A test to assess the dynamic evolution of preferences in marketing surveys

The subject of this paper is a two-stage hypothesis test, which has interesting applications in several situations related to the opinion research field. Such test is based on the components of a Bivariate Correlated Normal random variable. In particular, it is based on the exact distribution of their minimum and maximum modulus. This test was proposed for the first time by Duncan [Miller, 1981], and was recently improved by Pollastri [2008]. In this paper a variant of such test is described, since two samples in two different times or situations are considered. Two kinds of applications are provided to show the wide range of usage