Evaluating R&R of ordinal classifications with CUB model

The CUB model is a class of model for ordinal data obtained as a mixture distribution. It is adopted to assess the reliability of categorical measurement in several and different contexts, such as marketing and sensometrics. One of the main opportunities of this model is to include subjects’ covariates in its probability structure. Motivated by the work of De Mast and Van Wieringen (2010), in this paper we propose to modify the CUB model in order to define Repeatability and Reproducibility indexes (R&R) for ordinal data in business and industry framework.