Abstract:
|
Cluster analysis is a statistical method in which observations are grouped into different classes based on some measure of similarity and is a natural exploratory statistical method when working with multiple, isolated networks such as those seen in the social sciences. There are several ways to cluster social networks depending on how the network is summarized; in particular, we explore using feature vectors in which the paths among different types of nodes are enumerated as network summary statistics. This method is particularly useful in grouping networks based on the ways in which leaders and non-leaders interact, and we will use teacher advice-seeking networks to illustrate these methods.
|