On the Use of Alternative-Specific Designs in Choice-Based Conjoint Analysis

In this paper, we’d like to draw the attention of marketing academics and marketing practitioners to the use of alternative-specific designs (ASD) for choice-based conjoint (CBC) studies in marketing research. As opposed to transportation research, where the ASD approach for discrete choice experiments (DCE) or discrete choice modeling in general originated from and is still widely used, this design method rather tends to lead a shadowy existence in marketing research despite its obvious superiority for generating more realistic choice situations for respondents compared to the application of the generic designs. The latter are commonly employed in conjoint experiments in a marketing context, for example for product design and related market simulations and optimizations. We discuss why and when it is recommendable to use an ASD, propose a typology for characterizing different types of ASD structures, and work out the particularities of interpreting the estimation results of an ASD-CBC model and calculating related willingness-to-pay quantities. For illustration, we use an empirical example based on an ASD-CBC study on consumer preferences for electric vehicles in the UK. We further point to advanced modeling options for enriching or extending ASD-CBC models, including alternative-specific hybrid choice models and nonlinear utility specifications. Finally, we discuss the state-of-the-art how artificial intelligence (AI) is already influencing the domain of conjoint analysis and DCE.