Design for discrete choice experiments (Q2799906)

This article begins with an introduction which provides the background material. Then choice models, including the multinomial logit model, mixed logit model and other extensions are introduced. Turning to the design problem, first paired comparison experiments are considered with reference to models that involve only the main effects, and main effects as well as two-factor interactions. For both situations, optimal approximate designs and associated constructions are discussed, along with examples and tables. Next, designs for larger choice sets are studied with a discussion of issues relating to efficiency gains, relationship with block designs, and local optimality. Thereafter, computational approaches for the multinomial and mixed logit models are discussed. The article then dwells on ranking based methods and various miscellaneous problems, before ending with the concluding remarks.NEWLINENEWLINEFor the entire collection see [Zbl 1327.62001].