Node centrality measures serve as an additional way, beyond the visual of a graph, to understand social networks. These measures provide more detailed evidence about how the network itself, and the players within it (the nodes) function. For example, the eigenvector centrality metric gives us an idea of which nodes are most influential in the network, serving as an indicator for spotting important nodes. Additionally, the betweenness centrality metric is basically the idea of how regularly a given node serves as a bridge or intermediary between other nodes in the network. In this way, centrality measures provide information not just on the number of edges connected to each node (as can be seen in the visual graph like the one below), but also explains the role that each of the nodes play in the network, and how their role impact larger network functioning.
Ennett and colleagues (2006) provide an example of how centrality measures such as betweenness centrality and a few other centrality measures that we haven’t specifically discussed in class. In their study of substance use in adolescent peer groups, they describe betweenness centrality as “the extent to which an adolescent in-directly links pairs of adolescents who are not directly linked as friends” (p. 168). They found that 45% of the adolescents in their sample served as bridges between neighborhood peers, and that betweenness centrality was significantly and positively correlated with the number of neighborhood substance users. However, other measures of centrality were more likely to be significantly associated with personal substance use, suggesting that the adolescents who are most visible in the network are more likely to use substances than the ones who serve as intermediaries. In this example, we see how the use of betweenness and other centrality measures allow for the understanding of how nodes in the network function, and the relationship between network positionality and an outcome of interest.
Similarly, Russell and Smith (2011) show the importance of using centrality measures such as degree centrality in their investigation of the afterschool network program in Dallas, Texas. For example, through exploring the degree centrality of certain programs, there were able to identify which programs were vulnerable due to the fact that they had just one or two sources of funding, and tended to work in isolation. The image below shows the differences across the network in relation to their funding structures. Programs in the clusters on the left generally receive funding from one or two organizations; whereas programs in the cluster on the right have more diverse funding mechanisms and are more integrated into the community of afterschool program providers.
Results from these two studies demonstrate the different ways in which measures of centrality can be used to understand not only network shape but also how nodes within a network differ in relation to each other, how edges in a network develop into clusters, and how these clusters and measures better allow us to understand our research questions.
- Ennett, S. T., Bauman, K. E., Hussong, A., Faris, R., Foshee, V. A., Cai, L., & DuRant, R. H. (2006). The peer context of adolescent substance use: Findings from social network analysis. Journal of research on adolescence, 16(2), 159-186.
- Russell, M. G., & Smith, M. A. (2011). Networks Analysis of a Regional Ecosystem of Afterschool Programs. Afterschool Matters, 13, 1-11.