Empirically identify whether model-level and cross-model curvature coincide

Determine empirically whether real decision-makers' curvature parameters for model-level robustness and cross-model aggregation coincide, by separately identifying these two forms of ambiguity response in choice data.

Background

The theoretical results show that exact reduction to a single-layer entropic value requires matched curvature, \lambda=\xi, and that mismatched curvature cannot be repaired by changing the target curvature or baseline. However, the theorems do not establish whether actual decision-makers exhibit this equality.

The proposed unresolved empirical problem is to design choice data that separately identify responses to misspecification of a single model and responses to enlargement or contraction of the model set, allowing the matching restriction to be tested rather than imposed.