Evaluation Inflation on Transaction Websites

Abstract

How reliable are online ratings? This research examines whether online ratings might be inconsistent across websites with different perceived main functions, comparing transaction sites (websites primarily selling products, e.g., Amazon) with communication sites (websites primarily facilitating information exchange, e.g., Goodreads). Seven studies, including a large-scale secondary data analysis (N = 148,988 matched products; N = 309,930,122 ratings) and six experiments (N = 2,335), consistently reveal an evaluation inflation phenomenon: for the same product, rating on a transaction site leads consumers to give higher average rating and proportion of five stars, compared to communication sites or a control group (making no mention of platform). Despite being multiply-determined, process evidence highlights conformity as one important, previously underexplored contributor to this phenomenon: consumers perceive a norm to give perfect scores on transaction sites, and conform to this norm by boosting their ratings. Consistently, evaluation inflation is mitigated when an alternative norm is prompted through reviewer recognition badges, or when the consumer's need for conformity is attenuated, which can be achieved by rating products from out-group sellers or by fostering independent thinking. We conclude by discussing implications for the online review literature and social influence research, as well as for consumers, marketers, and policymakers.

Keywords: online ratings, social influence, conformity, attitude expression