Can the closeness of the mutual friends of both participants in a romantic relationship help determine the strength of that relationship? A study by Cornell University Computer Scientist Jon Kleinberg and Facebook Engineer Lars Backstrom sought to find out.
Kleinberg and Backstrom studied data from 1.3 million randomly selected, anonymous Facebook users, all of whom were at least 20 years old, with anywhere from 50 to 2,000 friends, and all of whom indicated in their profiles that they were in relationships, analyzing a total of some 379 million nodes and 8.6 billion links.
They found that by using dispersion to study the partners’ relationships, rather than embeddedness, they were able to more accurately identify the partners. Kleinberg and Backstrom described the difference between dispersion and embeddedness as follows:
Our central finding is that embeddedness is in fact a comparatively weak means of characterizing romantic relationships, and that an alternate network measure that we term dispersion is significantly more effective. Our measure of dispersion looks not just at the number of mutual friends of two people, but also at the network structure on these mutual friends; roughly, a link between two people has high dispersion when their mutual friends are not well connected to one another.
On a large random sample of Facebook users who have declared a relationship partner in their profile, we find that our dispersion measure has roughly twice the accuracy of embeddedness in identifying this partner from among the user’s full set of friends. Indeed, for married Facebook users, our measure of dispersion applied to the pure, unannotated network structure is more effective at identifying a user’s spouse than a complex classifier trained using machine learning on an array of interaction measures including messaging, commenting, profile-viewing, and co-presence at events and in photos. Further, using dispersion in conjunction with these interaction features produces significantly higher accuracy.
Kleinberg told The New York Times’ Bits blog in an interview:
A spouse or romantic partner is a bridge between a person’s different social worlds. We hadn’t had this view of it before.
And Kleinberg and Backstrom concluded:
Beyond these specific applications, our measures suggest new perspectives on basic questions in social network analysis. Overall, the notion that our mutual friends with a person may be clustered in a single context or may alternately span multiple contexts offers a novel type of trade-off in the study of tie strength. Certain important types of strong ties — including romantic and family relations — connect us to people who belong to multiple parts of our social neighborhood, producing a set of shared friends that is not simply large, but also diverse, spanning disparate portions of the network and, hence, correspondingly sparse in their internal connections. In this way, the notion of dispersion combines concepts of network closure (since there must be mutual network neighbors to bridge) and brokerage between groups (since the two ends of a link with large dispersion are jointly acting as brokers between disconnected mutual neighbors). The success of the measures resulting from this notion suggests that some of the ways in which closure and brokerage are intertwined in the structure of strong ties.
The analysis shows how these classes of strong ties produce an extremely clear structural signature, but subtle network measures different from the standard formulations are needed to extract this signature. Crucial aspects of our everyday lives may be encoded in the network structure among our friends, provided that we look at this structure under the right lens.
Readers: How much of an impact do you believe partners’ sets of friends has on the success of relationships?
Image courtesy of Shutterstock.