CONSUMER PRODUCT TESTING 767 features. Consumers, like people, emphasize the negatives. One solution is to give the consumer panel two products, both unfinished but having different negatives and positives. Whenever two products are submitted to consumers, they will form the parameters of a comparison test. If one is a good product and the other only a fair product, the comparison will tend to emphasize the extremes in this instance, the consumers would regard the good product as excellent and the fair one as poor. In any paired comparison both products should be of the same general quality. It would be difficult to test two shampoos with the same consumer panel. An acceptable solution to this problem is to use two similar panels, one for each product. The panels should be alike in all per- tinent variables: the same numbers of users and nonusers of cream rinses, for example the same age distribution patterns the same numbers of home permanent users the same numbers of nonusers and the like. When the returns are in, the balance of the two panels should be confirmed, and their findings can be compared. Multiple product testing with the same panel can be done only if sufficient time elapses between the use of each product. The time lapse prevents carryover from one product to the next. Second, the time lapse must not exceed the user's ability to remember and compare the various products. Multiple product testing is one of the most misused techniques in food product testing for several reasons: the test designers seem to feel that by alternating the position of use of each product with succeeding consumers, they have canceled out product interactions and carryovers. Such is not the case. They also seem to feel that consumers remember each product's properties equally well. This is not the case either. Consumers do a paired comparison of the first two products, sort out the major differences, and then assess the succeeding products according to the major differences they re- membered from the first two. The test designers frequently make an- other mistake by forcing consumers to respond to three or more prod- ucts tasted at one session, headcounting these preference rankings and misapplying statistical analyses, coming up with probability figures that are accepted as true. The point here is that statistics should not be applied to data from improperly conducted tests. Fortunately, most cosmetic products require repetitive use before they can be truly evaluated by consumers, and multiple cosmetic product testing is therefore contraindicated.

768 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS With random panels of consumers who run the gamut in intelligence, there is always some danger of introducing unwanted biases induced by sample codes. A panel of intelligent consumers obviates the need to be deeply concerned about findings based on a like or dislike of the code used for an otherwise unidentified sample. But from common sense codes should be avoided that might take on meaning, e.g., the letter X, the letter-number code A-l, the butter-score numbers 88 or 93 for margarine, or G-11 for a soap. Most test designers avoid color codes and simple letter or number codes they usually use two- or three- digit number-letter combinations. Use Period In food testing if only a simple "yes/no," "go/no-go" kind of result is needed, a product may be subjected to "one-shot" testing. This type of test is usually conducted in trafficked areas, such as a store, county fair, or bus terminal. This test resembles the man-in-the-street opinions poll. Persons who happen to pass by and who have the time and inclina- tion are invited to participate. Often they are asked to choose between two samples on a preference basis. Considerable numbers of consumers can be reached by such a test method, but it should be remembered that the data relate only to first impressions. Conceivably the aromas of two perfumes could be checked out in this way, if the bias inherent in such a consumer test population were recognized. But the actual properties of the perfumes and the con- sumer's changes in attitude could be learned only through a repetitive use test. For most products there is a first impression, a get-acquainted period, and a final impression. The final impression leads to final assess- ment, which in the marketplace would determine second purchase. A use test, then, should be conducted for a long enough period to attain a final impression. This might be as short as three days for an after-shave lotion or as long as four weeks for a scalp conditioner. It is difficult to sustain a panelist's interest for more than two weeks unless re-stimulation is offered. Recently the authors were able to hold a home-use consumer panel of 200 persons together for three months by check postals, fresh samples, and friendly letters. The check postals really served a dual purpose. They activated the panelists and provided information on changes in attitude and frequency of product use so that one could be reasonably certain of obtaining final impressions at the con- clusion of the test.