Abstract
This article presents an extension of the methodological development proposed by Chapman and Staelin (1982) for the estimation of individual-level binary logit preference models from conjoint-type data. These individual binary logit models are compared with traditional conjoint models, in terms of estimation accuracy and predictive efficiency. Both simulation and empirical results suggest that binary logit analysis yields models which are equivalent to those produced by conjoint analysis. The implications of these results for the potential reduction of data requirements for individual-level preference modeling are discussed.
Additional information
Notes on contributors
Rose S. Prave
Rose S. Prave is Assistant Professor of quantitative methods at the University of Scranton. She received her Ph.D. in Management Science from the Pennsylvania State University. Her research interests are in the areas of marketing research, total quality management, and forecasting. She has published in Decision Sciences, Production and Inventory Management Journal, and the Journal of Education for Business.
J. Keith Ord
J. Keith Ord is the David McKinley Professor of Business Administration and Professor of Statistics at the Pennsylvania State University. His research interests relate to the application and development of statistical models for business processes. He is co-author ofThe Advanced Theory of Statistics with Alan Stuart and has published in a variety of journals including Decision Sciences, Management Science, and the International Journal of Forecasting.