I propose a social structure theory that challenges the widely accepted role of preferential attachment and triadic closure as primary mechanisms of network formation. For that, I build upon Feld's concepts on social circles, Breiger's on the duality of actors and groups, and Hinde's on interactions and relationships. The theory emphasizes that ties between actors stem from social circles in which they participate; a notion straightforwardly modeled through two-mode and projected networks. Using recent results, aided by the analyses of empirical and artificial networks, I argue that observed macro-level structural properties of social networks emerge from homophily jointly with social activity – a mechanism introduced in this study. More specifically, homophily, overlapping (a by-product of homophily), and social activity interweave to generate strong ties, popularity and strength heterogeneity among actors, and high clustering that leads to community formation. I also show how overlapping increases segregation, from a networks viewpoint. This work's implications are broad and applicable to several social processes ranging from investigating social cohesion, tolerance, and child development to modeling epidemics.
