Assessing heterogeneity in consumer analysis across product similarities and within consumer’s differences

8 September 2017
 Download pdf

ASA 2017 - StatFood, Napoli, 7-9 September
Statistical Methods and Models for Classifying, Choosing and Experimenting Food and Wine
Convegno dell’Associazione Italiana per la Statistica Applicata
Session: Evaluating Consumers’ Perception

In consumer liking studies, the main objective is to identify the product or group of similar products that maximize consumer preferences and at the same time to analyze what are the properties of the products that most affect liking (Menichelli et al., 2013). In such studies, it is of crucial importance also to analyze the individual differences (Endrizzi et al., 2011) in order to identify consumer segments that tend to like the same types of products. The aim of this paper is to present a new statistical approach that allows to analyze both the a-priori assumed hetero- geneity between products and the latent heterogeneity between consumers. It is based on the use of Quantile Regression (QR) (Davino et al., 2013), that may can be considered an extension of classical least squares estimation of conditional mean models to the estimation of a set of conditional quantile functions. Recently, QR has been used in consumer studies for estimating the conditional quantiles of liking as functions of the consumer characteristics (Davino et al., 2015). In this paper, it is used to evaluate how much specific liking may influence disagreement in overall liking, which can be equally important as the predicted average liking. Furthermore, the proposed approach is also used to explore how the overall liking varies according to the dif- ferent products. The proposal allows to handle heterogeneity among units Davino et al. (2013) identifying a different dependence structure according to the considered product. The approach is illustrated by a case study based on real data.

« A chemometric approach to evaluate the concentration of polychlorinated biphenyls and heavy metals in M. merluccius from Mediterranean Sea | Tutorial: Quantile Regression with R »