Approximating the distribution of discrimination rates in replicated difference tests using Bayes’ rule

A recently proposed approach to use Bayes’ rule for triangle tests in order to improve the estimates of interest is extended to experiments with replications. It is shown how this results in an empirical distribution of the true discrimination rates of the assessors. Two examples involving replicated triangle tests with equal and unequal numbers of replications illustrate the approach, respectively. The results are compared with those of a chance-corrected Beta-binomial model.