Social Network Analysis

How and what we consume, like our behavior more generally, is shaped by our interactions and relations with other people and by networks generated through interactions and relations between the multiple actors in the various populations to which we belong. There is a growing recognition in social science that networks are central to a proper understanding of taste and consumption. This poses a methodological challenge, however, as most standard methods of data gathering and analysis are not well suited to the analysis of relational data, that is, data on relations between actors.

Social network analysis (SNA) is a set of techniques for handling, analyzing, and visualizing such data. It is a structural method that is used to examine the pattern of connection among a set of actors. As a formal method, it can be applied to any type of relationship and any set of actors. Researchers might study relations within sets of individual human beings, organizations, cities, websites, or any other “node” capable of engaging in the types of relationships that are of interest to us. Likewise, the relations in question might be relations of friendship, sexual contact, economic exchange, bullying, aggression, migration, URL connection, or acquaintance—again, any type of relationship that is possible between the specified entities and significant in terms of the research questions and interests.

SNA emerged in an overlap among a number of social science disciplines and a branch of mathematics known as graph theory. Over time, it has extended into other social science disciplines and has incorporated other mathematical aspects, most notably statistics. In addition, the growing popularity of the concept of networks and of graph theoretical methods of analyzing them across the academic field has led to various interdisciplinary debates (and turf wars), most notably with physicists.

There are three primary types of network data used in SNA:

• Ego-net data. An ego-net is the network that forms around a particular actor. Typically, analysis focuses on, or compares, the networks of various egos. This requires that for each ego, researchers derive a list of relevant alters and a list of all relevant relations between those alters.
• Complete network data. In this form of network analysis, researchers define a population, specify the type of relationship they are interested in, and investigate which members of the population have that type of relationship.
• Two-mode analysis. Typically the “modes” in question are actors and some context where actors might meet. The underlying idea is that actors are connected through, for example, their common participation in the same events, and events are connected through the participation of the same actor(s) in them. Technically two-mode networks can be decomposed into two complete networks: for example, research on the links between people and places might produce a network of people (linked through places) and a network of places (linked by people). Reference to two-mode networks, however, generally suggests an attempt to deal with both modes in the same network.

SNA procedures differ in accordance with the type of network being analyzed. In what follows, the paradigm example is complete networks, unless otherwise indicated.

...