On optimal confidence sets for parameters in discrete distributions

In discrete distributions, the coverage probability and the expected length of an interval estimator often depend on the unknown parameter of interest. Some authors have suggested that “good” interval estimators should have mean coverage probability near the nominal level and small mean expected length, where the mean is taken over all possible values of the parameter. This paper uses these criteria to precisely define an optimal interval estimator and finds it in the single-parameter discrete distribution setting.