Replication data for: Online Word of Mouth, Hit vs. Niche Products, and Social Dynamics

Post-purchase online word of mouth (WOM) has emerged as a major research area in marketing and information systems literature as it functions both as a predictor and driver of sales revenue. While the relationship between WOM and sales is now quite well understood, it is less clear what drives WOM itself. Of specific interest is the question of whether a population’s propensity to contribute to post-consumption WOM is dependent on various indicators of that product’s availability. The study of these product-level effects is complicated by simultaneous social dynamic-level effects which might also differ for products of varying availability and popularity. Furthermore, the bi-modal distribution commonly found in review ratings complicates the use of established summary statistics such as arithmetic mean and standard deviation. In this paper we study a dataset of almost 280,000 consumer reviews for 433 movies over three years to disentangle these product-level effects and social dynamic-level effects to better understand their effect on a population’s propensity to contribute to post-consumption WOM. We first show how the use of a simple arithmetic mean can lead to estimation problems. To resolve these problems and allow the study of social dynamics, we introduce a novel measure to capture the disagreement contained in non-normally distributed review ratings based on an expectation-maximization algorithm for finite mixture models. Using this measure, we find that consumers are more likely to post reviews for products that are perceived to be less available in the market but no effect for those perceived to be less successful in the market. Investigating the social dynamic-level effects, we find a positive effect for products that have already accumulated a larger volume of prior reviews and a positive effect of disagreement on consumers’ propensity to engage in post-consumption WOM. Investigating the interaction between the movie-level effects and social dynamic-level effects, we find that while high disagreement helps gen erate WOM for less popular (niche) products, it hampers that of highly popular (hit) products.