Preference Evolution in a Social Network

• Activity Number: 590
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Jing Wang*+ and Anocha Aribarg and Yves F. Atchade
• Companies: University of Michigan and University of Michigan and University of Michigan
• Keywords: social network ; viral marketing ; Bayesian analysis ; random field ; Markov Chain

Abstract

With the advent of social media, e.g., Facebook, understanding how consumers' preferences for products can be influenced and dynamically changed over time within a social network is crucial, as it helps firms design more effective advertisements on social media. We propose a discrete Markov chain model and a random field model to fit longitudinal choice data observed in a social network. Two methods to simplify the estimation of the random field model are introduced. We show that under certain conditions, the Markov chain model converges to the random field model. We designed a field experiment that mimics how consumers respond to interactive advertising on Facebook. We calibrate our models to the data using a Bayesian framework. Controlling for individuals' heterogeneous preferences, the results show that consumers' preferences are significantly influenced by their peers'.