduce some key theoretical concepts used in social network data analysis and identify several software packages that may be used in conducting an SNA.
Connection and Distance
When analyzing an actor's attributes and behavior, it is likely that the more connections one has within a network, the more one is exposed to more diverse information (Hanneman & Riddle, 2005; Scott, 2017). In addition, items ranging from information to disease spread more quickly if there are more connections among actors in a network. Further, people who are more highly connected may be able to better obtain and use their resources and access more diverse perspectives to resolve issues. Thus, SNA can be useful in examining the consequences of variations in the degree of connection among actors (Hanneman & Riddle, 2005). In doing so, two important concepts are relevant: connection and distance.
To examine the extent of connection in a network, a researcher must know the number of actors, the number of possible connections, and the number of connections that are actually present between actors in a network. Two important factors include differences in the size of networks and the extent to which actors are connected, which is known as density. The density of a network is simply the proportion of all possible ties that are actually present. For a valued network, density is calculated as the sum of the ties divided by the number of possible ties. The density of a given network provides insight about the ways in which information may diffuse between actors and how a certain actor may possess a high level of social capital and/or social isolation (Hanneman & Riddle, 2005).
The concept of distance refers to the characteristics of the network that are relevant to adjacencies: the direct tie from one actor to another. The larger the distance between actors, the longer it takes to diffuse information within a network. Distance is measured in terms of the number of “steps” from one actor to another; in other words, if two actors are adjacent to one another, the distance between them is 1, meaning it takes one step to go from the source to the receiver (Hanneman & Riddle, 2005).
Centrality and Power
The concept of centrality provides insight about power dynamics. From a social network perspective, one cannot have power without connections, and power may result from a person's ability to influence other actors in a network (Hanneman & Riddle, 2005). To study the ways in which a person's position in a network may influence power, we present three types of centrality: degree centrality, closeness centrality, and betweenness centrality (Borgatti et al., 2013).
Degree centrality focuses on how many connections a given actor has in a network. If a network contains directed relationships between actors, it is also important to distinguish between in-degree centrality and out-degree centrality. If one actor has four incoming connections, the in-degree centrality is four. If one actor has five outgoing connections, the out-degree centrality of that actor is five (Freeman, 1979). Degree centrality tends to measure only the immediate ties that an actor has, overlooking the importance of indirect ties to others in the network. In other words, one actor may be directly tied to a large number of other actors, but those actors may be disconnected from the whole network. In this case, this actor may be positioned as a central actor, but only in a local neighborhood (Hanneman & Riddle, 2005). To understand the indirect ties of an actor, closeness centrality highlights the distance of an actor to all others in the network. Additionally, the person who lies between two actors can have power because information can pass between those two actors only through this broker. Betweenness centrality measures the extent to which an actor serves as an intermediary between other actors in a network.
Social Positions and Structural Equivalence
To investigate network positions and social roles, social network scholars have developed the idea of structural equivalence (Scott, 2017). In its simplest form, if two actors have a similar pattern of relationships with other actors, these two actors can be said to have the same or a similar position in the network. Scott explains this type of structural equivalence using the concept of “substitutable” or “interchangeable” actors. That is, if one's position can be occupied by an actor who has similar relational ties, these two actors' positions are said to be structurally equivalent. In other words, identifying uniformities within social positions is key to understanding structural equivalence. Once positions within a certain network are identified, the relations among these positions can be explored. Hanneman and Riddle (2005) describe two types of equivalence in addition to structural equivalence: automorphic equivalence and regular equivalence. If two actors have the exact same relationships to all other actors within a network, these two actors are considered to be structurally equivalent. Automorphic equivalence entails a less strict definition of equivalence. That is, sets of actors can be automorphically equivalent by being positioned in local structures that have the same patterns of ties. For example, assume that both the manager of a sales department and the manager of the human resource department have three employees each. Even though these two managers do not share the same three employees, each has a similar pattern of ties with three employees. Finally, two actors are considered to be regularly equivalent if they have the same pattern of ties with members of other sets of actors who are also regularly equivalent. In other words, actors who are regularly equivalent do not have to fall in the same network positions, but they do have the same types of relationships with members of another set of actors (Hanneman & Riddle).
Social Network Analysis Software
Several software packages for conducting SNA have been developed in recent years. The following list includes software packages that are developed specially for SNA and have their own graphical user interfaces (GUIs): Cytoscape, EgoWeb 2.0, Gephi, KrackPlot, NetMiner, NodeXL, Pajek, SocNetV, StOCNET, and UCINET. Readers should be aware that there are also social network analysis packages designed for programming languages such as R (i.e., a statistical computing software) or Python.
Here is a list of SNA software packages: