Abstrakt: It is often suggested that when individuals within groups hold different opinions, the individual opinions should be aggregated to form a collective or group-level opinion. This paper compares two aggregation strategies, linear pooling (or weighted averaging) and what I call opinion agglomeration (or collective imprecise probabilities). Within the literature, aggregation strategies are usually assessed by some set of desirable criteria they satisfy or fail to satisfy. In continuing this tradition, I provide a novel pragmatically-focused criterion for opinion aggregation, entailing that no aggregation strategy should invite anticipation of regret from a group’s perspective upon committing to the betting prices for contracts on events yielded by aggregated opinions. I show that only opinion agglomeration satisfies this constraint and give normative and empirical arguments in defense of the criterion.
Spotkanie odbędzie się w języku angielskim.