Your Hometown Matters: Popularity-Difference Bias in Online Reputation Platforms

• 成果类型: Article
• 署名作者: Kokkodis, Marios; Lappas, Theodoros
• 署名单位: Boston College; Stevens Institute of Technology
• 刊物名称: INFORMATION SYSTEMS RESEARCH
• ISSN/ISSBN: 1047-7047
• DOI: 10.1287/isre.2019.0895
• 发表日期: 2020
• 页码: 412-430
• 关键词: word-of-mouth; destination image; consumer reviews; product reviews; social network; self-selection; IMPACT; ratings; performance; TOURISM

摘要

We study a new source of bias in online review platforms that originates from the popularity difference between the traveling reviewer's hometown and destination (popularity-difference bias). In particular, we model popularity-difference bias as a function of two opposing forces: (1) the travelers' evaluation of performance and (2) the travelers' expectations. The net result of these two forces leads to two competing views regarding the nature of popularity-difference bias: the first view is performance-dominant, whereas the second one is expectation-dominant. Through analyzing a large set of restaurant reviews from a major online reputation platform, we find empirical evidence in support of the performance-dominant view. Specifically, we find that popularity-difference bias affects both the assigned rating and the text-encoded sentiment of a review. When reviewers travel to a less popular location than their hometown, popularity-difference bias is negative. To the contrary, when reviewers travel to a more popular location than their hometown, popularity-difference bias is positive. Popularity-difference bias affects the average rating of restaurants up to 11%. As a result, a restaurant's ratings skew lower if the restaurant tends to attract guests from more popular locations, whereas they skew higher if the restaurant tends to attract guests from less popular locations. This effect on ratings alters the probability that an average customer will consider a restaurant by up to 16%. Finally, awareness of popularity-difference bias allows managers to improve the design of their ranking systems: we show that such improvements can lead to up to 12% higher reviewer satisfaction, and up to 24% more diversified top-restaurant recommendations.