Our life experiences are shaped in contemporary societies by various types of social relationships, from followers in social media to ties with friends and kin. These relationships provide support, resources, and information, acting as a window through which we view the world and exposing us to certain realities over others. The myriad of ties surrounding each individual interact to generate broader population networks, that have the potential of supporting and integrating society members but also of dividing them. In this dissertation, I shed light on this potential by studying homophily among different types of social ties. In particular, I study follower relationship on Twitter and offline ties of various strength, from core ties to acquaintances. I focus on variables generating important cleavages in society, such as socioeconomic status, race and ethnicity, age, or partisanship, and on the role of residential patterns.

I leverage three unique datasets that allow examining patterns in social networks previously unexplored in the literature. In the first project, I use data on Twitter users matched to US voter file records to evaluate how physical proximity and homophily by age, sex, race/ethnicity, and partisanship structure follower ties. In the second project, I work with data from a large-scale survey with question on the core ties of respondents. I take advantage of the large sample sizes and the geographical granularity of this survey to investigate the correlates of core network homogeneity by race and ethnicity in the US and the role of opportunity structures. Finally, my third project uses a probability survey from Spain with detailed questions on the extended ego-network of respondents to describe how the homogeneity of ties varies with relationship strength.