Hypothesis Testing for Coefficient of Variation in an Inverse Gaussian Population

• Activity Number: 690
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Computing
• Author(s): Debaraj Sen*+ and Yogendra P. Chaubey and Krishna K. Saha
• Companies: Concordia University and Concordia University and Central Connecticut State University
• Keywords: Maximal invariant ; Invariance ; Neyman-Pearson Lemma

Abstract

The inverse Gaussian distribution provides an attractive family of probability densities in modeling the coefficient of variation (CV) as it may conveniently be parameterized in terms of the mean and CV. Tests for mean and dispersion parameters have been investigated for this family in the literature, however, the coefficient of variation has not received much attention in this respect. Noting that the coefficient of variation plays a very important role in many practical data analysis situations, this article considers the uniformly most powerful invariant test for the problem. Some approximations to the distribution of the resulting test statistic have been investigated.