Tarver, Mae-Goodwin (1987, ASQC) Quest Associates, Ltd., Park Forest, IL
Man continuously receives "messages" from his environment through natural communication channels such as the visual, auditory, tactile, gustatory and olfactory nerves. Consumers judge the subjective quality of products using these "messages". For example, the exterior finish of refrigerators, stoves, washers, and automobiles are examined by consumers for visual defects, for uniformity of color, and for gloss. Fabrics are examined for "drape", softness, color, and luster. Food products and beverages are appraised for flavor, odor, clarity (beverages) and even for characteristic sounds ("fizz" of carbonated beverages, "crunch" of crackers, etc.) Adequate evaluation of subjective quality depends upon human ability to detect small as well as large differences in product quality. In other words, perceived quality of subjective characteristics depend upon the amount of information which can be transmitted through "noisy" human sensory channels.A method is described in this paper for selecting consumer panelists for appraising subjective quality characteristics. This method quantifies a panelist's inherent potential for sensory discrimination and is based upon 1) the signal detection concept of psychophysics and 2) the fact that the human sensory system is analogous to Shannon's "noisy" communication channel. The major problem in selecting panel members is to discover which persons have this inherent ability to detect sensory signals in the presence of noise and to find those whose signal recognition power can be strengthened by training.Selecting flavor panel members by a Two-Sample Different Test is used in this paper to illustrate these concepts for the following reasons: 1) Flavor is a familiar quality characteristics to virtually all consumers, 2) the test method is relatively simple and the analysis of the resulting data can be computerized, and 3) it is possible to calculate in advance the values of the information category boundaries, provided the number tests to be run are fixed in advance. It is also possible to attach a monetary value to each panelist's results i the cost per test is known in advance.