ABSTRACT
One of the principal functions of industrial food analysis is to reduce the risk of encountering a negative market judgment. Consider the problem of comparing two similar products in order to determine which is preferred by taste. Problems of this kind abound in industrial practice, when one ingredient is replaced by another, its concentration is changed, or a manufacturing process is altered. Among the more discriminating assessors, some will truly prefer one product and some will prefer to the other: de gustibus non disputandum. The risk of idiosyncratic decision increases as the number of assessors decreases and attains a maximum with only one person. An alternative approach would be to screen the population of potential assessors, and to disqualify those less sensitive from taking part in the preference-decision process. In theory, a finite difference exists between any two products, and can be detected only if a large enough population of assessors is used.
