How perceived variety impacts on choice satisfaction: a two-step approach using the CUB class of models and best-subset variable selection

In consumer research, marketing, public policy and other fields, individuals' choice depends on the number of possible alternatives. In addition, according to the literature, the choice satisfaction is influenced not only by the number of options but also by the perceived variety. The aim of the present study is to apply a novel approach to model perceived variety, in order to better understand the perceptions of individuals about the variety of the possible choice options and to model the impact of perceived variety and individuals' characteristics on the choice outcome satisfaction. We resort to the class of cub (Combination of Uniform and Binomial random variables) models for rating data that model the respondents' decision process as a combination of two latent components, called feeling and uncertainty, that express, respectively, the level of agreement with the item being evaluated and the human indecision surrounding any discrete choice. The model applied in this paper is an alternative to the most common models used in the studies of human judgments and decisions, whenever attitudes, perceptions and opinions are measured by means of questionnaires having questions with ordered response categories. The chosen approach is composed of two steps: (1) we construct measures of feeling and uncertainty of perceived variety by means of cub and (2) we investigate their impact (eventually together with personal characteristics) on choice satisfaction. The R FastCUB package is exploited to select the best set of covariates to include in the final model.