CASOS Working PAPER

"Multivariate Methods for Inter-Structural Analysis" (PDF file)
Authors: Carter T. Butts & Kathleen M. Carley

Abstract
Inter-structural analysis - the analysis of data sets consisting of multiple social structures - is an important extension of classical social network analysis. Building on past research regarding metric inference and graph covariance analysis, we here present a general approach to the analysis of social structure sets with arbitrary labeling assumptions. We define a family of structural distance and analogous structural covariance measures, and show that these measures satisfy the conditions necessary for their use in conjunction with traditional multivariate analysis procedures. Demonstrations of the use of various exploratory data analytic procedures on inter-structural data sets are provided, and suggestions are made regarding the further development of this approach.