Article ID: | iaor200968850 |
Country: | United States |
Volume: | 55 |
Issue: | 6 |
Start Page Number: | 1003 |
End Page Number: | 1017 |
Publication Date: | Jun 2009 |
Journal: | Management Science |
Authors: | Park Young-Hoon, Ding Min, Bradlow Eric T |
Keywords: | statistics: experiment |
We propose a new alternative preference measurement method, barter conjoint, to contrast with traditional choice-based conjoint (CBC) approaches. Barter conjoint collects a substantially larger amount of data compared to CBC and allows for information diffusion among respondents. We conducted two empirical studies that compare CBC (with and without incentive alignment) and barter conjoint. The studies employed a total of three product categories, each with two validation tasks (one follows immediately and one conducted two weeks later). Our results confirmed prior research that incentive alignment, in general, substantially improves out-of-sample predictive performance of CBC. Furthermore, we found barter conjoint performs substantially better than the incentive-aligned CBC. However, in the spirit of “no free lunch,” barter conjoint is more taxing (in various ways) than CBC, suggesting a potential trade-off between consumer resource allocation (at the time of the task) and (managerial) predictive accuracy downstream. Given that this is the first study on barter conjoint, we discuss various limitations of the current implementation and fruitful directions for future research.