|
Activity Number:
|
280
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 14, 2002 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and Marketing
|
| Abstract - #301814 |
|
Title:
|
A Random Coefficients Multinomial Logit Model for Choice-Based Conjoint Studies
|
|
Author(s):
|
James Lymp*+ and W. O'Fallon and Sherine Gabriel and Veena Nayar and Kent Seltman and Megan Maurer
|
|
Affiliation(s):
|
Mayo Clinic and Mayo Clinic and Mayo Clinic and Mayo Clinic and Mayo Clinic and Mayo Clinic
|
|
Address:
|
200 First St. S.W., Rochester, Minnesota, 55905,
|
|
Keywords:
|
Conjoint Analysis ; Discrete Choice Modeling
|
|
Abstract:
|
Choice-based conjoint studies are used in economics and marketing to assess the relative contributions of various product attributes and to predict consumer behavior. In these studies, subjects are given a set of questions. Each question is a scenario containing several choices from which the subject makes a selection. The choices are characterized by various attributes and particular levels of these attributes. Based on these stated preferences of the subjects, inference is made on the effects of various attributes and their levels on decision making. A commonly used model for these studies is the multinomial logit (MNL) model. We develop the MNL model for choice-based conjoint studies and relate the MNL model to well-known biostatistical models. We also describe a Markov Chain Monte Carlo method for adding random coefficients to the MNL model. We describe how the MNL model can be interpreted in terms of odds ratios and attributable risk estimates. We fit the random coefficients MNL model to data from a choice-based conjoint study on patient preference for tertiary care medical centers.
|
