Simply Statistics A statistics blog by Rafa Irizarry, Roger Peng, and Jeff Leek

When overconfidence is good

A paper came out in the latest issue of Nature called the “Evolution of Confidence”. The authors describe a simple model where two participants are competing for a resource. They can either both claim the resource, only one can claim the resource, or neither can. If the ratio of the value of the resource over the cost of competition is good enough, then it makes sense to be overconfident about your abilities to obtain it. 

The amazing thing about this paper is that it explains a really old idea “why are people overconfident” with really simple models and simulations (done in R!). Based on my own experience, I feel like they may be on to something. You can’t get a paper in Nature if you don’t send it there…