ITA Software's Technical Seminar Series

Talk Topic: Experts vs. Aggregators: The Evolution of Endogenous Influence
Tuan Phan, Harvard Business School doctoral candidate
August 4, 2010

Abstract

Marketing researchers and practitioners are interested in finding individuals, or "influentials," in social networks who may have disproportionately higher level of influence over others. Whereas one stream of literature suggests that these influentials are those with personal characteristics such as charisma, "coolness", and expertise, an alternative stream shows that they are those with advantageous network structure by connecting to others that matter. Our study seeks to bridge these two streams by showing that the network structure evolves to create influentials based on the agents' information. Through the endogenous link formation (ELF) process, agents drop links with inferior information quality for better ones. We model two agent types: experts as those who receive information to adopt, and novice agents as those who rely on observational learning to adopt. In an environment where agent types are not communicated, we find that influentials, measured by the number of followers, emerge from those who collect information from several sources and not from experts with exogenous information. These "aggregators" collect and provide higher quality information from multiple sources than that from a single expert. However, in a follow-up study, communication noise moderates the influence of aggregators and increases the strength of experts because of accumulation of error. As a result, novice agents face a trade off between access to multiple experts through aggregators, and the accuracy of that information. For managers, our findings suggests that: 1) aggregators may act as strong influentials in perfect communication medium such as on-line context and digital communication, 2) on the other hand, experts may act as strong influentials in noisy medium such as verbal communication where direct access to information is valued.

Bio

Tuan Phan is a doctoral candidate in the Marketing Unit of Harvard Business School. He is working in the emerging field of computational marketing.