Delving into Choice Internals: A Joint Discrete Choice/Attribute Rating Model

Multi-attribute modeling has rapidly progressed from being a novelty to becoming a widely used tool of economic research. When both choice and attribute ratings data are available, a model that makes joint use of both offers informative inference opportunities. In the present study we develop a joint model which utilizes both choice and ratings data, allows for scale usage heterogeneity, is robust to violations of utility continuity and completeness. The model is used to obtain WTP estimates for genetically-modified content and country-of-origin attributes in an survey-based study of Canola oil labeling. The median survey respondent's WTP for non-GM Canola oil was found to be CA$0.92/liter. The median WTP for non-specification of GM content was found to be approximately 80% of the WTP for the explicitly non-GM-labeled product. The median WTP to purchase Canada-made Canola oil versus a U.S. product was estimated to be CA$0.86/liter