|
Abstract:
|
In some product categories, consumers’ individual preferences may consist of multiple distinct tastes defined over product attributes. Capturing heterogeneity among multitaste consumers requires new models, as a consumer simultaneously belongs to multiple segments. This is a different type of heterogeneity than that captured by existing models, such as mixed logit or latent class models, which estimate one taste per individual. In this paper, we propose a model that allows individual consumers to express multitaste preferences, and we provide an estimation procedure that scales to high-dimensional attribute spaces. Through extensive simulation experiments, we demonstrate that the proposed algorithm accurately recovers parameters, whereas single-taste benchmark models underfit and generate a misleading representation of both population- and individual-level preferences. We apply the algorithm to a large dataset of recipe choices to uncover rich patterns of preference heterogeneity. The proposed model fits the data better than single-taste benchmarks and provides additional individual-level insights.
|
