The observation that individuals tend to be friends with people who are similar to themselves, commonly known as homophily, is a prominent and well-studied feature of social networks. While homophily describes a bias in attribute preferences for similar others, it gives limited attention to variability. In this work, we observe that attribute preferences can exhibit variation beyond what can be explained by models of homophily. We call this excess variation monophily to describe the presence of individuals with extreme preferences for a particular attribute possibly unrelated to their own attribute. We observe that monophily can induce a similarity among friends-of-friends on a network without requiring any similarity among friends. In order to independently simulate homophily and monophily in synthetic networks, we contribute a new model of social network structure that we call the overdispersed stochastic block model (oSBM), an extension of the classical stochastic block model. We use this model to demonstrate how homophily-based methods for predicting attributes on social networks based on friends, “the company you keep,” are fundamentally different from monophily-based methods based on friends-of-friends, “the company you’re kept in.” To illustrate the differences between homophily and monophily-based prediction we place particular focus on predicting gender, where homophily can be weak or nonexistent in practice. These findings offer an alternative perspective on network structure and attributes in general and prediction in particular, complicating the already difficult task of protecting privacy on social networks.
This is joint work with Johan Ugander, Assistant Professor in the Management Science & Engineering Department at Stanford University (Website: https://web.stanford.edu/~jugander/). The paper associated with this talk is at https://www.nature.com/articles/s41562-018-0321-8.