Duality Revisited: A New Methodology for Bipartite Networks

Abstract

We propose a novel method to explicate the emergent structure of a field of social groups starting with a random sample of individuals. We illustrate our approach by leveraging a two mode ego-occupation network data collected on a random sample of 6,000 Japanese. Results show that the occupational field of Japan has a segregated and polarized emergent structure. Our method is based on the assumption that homophily drives individual sociability. We use an agent based modeling to test this assumption and identify two key mechanisms - status and geographical homophily - that produced networks consistent with the observed emergent field structure.