Department of Mathematics, University of Tennessee, Knoxville, IN 37916, USA;Department of Statistics, Florida State University, Tallahassee, FL 32313, USA
Abstract:
Extremal problems in large deviations of the F-statistic are considered. It is shown that the slowest rate of convergence over a specified class of distributions of the F-statistic is slower than exponential, and that the Bahadur efficiency of the F-statistic with respect to some distribution-free competitors is identically zero.