Seminar on History and Philosophy of Science
Abstract: This paper develops a new approach to combining individual beliefs. We call it the evidence first method. Instead of combining credences alone, as the prevailing approaches in the literature, we focus instead on eliciting the group's full probability distribution on the basis of the evidence available to its individual members. This method has at least three main benefits: (1) it captures the weight, or resilience, of a group's belief; (2) it is sensitive to correlation among its individuals; and (3) it is commutative under updating. We also show that under some (but not all) circumstances, this method coincides with weighted averaging, and with the Upco rule developed in Easwaran et al. (2016).
Co-authored with Anil Gaba (INSEAD), Ilia Tsetlin (INSEAD), Robert L. Winkler (Duke)
Starting July 2021, Babic will be an Assistant Professor of Philosophy and of Statistics (joint) at the University of Toronto.