Biased individuals, unbiased juries? An experimental analysis of Condorcet Jury Theorem

Supervisor: Marco Piovesan - University of Copenhagen
Date and time: Monday, May 18, 2009 at 1:00 PM - Biblioteca DSE, Palazzina 32 - Ex Caserma Passalacqua
Contact person: Angelo Zago
External reference
Publication date: March 24, 2009
Department

Summary

According to the Condorcet Jury Theorem (CJT), voting yields “positive information

aggregation” in a common interest situation: groups make better decisions than individuals

could have made, provided that individuals are more likely to be right than wrong. However,

if individuals are biased, i.e. if they tend to systematically choose the inferior option, the CJT

predicts poorer outcomes for groups than for individuals. We show that information

aggregation can cut either way. In particular, we find that groups do worse than individuals

when individual voters are biased. In addition, we test how providing social information

shapes voting outcomes in the presence of biases. In treatment

facts, voters learn thefacts are de-biasing and improve group performance as they alert voters toopinions, voters learn how popular theopinions

deteriorate group performance since biased voters find confirmation for their opinion and

non-biased voters may yield to conformity. We find support for both hypotheses.

percentage of individuals who provided the correct answer in a previous session. We

hypothesize that

how common biases are in the electorate. In treatment

alternatives were among voters in a previous session. We hypothesize that