Bridging Designs for Conjoint Analysis: The Issue of Attribute Importance

Conjoint analysis studies involving many attributes and attribute levels often occur in practice. Because such studies can cause respondent fatigue and lack of cooperation, it is important to design data collection tasks that reduce those problems. Bridging designs, incorporating two or more task subsets with overlapping attributes, can presumably lower task difficulty in such cases. In this paper, we present results of a study examining the effects on predictive validity of bridging design decisions involving important or unimportant attributes as links (bridges) between card-sort tasks and the degree of balance and consistency in estimated attribute importance across tasks. We also propose a new symmetric procedure, Symbridge, to scale the bridged conjoint solutions.