In Defense of Fishing

Abstract

Using an example from animal cognition, I argue that the problems of bias—inherent in choosing null hypotheses or setting Bayesian priors—can sometimes be avoided altogether by collecting more and better observational data before setting up tests of any sort.