A truth-serum for non-Bayesians : correcting proper scoring rules for risk attitudes

Proper scoring rules provide convenient and highly efficient tools for eliciting subjective beliefs. As traditionally used, however, they are valid only under expected value maximization. This paper shows how proper scoring rules can be generalized to modern ("nonexpected utility") theories of risk and ambiguity, yielding mutual benefits: the empirical realism of nonexpected utility is introduced in proper scoring rules, and the beauty and efficiency of proper scoring rules is introduced in nonexpected utility. An experiment demonstrates the feasibility of our generalized proper scoring rule, yielding plausible empirical results.