The consumer preferences of Parmesan cheeses in foreign countries: a non parametric analysis using CUB models

In the last decade the exports of parmesan cheese has been reinforced in the north American countries (US and Canada) because of a strong increase in consumption. However, factors boosting this consumption are not always so obvious. For instance, in US both Reggiano and Grana Padano are generally classified as “Parmesan Reggiano and Parmesan Padano” where the term Parmesan is well-known while the differentiation between Reggiano and Padano is often not appreciated or not so clear for US consumers. This paper analyzes data of a survey carried out on US and Canada about the Parmesan’s consumption through a class of mixture models with covariates know as CUS models. The survey, started in 2011, was done in some restaurants chains in US and Canada. A brief questionnaire was fill out by 540 consumers to get information about the of knowledge and appreciation of Parmesan, purchase features as well as factors influencing the purchase and willingness to pay. Information about consumer’s profile were also collected. A descriptive analysis of result show that the Parmesan (as general cheese) is known and appreciated by the majority of respondents, but most of them do not know the difference between Parmigiano Reggiano and Grana Padano. To deeply investigated a CUB model was employed. In particular, the CUB model allows us to better explore the relationship between the consumers and items (answers) and to summarize the distribution of preferences (ranking) expressed by the respondents (as 7 point Likert scale). Using the qualitative judgments of preferences and the overall distributions of rating, the CUB model makes comparisons in a inference context and it associates the consumers features to their evaluations. The CUB model is able to explain, in a direct way, the option that a subject expresses by interpreting it as a mixture of feeling and uncertainty. Feeling and uncertainty and latent components are, therefore, essentially continuous and never observable, and allow a connection with the individual characteristics (covariates) of consumers. The CUB has firstly investigated the following questions: "Level of knowledge of Parmesan", "Degree of appreciation of Parmesan", "Frequency of purchase", "Knowledge of the differences between Parmesan Padano and Parmesan Reggiano" and "Willingness to pay for Parmesan.” The analysis starts by analyzing all consumers and it generates alternative models according to differences in consumers ratings; next the analysis chooses the best models outcome and introduces covariates (consumer’s features) to recognize factors affecting feeling and uncertainty. Firsts results seem to shows a variability in ratings expressed by consumers, except for the level of knowledge where the distribution of scores is similar to the distribution of probability assessed by the model. Secondly, the introduction of covariates (such as country, age, etc.) has identified the effect of factors affecting consumer ratings. Among results, the CUB model shows a different consumer behavior in consuming Parmesan while the comparison between US and Canada revealed significant differences in consumer habits and expectations.