Canonical correlation analysis: Assessing links between multiplex networks

Abstract

Ignoring the multiplex nature of social networks might result in sets of empirical findings that are misleading or contradictory. It would be helpful to adapt a variety of multivariate statistical techniques to handle network-oriented data. A canonical correlation analysis (CCA) makes a particularly interesting example since it is the most general form of the general linear model. In an example involving 317 banks, a CCA is applied to two multiplex networks: interdependence and cooperative alliances. Four significant patterns of association (orthogonal linear functions) are identified.