Are we all overconfident in the long run? Evidence from one million marathon participants

In this project we sought to contribute to extant literature on overconfidence by identifying it in a large, heterogeneous sample making familiar, repeated choices in a natural environment which provides direct feedback. In Study 1 we elicited predictions of own finishing time among participants of the 2012 Warsaw Marathon. Their prediction errors turned out to be very highly correlated with the change in pace over the course of the run. In Study 2 we thus took this change in pace as a proxy for self-confidence and used existing field data of around one million participants. Both studies indicate that males as well as youngest and oldest participants tend to be more confident. In Study 2 we are also able to investigate national and cultural dimensions, confirming previously reported findings of relative overconfidence in Asians and providing some novel results, i.a. that relatively conservative societies tend to be more self-confident.