A HIDDEN MARKOV MODEL OF ONLINE RATING DYNAMICS

Abstract

Consumers communicate opinions about their product usage experiences in online platforms. Whereas researchers have demonstrated the impact of such online opinions on product sales and started to investigate the underlying drivers that affect an individual’s decision to contribute these opinions, little attention has been paid to the behavioral changes of the individuals over time. A reviewer can differ from herself when accumulating rating histories on the same online platform in the posting behaviors. We develop and estimate a nonhomogeneous hidden Markov model to account for the transitions among latent states and consequential behavioral changes of an individual reviewer. Applying the model to online movie ratings, we uncover 3 states a reviewer may probabilistically reside in her tenure and estimate the effect of reviewers’ rating histories on the transitions between these states. Moreover, we show that reviewers in each latent state are affected by the rating environment in substantively different ways.