Consumer Choices Based on Online Ratings: The Diagnosticity of a Discrepant Rating

Abstract

This paper examines how consumers integrate other people’s opinions to evaluate an unknown product or service. The use of these ratings has mushroomed with the prevalence of online ratings by consumers in domains ranging from restaurants to movies. However, their credibility or diagnosticity is unknown and/or even skeptical, since they are given by anybody. We examine how consumers judge which ratings are more diagnostic among inconsistent ratings without knowing who the rating providers are. Specifically we focus on how consumers judge a discrepant rating’s diagnosticity, because a discrepant rating, a rating very different from other ratings, can be either negligible or influential on product evaluations. A two-stage aggregation process (rating screening stage and rating integration stage) is proposed to explain when a discrepant rating is discounted as an outlier and when valued as a loss or gain signal. Four experiments show that the diagnosticity judgment is moderated by the rating’s discrepancy level (Study 1,2,3). When a rating is highly discrepant, it is discounted as an outlier, but when a rating is less discrepant, it is judged to be valid and valued as a loss or gain signal (Study 1). Consumers discount a rating, if it is highly discrepant, even when data size is very small (Study 2). When ratings are negatively skewed, discrepancy discounting and loss aversion generate conflicted weights for a discrepant rating. Whether the rating is discounted or valued depends on the relative strength of discrepancy and loss aversion (Study 3). As a psychological factor affecting the diagnosticity judgment, a consumer’s expectation of ratings’ similarity was examined (Study 4). The greater the expected heterogeneity of preferences, the greater the value assigned to majority ratings, the smaller the value assigned to a discrepant ratings. Marketing implications for the format and display of online ratings are discussed.