All Times ET
Keywords: CUB models, Ordinal scales, Surveys
Feelings, emotions, and preferences are complex psychological processes of interest in several disciplines, such as marketing, economics and business. These human feelings are usually captured through surveys in which ordinal scales, such as Likert-type scales, are commonly adopted. Likert-type scales are made up of a set of items, usually formulated in terms of linguistic expressions coded into natural numbers, characterised by a rank order. Unfortunately, this kind of scales return vague and imprecise information. Firstly, individuals are asked to convert their thoughts into a linguistic expression and then to a natural number and these conversions can be inaccurate. Secondly, the items of a Likert-type scale can be subjectively interpreted by respondents due to their personal background and knowledge about the phenomenon investigated, their culture and understanding of the question. Fuzzy sets theory and CUB (Combination of discrete Uniform and shifted Binomial random variables) model are the two common ways to dial with the imprecision and vagueness of both Likert-type scale and human thinking. The CUB approach aims to model the final answer as a mixture of two internal aspects, feeling and uncertainty. The CUB model has been extensively studied in the last 15 years and several variants have been proposed. This article offers a review of the CUB models suggested so far in the literature. From this review, it emerges that most of the models focus on a single question and only two models, the SCUB model and the Copula CUB model, have been suggested to analyse multiple questions simultaneously. Knowing the practical importance of investigating a bundle of related questions together, in this work we will investigate the possibility to suggest a new multivariate CUB model.