Replicated triangle and duo–trio tests: Discrimination capacity of assessors evaluated by Bayes’ rule

Discrimination rates of panellists performing replicated difference tests are estimated in the present paper according to Bayes’ rule by considering the successive replications as different steps and using the posterior distribution obtained in each step as prior distribution of the following step. Data are also successively obtained in real situations and, thus, this approach imitates what happens in practice and differs from other published approaches which consider all replications as if they were simultaneously observed. Before performing the first step, a non informative prior distribution is used as density function of the discrimination rate but after the first step has been completed each prior distribution (posterior distribution obtained in the previous step) is informative. The density functions of the discrimination rates are not proper beta distributions but the sections of beta distributions corresponding to the interval of the independent variable from 1/3 to 1 in triangle tests and from 1/2 to 1 in duo–trio tests.