Manuskripte aus den Instituten für Betriebswirtschaftslehre der Universität Kiel No. 409
Basic confounding patterns for full-profile conjoint analyses based upon ranking are examined. It is shown that commonly used orthogonal main-effect designs can lead to biased part-worth estimates, especially to an underestimation of less important variables. An alternative design procedure is developed to overcome this flaw. The model is tested by means of simulation analyses and is applied to a marketing research study. Some guidelines for applications are provided.