The Resonance of Reciprocity: How Reciprocity Informs Social Network Structure

• In Event: Analytic and Computational Approaches in Social Psychology (Co-sponsored by Mathematical Sociology Section)

Abstract

Sociologists and social psychologists have recognized the power of reciprocity in social processes for decades, with Alvin Goulder in 1960, for example, proposing a “norm of reciprocity.” According to this norm, people are expected to respond in kind to actions directed towards them by another, typically referring to positive, prosocial actions. Social exchange theory maintains that reciprocity endures due to its social rewards and the costs accrued with failure to reciprocate beneficial interactions. In recognition of the foundational role of reciprocity, statistical analyses of social networks routinely control for this process. Even though the effect of reciprocity is almost always sizeable, the overabundance of mutual dyads is continuously taken-for-granted in social networks research. The purpose of this paper is to examine systematically the degree to which the presence of reciprocity shapes social network configurations. We accomplish this by estimating exponential random graph models (ERGMs) on a sample of 120 social networks. The networks represent diverse genres of interaction and include relationships defined by both amity and conflict. Then, we rely on our ERGM estimates to simulate a series of networks and quantify the extent to which local tendencies toward reciprocity inform the broader macro-level structures of networks (e.g., centralization, clustering, connectedness). We find that reciprocity tends to reduce network centralization, shorten indirect paths, and increase clustering. These effects are significantly more prevalent in positive and in-person networks, as compared to negative and online graphs. Our findings imply that the benefits associated with positive social ties extend to a wider range of people when reciprocity is present. In other words, the effects of reciprocity can extend beyond the immediate dyad to influence broader, network-level topologies that often improve population-level outcomes.

Authors

• Diane H. Felmlee, Pennsylvania State University

• Cassie McMillan, Northeastern University