Meta-Inductive Probability Aggregation

Abstract There is a plurality of formal constraints for aggregating probabilities group individuals. Different characterise different families aggregation rules. In this paper, we focus on the linear and geometric opinion pooling rules which consist in linear, respectively, weighted averaging individuals’ probabilities. For these families, it debated weights exactly are to be chosen. By applying results theory meta-induction, want provide general rationale, namely, optimality, choosing success-based way by scoring A major argument put forward against weighting that heavily depend chosen rule. However, as will show, main condition optimality meta-inductive so holds under most standard rules, more precisely all based convex loss function. Therefore, whereas exact choice rule probability might respective purpose such an aggregation, epistemic rationale behind generally valid.