Research Abstracts Online
January 2009 - March 2010
University of Minnesota Twin Cities
Curtis O. Carlson School of Management
Department of Operations and Management Science
PI: William Li, Associate Fellow
Co-PI: Christopher J. Nachtsheim
Discrete Choice Experiments With Fatigue and Model-robust Designs
The first part of this project is a continuation of previous research. The researchers are investigating internet-based discrete choice experiment (DCE) surveys, which are a popular form of marketing research due to the wide availability of on-line respondents. These methods are usually cost-efficient, but the fatigue of respondents can be a major concern. While efficient experimental designs have been proposed and used for such survey methods, none of existing designs directly take into account the impact of the fatigue effect. This study was motivated by an internet-based survey that polled respondents on their choices of a glass supplier. The researchers estimate the fatigue effect via a mixed multinomial logit model. They then study the impact of fatigue on different experimental designs, using several penalty functions for the fatigue effect.
The second part of the project involves a new class of model-robust DCE. The existing "optimal” DCE designs usually assume a given model. When the true model is unknown, especially with the presence of interaction effects, such optimal designs may not perform badly. These researchers have developed a new criterion based on previous work and extend the idea to the construction of DCE designs.
Kenny Ye, Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, New York, New York