In Praise of Subjectivity?
Bayesian analysis requires that probabilities are subjective. Attempts to escape this apparently unwelcome fact are numerous, and they are ultimately misguided because science itself is necessarily subjective. Instead, we should embrace the opportunity to incorporate additional knowledge into the analysis through the prior distribution. But that doesn’t make subjectivity admirable in itself. Science strives towards objectivity and so should we – all subjective input should be based on careful, rigorous judgment. In this talk, I will enlarge on these ideas with various examples.