Estimated confidence under the validity constraint (Q1184218)

The decision theoretic estimated confidence approach is examined. Particular attention is paid to results under the frequentist validity constraint \[ E_ \theta I(\theta\in C_ X) \geq E_ \theta\gamma(X), \] where \(\gamma(X)\) is a data dependent estimate of the value of the coverage function \(I(\theta\in C_ X)\) and \(C_ X\) is a confidence set for the parameter \(\theta\) based on the random variable \(X\). The main result shows that the usual constant coverage probability estimator for the usual confidence set of a linear model is admissible under the frequentist validity constraint. The matter of inadmissibility without this constraint is also discussed.