Search

The extant marketing literature has demonstrated the impact of consumer reviews on firm performance outcomes at the aggregate level and focused primarily on the role of star ratings, not review content. Only a few papers have investigated how consumers make choice decisions based on review content. In this research, we examine how consumers select a review to read, how they interpret the review’s content, and how the review may affect their choices. Inspired by the psychology theory on pre-decision confirmation bias, we conducted an incentive-aligned review-based choice experiment to test i) whether confirmation bias influences consumers to select more positive reviews of a product offered by their preferred vs. non-preferred brand to read (selection bias) and ii) whether they interpret information in favor of their preferred brand’s product (evaluation bias). We then incorporate both mechanisms of confirmation bias into a structural choice model with quality learning and show how confirmation bias can manifest as belief updating bias. With the structural model, we further perform counterfactual experiments to show how a retail platform can re-design its review display system to minimize confirmation bias.

The extant marketing literature has demonstrated the impact of consumer reviews on firm performance outcomes at the aggregate levels and focused primarily on the role of star ratings, not review content. Only a few papers have investigated how consumers make choice decisions based on review content (e.g.,Liu, Lee and Srinivasan 2019; Wu et al. 2015). In this research, we aim to examine how consumers select reviews to read, how they interpret the review content, and how it may affect their choices. Given unobserved orders of reviews read in clickstream browsing and purchase data, we conducted an incentive-aligned review-based choice experiment to test i) whether confirmation bias as a result of endowed brand preference influences consumers to select more positive reviews of a product offered by a preferred vs. less preferred brand to read and ii)whether they interpret information in favor of their preferred brand’s product.We confirm these hypotheses based on model-free evidence. We further build a choice model with quality learning to examine how researchers may capture these biases in observational data by relaxing standard learning model assumptions. Based on this model, we conceptualize that confirmation bias in favor of a brand may affect prior beliefs about the quality of its product, review selection, review interpretation, and the extent to which consumers update quality beliefs based on review information. The model also allows us to calculate consumer welfare loss due to confirmation bias in review selection and information processing.