ENBIS: European Network for Business and Industrial Statistics

Teacher: Roselinde Kessels

Choice experiments, or more formally, choice-based conjoint experiments, measure the trade-offs people make in choosing between alternative products by quantifying people's preferences with respect to pre-specified product attributes. The design of these experiments is different from any classical design based on the principle of orthogonality. Moreover, choice-based conjoint designs based on orthogonal arrays have substantial drawbacks.

Currently, the Choice Design platform in JMP 8 handles the design of choice experiments for the multinomial logit model best using Bayesian optimal design techniques meeting the specific experimental needs. This course provides an introduction to the multinomial logit model, an overview of the Bayesian design methodology used and a hands-on approach to the creation of Bayesian choice designs in JMP 8. Through a real-life case study the analysis of choice data is also demonstrated.