Replicated Difference and Preference Testing with Applications to Claims Support
Bi, J. and Ennis, D. M. (1998). IFPress, 1(2) 2-3.
As part of an effort to compare a new deodorant with a competitor’s product, 10 experienced judges evaluate the two products on the left and right arms of 30 subjects. In a counterbalanced design in which the two products are alternately placed on the left or right arms within a subject, each judge reports the least malodorous arm.In this example, there is the possibility that there may be trial-to-trial differences in malodor. The chemical reactions of subjects to the deodorants may differ so that one product may be less effective on some subjects than others. This type of result leads to the need to account for inter-trial variability either because of the need to provide defensible claims in the face of inter-trial variation or because this type of variation is of fundamental interest itself, as may occur when one is identifying preference segments.
Figure 1. Psychometric functions for the 2-AFC and duo-trio.