Information Quality of Online Reviews in the Presence of Potentially Fake Reviews

Abstract

Online reviews are important in the evaluation of product quality. This paper seeks to assess information quality of online reviews using the TripAdvisor data for Korean hotels. We first estimate the review model developed by Dai, Jin, Lee, and Luca (2012) and show that high-quality reviews contain most of the information for the quality of hotels. Second, we assess the degree of distortions caused by fake reviews through numerical experiments and show that the distortions of fake reviews are serious. Third, we compare the simple average and weighted average aggregation methods. Weighted average method is better than simple average in finding the quality of hotels but it is more vulnerable to fake reviews. Fourth, we suggest excluding low-quality reviews to deal with fake reviews and show that the benefit of avoiding serious distortions from potentially fake reviews is greater than the cost of losing information from low-quality reviews.