B1658
Title: A Bayesian test for the equality of two coefficients of variation
Authors: Francesco Bertolino - University of Cagliari (Italy)
Silvia Columbu - University of Cagliari (Italy)
Mara Manca - University of Cagliari (Italy) [presenting]
Monica Musio - University of Cagliari (Italy)
Abstract: A Bayesian Discrepancy Test (BDT) for the equality of two coefficients of variation (CVs) concerning independent populations is proposed. It is based on the Bayesian Discrepancy Measure (BDM), a bayesian measure of evidence recently introduced. When taking into account some distributions, such as Lognormal, Poisson, Gamma, Pareto and Weibull, the problem of testing the equality of the CVs of two independent populations is reduced to testing the equality of one of the parameters of the populations considered and the BDT can easily be applied. This simplification does not hold when considering two normal independent populations, but the comparison of their CVs can still be readily managed by using the BDT. The advantage of the proposed procedure is that it does not rely on asymptotic assumptions, unlike most frequentist statistical tests addressed in literature.
