Modelling Consumers Preferences Using Risk Perceptions and Subjective Knowledge

Abstract

Latent class models constitute an alternative approach to the popular mixed logit models to capture sources of individual preference heterogeneity to explain discrete choices. This article applies the latent class model to incorporate random taste variation across individuals or consumers, encompassing the role of lexicographic decision-making rules in the modeling of discrete choice responses. Furthermore, it provides a deeper understanding of processing information rules, by including latent variables related to individuals' subjective knowledge and risk perceptions in the explanation of the probability of adopting a specific simplifying decision rule or heuristic to select the most preferred alternative. The results reveal that accounting for information processing strategies avoids misspecification bias and yields more plausible behavioral outputs, namely deflated willingness to pay measures. Thus, providing a closer comprehension of the consumers' behavior and choice. These results were derived from a discrete choice experiment applied to a sample of regular wine consumers in Switzerland.