The authors propose and illustrate an extension of the method of Hierarchical Information Integration (HII). HII allows one to handle large numbers of attributes in conjoint tasks by designing subexperiments that include subsets of attributes. It assumes that individuals can use general attributes or decision constructs to summarize their impressions of these subsets, which could be clusters of detailed, managerially relevant attributes. The proposed extension involves the design of sub-experiments that include attributes plus summary evaluations of remaining constructs. Advantages are that subexperiments can be analyzed separately but also jointly to estimate one overall preference or choice model; a more flexible and easy task is obtained; and one can test the assumed hierarchical decision structure. The authors illustrate the approach with an application that models consumer choice of shopping center. In this application, results partially support the hierarchical structure and predictive validity. Finally, the authors discuss implications for further research.

An analysis proposes and illustrates an extension of the method of Hierarchical Information Integration (HII). HII allows one to handle large numbers of attributes in conjoint tasks by designing subexperiments that include subsets of attributes. It assumes that individuals can use general attributes or decision constructs to summarize their impressions of these subsets, which could be clusters of detailed, managerially relevant attributes. The proposed extension involves the design of subexperiments that include attributes plus summary evaluations of reamining constructs. Advantages are that subexperiments can be analyzed separately but also jointly to estimate one overall preference or choice models, a more flexible and easy task is obtained, and one can test the assumed hierarchical decision structure. The analysis illustrates the approach with an appplication that models consumer choice of shopping center. In this application, results partially support the hierarchical structure and predictive validity. Finally, implications for further research are discussed.