24 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS or appearance or the substitution of cheaper, or more stable, or more easily obtained materials in the formulation. Mrs. Ludford (1) has already outlined some of the problems involved in product testing, and I have no wish to repeat them. I feel it is important however to stress that any product which reaches the market should have already been subjected to, and emerged successfully from, prior testing by at least two and frequently all three of the following alternatives which I quote from a private document (3): a Expert panels, which can assess the magnitude of the effect of changes in formulation but only rarely their acceptability b postal panels, which can assess the relative preference between two products but only on rare occasions and with special test designs, the motivation behind the preference c consumer tests involving personal interviews, which can assess the relative preference of two products and can often in addition establish the motives leading to the preference and relate it to the market conditions. The standard technique at the consumer stage is the double placement blind test, the modified product being pitted against the existing version. It may well be tested against competition as well if the reformulation was designed to match or gain ascendancy over it. Whilst the technique is reasonably satisfactory if correctly administered by fully trained interviewers who are able to probe respondents' views without biassing them in any way, it is extremely important that the sample of respondents is carefully defined. Clearly whatever sample is chosen it should wherever possible be representative of its universe. The problem lies in defining the universe. For example, in a test involving two hair sprays, one's own brand and the leading competitor, and designed to establish whether an improved product has better holding properties than the competition, should one test among representative samples of a. all adult women, or b. women between 16 and 50, or c. women with modern hairstyles, or d. users of the two brands only, or e. current hair spray users only (and how to define 'current'?), or f. some combination of two or more of these? No universal rule can be laid down the decision must be made ad hoc in the light of knowledge of the market and the specific objectives against which the modified product was formulated. It is perhaps worth stressing

PRODUCT TESTING IN THE SPHERE OF PRODUCT DEVELOPMENT 25 that the test results should be interpreted with the sample employed very much in mind if misleading conclusions are to be avoided. It often happens in double placement blind product testing that the test product is not preferred either overall or in any of the sub-samples whether these be of respondents in particular demographic categories or users of particular types or brands of the product in question. Should the test product be rejected, or reformulated and re-tested, in such circum- stances? There can be no all--embracing answer to this question either. Action should be related to the objectives of the product test, but it may be worth considering the possibility of a. repeating the test with the same sample of respondents to ascertain the stability of their preferences, to isolate the discriminators among them and to measure intensity of preference in an attempt to throw further light on the meaning of the test results b. re-interviewing the minority who preferred the test product to probe more deeply the reasons for their choice and the strength of their opinions about it c. carrying out an extended usage test to ascertain whether the test product gains in popularity with greater familiarity. Whether or not any of these additional steps are taken must be decided in the light of the original objectives. Further action will only be taken generally speaking when it is thought that the test product might have a minority appeal which could prove profitable. One of the aims of this second stage research would then be to delineate this minority in meaningful terms, whether they be the more usual demographic or usership categories or psychological or behavioural traits. It follows that such action would not normally be taken if the test product were designed to compete in the same market as the brand leader against which it was tested, or if the double placement test results indicated a product deficiency on any essential characteristic. I should add here that in the present state of development of consumer research techniques no valid yet practical and simple means of assigning people to useful psychological categories, e.g. extroverts and introverts, has yet been found. Such methods as exist tend to be far too cumbrous for general use. What is needed is a means of achieving the desired ends merely by asking three or four questions. A great deal of work has already been done in this area and there are distinct possibilities of success.