Provision, Interpretation and Effects of Feedback in Reputation Systems.

Abstract

This dissertation addresses three questions that arise out of the voluntary and subjective provision of feedback and its subjective interpretation. The first study is an experimental investigation of a mechanism designed to give buyers incentives for feedback provision. It starts from the intuition that if buyers' feedback giving histories are made available to sellers, reputation effects will arise on the buyer side and act as incentives for buyers to provide more feedback. In a controlled laboratory experiment I examined whether this mechanism had the desired effects on buyer and seller behavior, and found that the mechanism had mixed success. It resulted in higher levels of feedback, but not higher levels of trust, trustworthiness or gains from trade. In the second study, I conducted an empirical analysis of transaction and feedback data from eBay, to understand the impact of a negative feedback in the seller’s profile on his own future behavior, and behavior of the buyers he gets matched with. I found evidence that after sellers got negative feedbacks; they were less likely to list items again, at least for a while. For sellers who continued to sell, the presence of a negative feedback in the recent past made buyers more willing to give another negative feedback. In the third study, I examined the effect of culture on behavior in Reputation Systems. In a controlled laboratory experiment using student subjects having two different nationalities, I found that the differences in trust and trustworthiness between the two groups prevailed in spite of the reputation system. I also found that there were systematic differences in how people from the two groups interpreted comparable reputation profiles.