Estimating Online Reviews Adoption: A Bayesian Network Approach

Abstract

We use Bayesian Networks (BN) to estimate how the characteristics of online reviews and product involvement affect perceived message credibility and message adoption. An experimental design and sample data from 236 individuals with knowledge and interest in online product reviews are used in this study. Participants were asked to read and evaluate eight product reviews representing different combinations of online message strength, message framing, and source credentials and report the perceived credibility and likelihood of message adoption. We define the model structure using the existent theoretical knowledge and refine it using the Bayesian Network learning structure. We capture the most likely structure between the informational factors of online reviews, involvement and message adoption. By using BNs, we show how to predict and make diagnostic inferences of adoption of online messages given different scenarios of message characteristics and source credibility.