Confidence Envelopes for Monotonic Functions: Principles, Derivations, and Examples.

摘要

The bounds of plausible behavior for a multiparameter monotonic function can be expressed as a stochastic confidence envelope around the function. This paper presents the derivation of a general method for constructing such bounds. The bounds described are first specialized to the linear case, yielding results identical to known results found by other means. The method in the more general nonlinear case is then illustrated on three logistic item response models.