Does the Better-than-Average Effect Show that People are Overconfident? : Two Experiments

We conduct two experimental tests of the claim that people are overconfident, using new tests of overplacement that are based on a formal Bayesian model. Our two experiments, on easy quizzes, find that people overplace themselves. More precisely, we find apparently overconfident data that cannot be accounted for by a rational population of expected utility maximizers who care only about money. The finding represents new evidence of overconfidence that is robust to the Bayesian critique offered by Benoît and Dubra (2011). We discuss possible limitations of our results