Predictive validity across product categories with similar attributes.

If the conference organizers make the offer, this paper will go onto the Journal of Choice Modeling In the study of consumer behavior the emphasis can be on attributes appearing in multiple product categories. The general area of diet, ‘lite’, low fat and low cal is an example of such an attribute but so too is brand and pack size. A single brand can hold a strong position in several categories. A consumer choosing bulk packs or family size may do so consistently. Even in heterogeneous willingness-to-pay studies the disutility for cost may or may not be consistent across categories. In the different categories the attribute might be described similarly but not the same. Low cost is an attribute that will manifest differently in each category due to differences in reference prices. Conversely, for some attributes the descriptions might be identical. The same brand name might be printed on all packs using the same font, color, size etc. Nevertheless, at least initially, a sensible approach is to assume that regardless of how similar the attribute descriptions might be there will be differences across categories to the consumer, as perceptions, and thus differences in the impact on choice. Consequently, the validity and accuracy of research using discrete choice experiments (DCE) is improved if the similar attributes are analyzed using separate DCEs for each category, rather than a single and combined experiment where the choice se