Social Influence on Future Review Sentiments: An Appraisal-Theoretic View

Abstract

Given the importance of online reviews, as evidenced by extant research, we studied an understudied area of the impact of past reviews on sentiments of future reviews. Past studies in the emergent area of incidence of reviews have investigated the generation of ratings but there seems to be a lacuna in the second dimension of the reviews (i.e., the textual sentiments). Through a controlled experiment (N = 232), we analyzed the impact of past reviews on reviewers’ subsequent texts by using the unique lens of social appraisal theory. We found the influence of the selection of reviews on most e-commerce websites could strongly bias subsequent written review sentiments, and this effect is more pronounced when the reviewer experienced higher disconfirmation. We also observed review writers tend to post extreme reviews in the absence of any benchmark or prior reviews. The study results extend the understanding of users’ reviews under social influence and enhance the appraisal theoretic understanding of review generation.

Key words and phrases:

• social appraisal theory
• sentiment analysis
• online reviews
• controlled experiment
• e-WoM
• user reviews

Notes

1. Phone with 5.5 inch Corning gorilla glass AMOLED touchscreen, with USB type C connector, 3GB RAM and 64GB ROM.

2. The Likert scale had a range from 1 to 7, where 1 denotes an extremely negative and 7 denotes an extremely positive review. A similar Likert scale that had a range from 1 to 7 was used for credibility rating, where 1 denoted minimum credibility and 7 denoted maximum credibility. The 7- point scale is structurally similar to a 5-point scale, but it allowed more variability and nuance-capturing capability for the sentiments.

3. To help the human coders understand and code credibility, the definition of credibility and the components that build up credibility, as described in Cheung et al. [10], were explained to the subjects.

4. We did not attempt to resolve the differences between the coders because we were deploying statistical means along with standard deviation to identify the most positive and negative reviews.

5. PPP: purchasing power parity, retrieved from the World Bank website. It has been provided to enable readers across different geographies to evaluate and compare the compensation provided to participants.

6. The research assistants were trained to code sentiments, and Cohen’s Kappa was used to ensure that the scoring scales between different coders did not diverge.

7. To ensure parity between our experimental results and Amazon review analysis, we also ran the experimental reviews through SentiStrength and used the software-provided sentiments for analysis. There was no significant change in results. The detailed results for this analysis can be obtained from the authors on request.

Additional information

Notes on contributors

Ashish Kumar Jha

Ashish Kumar Jha ([email protected]) is an associate professor of Information Systems at Rennes School of Business, France. His research interests lie in business analytics, Big Data, and technology innovation in IT firms. He has published in such journals as Information & Management, International Journal of Production Economics, and Computers in Human Behavior, among others. Dr. Jha has also presented papers at the top conferences, such as International Conference on Information Systems, Decision Sciences Institute annual meeting, and others. He is an associate editor for Journal of Organizational Computing and Electronic Commerce.

Snehal Shah

Snehal Shah ([email protected]) is a professor of Organization Behavior, chairperson of the Fellow Program, and head of research at SP Jain Institute of Management & Research, India. Her primary research interest is diversity and inclusion, with a specific focus on corporate governance and career choices. Her other research interests include Eastern wisdom traditions and intersectionality of Organizational Behavior concepts in other management domains. Dr. Shah has published in the Journal of Applied Psychology, Journal of Organization Behavior, and Journal of Management Development. She has also presented papers at top conferences, such as the Academy of Management and Marketing Science annual meetings, and has written book chapters.