Activity Number:
|
134
|
Type:
|
Invited
|
Date/Time:
|
Monday, August 12, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Business & Economics Statistics Section*
|
Abstract - #300284 |
Title:
|
A Learning-based Model for Partial Profile Conjoint Data
|
Author(s):
|
Eric Bradlow*+ and Teck-Hua Ho and Ye Hu
|
Affiliation(s):
|
University of Pennsylvania and University of Pennsylvania and University of Pennsylvania
|
Address:
|
3620 Locust Walk, Suite 1400, Philadelphia, Pennsylvania, 19104-6371,
|
Keywords:
|
|
Abstract:
|
We investigate how respondents infer missing information about product attributes in a conjoint partial profile by developing a general Bayesian learning model for the imputation process that nests several extant models as special cases. The advantage of this approach over previous research is that it simultaneously allows for learning across time, learning across attributes due to covariation, and differential individual-level learning. A second goal is to evaluate whether the partworth utilities derived using partial profiles are systematically different than those obtained from full profile data. This is examined by an experiment designed to explore the potential biases.
|