The Sound of Silence in Online Feedback: Estimating Trading Risks in the Presence of Reporting Bias

Abstract

Most online feedback mechanisms rely on voluntary reporting of privately observed outcomes. This introduces the potential for reporting bias, a situation where traders exhibit different propensities to report different outcome types to the system. Unless properly accounted for, reporting bias may severely distort the distribution of public feedback relative to the underlying distribution of private transaction outcomes and, thus, hamper the reliability of feedback mechanisms. This study offers a method that allows users of feedback mechanisms where both partners of a bilateral exchange are allowed to report their satisfaction to “see through” the distortions introduced by reporting bias and derive unbiased estimates of the underlying distribution of privately observed outcomes. A key aspect of our method lies in extracting information from the number of transactions where one or both trading partners choose to remain silent. We apply our method to a large data set of eBay feedback. Our results support the widespread belief that eBay traders are more likely to post feedback when satisfied than when dissatisfied and are consistent with the presence of positive and negative reciprocation among eBay traders. Most importantly, our analysis derives unbiased estimates of the risks that are associated with trading on eBay that, we believe, are more realistic than those suggested by a naïve interpretation of the unusually high (>99%) levels of positive feedback currently found on that system.