Overconfidence as a social bias: Experimental evidence

•We compare the overconfidence for an individual and a social, observational experiment.•We find realistic confidence levels in the individual setting on average.•Introducing the observation of other subjects considerably increases the overconfidence.•The social setting reduces underconfidence compared to the individual setting.•Our results suggest that overconfidence is a social rather than an individual bias.

The overconfidence bias is discussed extensively in economic studies, yet fails to hold experimentally once monetary incentives and feedback are implemented. We consider overconfidence as a social bias. For a simple real effort task, we show that, individually, economic conditions effectively prevent overconfidence. By contrast, the introduction of a very basic, purely observational social setting fosters overconfident self-assessments. Additionally, observing others' actions effectively eliminates underconfidence compared to the individual setting.