Title: Improved Confidence Intervals for the Ratio of Coefficients of Variation of Two Lognormal
Abstract: There are practical situations where one encounters data that are skewed and contain a relatively high proportion of zeros. These type of data can be modeled with lognormal distribution. The problem of estimating the ratio of coefficients of variation of two independent lognormal populations is considered. We proposed two closed-form approximate confidence intervals (CIs), one is based on the method of variance estimate recovery (MOVER), and another is based on the fiducial approach. The proposed CIs are compared with another CI available in the literature. Our new confidence intervals are very satisfactory in terms of coverage properties even for small samples, and better than other CIs for small to moderate samples. The methods are illustrated using an example.