UC Riverside

Consumers increasingly depend on online word-of-mouth to inform their purchase decisions. Thanks to the recent trend toward adding social network features to online recommendation platforms and vice versa, potential consumers can now see reviewers’ profiles and rating histories and use that information to seek out advice from users with similar preferences. We propose that consumers taking advice from a reviewer will do so according to how similar the reviewer’s preferences seem to their own, based on the degree to which they both liked or both disliked the same products or experiences in the past. We show that people make two systematic errors when making affective forecasts about potential products in the presence of preference similarity information. First, they tend to underestimate the degree of preference similarity with the reviewer, especially for mutual disliking. Second, they tend to overweigh her advice relative to a Bayesian advice-taker, particularly for negative advice. These errors bias affective forecasts in opposite directions, even cancelling each other out for negative advice, and can lead to surprisingly accurate forecasts.