This document summarizes research analyzing bias in eBay's online feedback system. The research uses a dataset of 51,052 rare coin auctions from 2002. Two models are presented: 1) a simultaneous equations model estimating the probabilities of different transaction outcomes and feedback patterns, and 2) adding feedback timing information. Both models estimate lower positive transaction outcomes than positive feedback posted, indicating bias. The research also finds evidence of reciprocity in feedback between buyers and sellers. In conclusions, the research presents a methodology for extracting information from silent transactions and analyzing how reciprocity impacts negative and positive feedback reporting behavior.