Is the crowd's wisdom biased? A quantitative asessment of three online communities

Abstract

This paper presents a study of user voting on three websites: Imdb, Amazon and BookCrossings. It reports on an expert evaluation of the voting mechanisms of each website and a quantitative data analysis of users' aggregate voting behavior. The results suggest that voting follows different patterns across the websites, with higher barrier to vote introducing a more of one-off voters and attracting mostly experts. The results also show that that one-off voters tend to vote on popular items, while experts mostly vote for obscure, low-rated items. The study concludes with design suggestions to address the "wisdom of the crowd" bias.