|
Activity Number:
|
443
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics and Marketing
|
| Abstract - #301457 |
|
Title:
|
The Application of Poisson Race Model to Conjoint Choice Study with Multiple Alternatives
|
|
Author(s):
|
Shiling Ruan*+ and Steven N. MacEachern and Thomas Otter and Angela M. Dean
|
|
Companies:
|
U.S. Food and Drug Administration and The Ohio State University and Goethe University Frankfurt and The Ohio State University
|
|
Address:
|
Division of Biostatistics, Center for Devices and Rodiological Health, Rockville, MD, 20850,
|
|
Keywords:
|
Poisson race model ; conjoint choice experiment ; multinomial logit model ; attribute-based process ; stochastic process ; dependence structure
|
|
Abstract:
|
Poisson race model with a dependence structure has been shown to model the conjoint choice data with two alternatives. Drawing upon the results of the model with two alternatives, this paper discusses the formulation of Poisson race model with a dependence structure for more than two alternatives. This model takes into account the various dominance relationships among the attributes of alternatives and derives an attribute-based choice process with a stochastic model. The model is then applied to real conjoint choice data with a mixture of two or three alternatives in a choice sets. The model parameters are estimated by hierarchical Bayesian methods. The Poisson race model demonstrates its ability to capture interesting choice behaviors with superior predictive performance as compared to traditional choice models including the multinomial logit model.
|
