Accounting for model uncertainty in individualized designs for discrete choice experiments

L. Deldossi, G. Consonni, E. Saggini

We consider the issue of finding optimal designs for Discrete Choice Experiments (DCE). In particular we deal with individualized designs, which are sequentially generated for each person, using responses from previous choice sets in order to select the next best set in a survey. Currently the optimal design for a DCE is predicated on a specific mixed logit model. Using a unique model represents a major limitation which we overcome by allowing for a collection of different models characterized by distinct linear predictors (e.g involving only main effects or including some interactions). In this new setting, model choice becomes a major concern. Accordingly, we specify the “utility” of selecting a choice set as the mutual information between the model indicator and the predicted observation at that set. By maximizing this utility over alternative choice sets model discrimination is enhanced.

Palabras clave: Optimal Design, Model Discrimination, Entropy, Bayesian sequential Design, Sequential Monte Carlo algorithm

M10.1 Sesión Hispano-Italiana: Diseño Óptimo de Experimentos
6 de septiembre de 2016  17:00
0.02 - Aula de proyectos 1

J. A. Moler Cuiral, F. Plo Alastrué

K. Mylona, M. Borrotti, F. Sambo, S. Gilmour

F. Rapallo, R. Fontana, M. P. Rogantin