Field Experiments of Preferential Attachment

Abstract

The preferential attachment hypothesis posits that a node’s propensity to be linked to by other nodes increases with the number of incoming links it has already accumulated. Compellingly demonstrating the presence of such positive feedback in connectivity in longitudinal records is made difficult by confounding temporal processes and unobserved nodal attributes. This challenge can be addressed with a field experimental strategy in which the experimenter bestows links upon random nodes and compares subsequent attachment to these experimental nodes to a control group. In recent studies the authors reported on four such field experiments representing distinct empirical environments and theoretical contexts. In each environment the hypothesis was confirmed, suggesting that preferential attachment is a common feature of network growth in many contexts. This chapter presents results from a novel test in a fifth environment, eBay, where preferential attachment was again found to operate.