Modeling product attitude formation process in online word‐of‐mouth

Purpose - Word‐of‐mouth (WOM) has been found to significantly influence consumers' decision making. Much attention has been paid to the effect of WOM characteristics such as the number of postings, the dispersion of online conversations, the reputation of the reviewers, and the review quality on product sales. Little research, however, has examined the interaction process of online reviews. The purpose of this paper is to investigate the consumers' product attitude formation process in online WOM. Three research questions will be addressed in this paper, i.e. the effect of prior responses on the following repliers' product attitude, the negativity effect and the role of main messages in shaping consumers' product attitude formation process. Design/methodology/approach - The product attitude formation process of online WOM is investigated using the data of product reviews (main messages) and their corresponding responses. The paper collected 26 product reviews from various web sites and kept the first 40‐50 responses for each review, which resulted in 26 main messages and 1,173 observations (i.e. responses) in total. A hierarchical Bayesian ordinal choice model is then specified and estimated with the Markov Chain Monte Carlo method to address the research questions and to capture the main message heterogeneity. Findings - The paper finds that the impact of prior responses (e.g. the proportion of positive and negative responses) on the product attitudes of the following responses differs significantly across products. This heterogeneity can be well explained by the characteristics of the main messages at the second‐level specification. Thereby, factors that influence consumers' product attitudes in the interaction process of online WOM include prior responses and main message characteristics. Another interesting finding is that positive responses have larger impacts on product attitudes than negative ones. Originality/value - This research contributes to both academic research and the firms' online WOM management. Theoretically, this research is the first attempt to examine the formation process of attitudes toward new products in online communications. This research contributes by modeling how the dynamic process of online WOM influences new product attitudes. Furthermore, inconsistent with the “negativity effect” proposed by researchers (e.g. Skowronski and Carlston), the paper finds that positive responses matter more than negative ones in online communications. In addition, the way the paper configures the data for online communications is innovative and provides a perspective to quantitatively model the communication process. Managerially, this research provides implications for firms to intervene in the online communication process and influence consumer attitude of new products.