Are preferences for food quality attributes really normally distributed? An analysis using flexible mixing distributions. (September 2018)

• Record Type: Journal Article
• Authors: Caputo, Vincenzina; Scarpa, Riccardo; Nayga, Rodolfo M.; Ortega, David L.
• Is Part Of: Journal of choice modelling. Volume 28(2018)
• Page Start: 10
• Page End: 27
• Publication Date: 2018-09
• Subjects: Flexible taste distributions -- Mixed logit -- Logit mixed logit -- Food preferences -- Preference heterogneity; Decision making -- Periodicals; Social choice -- Periodicals; Decision making; Social choice; Periodicals; 302.13
• DOI: 10.1016/j.jocm.2018.04.003
• Languages: English
• ISSNs: 1755-5345

Abstract

Abstract: We empirically question the commonly invoked assumption of normality of taste distribution in mixed logit models with continuous random parameters. We use a WTP-space random utility discrete choice model with flexible distributions on data from two choice experiments regarding beef with nested set of quality attributes. We specifically focus on distributional features such as asymmetry, multi-modality and range of variation, and find little support for normality. Our results are robust to attribute dimensionality in experimental design. Implications of our results for practitioners in the field are discussed. Highlights: We empirically question the distributional assumption of normality of taste distribution in mixed logit models. We use a WTP-space random utility discrete choice model with flexible distributions. We specifically explore estimates' sensitivity to the definition of the random coefficients' range of variation. We explore different distributional features across cue and independent attributes when extending the attribute space. Results from this study indicate that non-normal distributional features prevail. Our findings highlight the advantage of using flexible distributions in mixed logit analysis to check the robustness of distributional assumptions.