Birds of a Feather: Homophily in Social Networks

The question

How similar are the people we are connected to, why, and what does that do to what we know?

The method

A review of eight decades of network research rather than a new study — the paper that consolidated a scattered literature into one principle and became the standard citation for it [1].

The findings

The principle is stated flatly: “homophily is the principle that a contact between similar people occurs at a higher rate than among dissimilar people” [1]. It is not a friendship phenomenon. It “structures network ties of every type, including marriage, friendship, work, advice, support, information transfer, exchange, comembership, and other types of relationship” [2] — advice and information transfer are on that list by name.

The consequence for anyone trying to learn something from people is the sentence to remember: “homophily limits people’s social worlds in a way that has powerful implications for the information they receive, the attitudes they form, and the interactions they experience” [2]. Because “the pervasive fact of homophily means that cultural, behavioral, genetic, or material information that flows through networks will tend to be localized” [1], what reaches you is a function of where you sit.

On which dimensions: “homophily in race and ethnicity creates the strongest divides in our personal environments, with age, religion, education, occupation, and gender following in roughly that order” [2]. The causes are structural, not merely preferential. “Geographic propinquity, families, organizations, and isomorphic positions in social systems all create contexts in which homophilous relations form” [2], and institutions do the heavy lifting: “school, work, and voluntary organizational foci provide the great majority of ties that are not kin” [3]. The pattern is then maintained as well as produced, because “ties between nonsimilar individuals also dissolve at a higher rate” [2].

The limits

This is a review, and most of the literature it reviews is cross-sectional American survey data on friendship and confiding ties. The authors’ own closing argument is that the field needs dynamic data and better treatment of ties that serve several purposes at once. It describes a strong regularity; it does not give you a correction factor.

• You are drawing your Calibration Call list from the people you already know. Without this paper, five names from five different people feels like five perspectives. Homophily says those five were drawn from a social space that is already sorted by school, employer and position — which the paper says is where the great majority of non-kin ties come from.
• You are asking for referrals and getting them fast. Easy referrals are a symptom: the people easiest to reach are the ones closest in social space, and their view of the role is correlated with the view you already have.
• You want your sample to actually widen. The paper tells you which levers matter, because it tells you where ties come from. A person who entered the role by a different route, at a different kind of company, in a different city, is a genuinely different draw. A second name from the same team is not.