SIMULTANEOUS ESTIMATIONS OF MULTIPLE PRODUCT SIMILARITIES USING A NEW DISCRIMINATION PROTOCOL

This research investigated new paradigms that permit the simultaneous comparison of more than two samples in a discrimination study. Three successive experiments were conducted. All involved noncarbonated orange beverages. In experiment I, Torgerson's method of triads was found to be more discriminating than the multiple dual‐pair method and was used in the subsequent two experiments. In experiments II and III, subjects discriminated among stimuli using the Torgerson's method as well as traditional duo–trios. In experiment II, the univariate Thurstonian model with four distributions was found to provide a suitable fit of the data, and the d′ values obtained using the traditional duo–trio methodology were not found to be significantly different from those obtained with Torgerson's method. In experiment III, a multivariate, but not univariate, model provided a good fit of the data. Furthermore, d′ values from the Torgerson's method were not found to be significantly different from those obtained using the traditional duo–trio methodology.PRACTICAL APPLICATIONS This research supported the use of a Thurstonian model for Torgerson's method of triads and uncovered the usefulness of the method when comparing more than two samples using a discrimination methodology, which has applications in situations involving samples with inherent intra‐product variations. Flexibility in the models available also permits the estimation of the dimensionality of the differences among the stimuli involved, providing valuable information that can be obtained more efficiently than running multiple pair‐wise traditional discrimination trials.