Intervals for Ratios of Variance Components in Nested Designs

The purpose of the paper is to consider interval estimation for ratios of variance components in two-way nested designs. We propose a novel approach for the interval estimation using the concepts of fiducial inference. A simulation study is conducted to investigate the coverage probabilities of the resulted intervals to the Modified Large Sample (MLS) intervals by Burdick and Graybill [1]. Relative to the MLS approach, the fiducial approach is closer to the nominal coverage for a two-sided interval. The fact shows that the fiducial approach is a suitable candidate for making inference about ratios of variance components.