THE ROLE OF CONSUMER STUDIES IN RESEARCH 51 on a questionnaire. I advise that questionnaires should be kept as short as possible in order to give the required information. DR. I. SAMES: In page 008 under "questionnaires" you sub-divide a question into seven categories of likes and dislikes. If one were to make that into nine and have the prefix "I think it is fabulous", and have the suffix "I think it is lousy" you have covered the whole gambit of human expression. I wonder whether making the question into a type of census form renders question fatigue more likely and question objection more probable. Would you not agree that so many sub-divisions are un- desirable? TI•E LECTURER: The example cited is only one question, which indicates the kind of answer required. It looks long but only involves the panelist in reading through the statements to see which comes nearest to his or her opinion, and ticking DR. K. SAMos: I think you have to be extremely "intelligent, literate and articulate" to differentiate between disliking a thing moderately and disliking it slightly. What are your views on the optimum duration of the test before the panel are invited to complete their questionnaires? Are you prepared to wait for your panel to use all the product to form their opinion, or do you wish them to try it for a period of one week, two weeks, or longer, and then to complete the questionnaires? It is my personal experience, which I understand is not specific, that if a product is used only for part of its life you can get a falsely good impression the defects would have appeared had you gone on using it. T• L•CTUR•R: If yOU want initial reaction, a short test which just gives panelists the opportunity to try the product is satisfactory a longer test is necessary if you require reactions when panelists get to the end of the pack or after a long period of use. In order to administer the test it is easier to set a time limit rather than let each individual continue until they have finished the pack, because of the different rates at which people use products. However, if should be possible to ensure that the test goes on long enough for most people to use up all the product. MR. N.J. V• ABBg: In a Round Robin design, one sometimes obtains the finding that A is better than B, B is better than C, but C is better than A. How would you interpret such a result? T• L•CTUR•R: Such apparent anomalies can sometimes be explained by a chang- ing standard of judgement, which may be shown up in comments on the question- naire. It might be necessary to interview some of the panelists to try and get more information, and such checking can lead to useful new ideas. DR. T. J. ELLIOTT: Do you have any knowledge of the variation in results depend- ing on how one carries out the paired comparison test. I gather that if one wanted to evaluate a shampoo, the best way of doing it is to use shampoo A on one half of the head and shampoo B on the other, under the control of a professional hairdresser. If you are giving the panel two products to evaluate would you prefer if they used the two products simultaneously or would you prefer to let them use one product for a week, and the second product for the consequent week? How infallible is human memory? If one uses a product for a week and at the end of the week changes to another product, the one week of formative experience on product A is probably going to influence the reaction to product B, whereas if the two products are used simultaneously, a different result is obtained. Do you have any knowledge of this?

52 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS THE LECTURER: On a small panel in the laboratory it may be possible to test products 'one on each side'. Panelists would need to get used to this method of testing which is reasonably objective. For consumer opinions, which are subjective, it is necessary to have only one product assessed at a time so that it can be used in the usual way. DR. A. W. MIDDLETON: Reverting to your list in page 008 - what is the effect of the order in which these questions are asked? THE LECTURER: I suspect that panelists tick the first statement that they agree with. The order is, therefore, very important and placing favourable statements first on the questionnaire may lead to a more favourable rating for the product than if unfavourable statements were placed first. This is something that I would like to test. DR. K. H. HARPER: I gather that the Inethod of sequential analysis offers many advantages in a study where you are asking for a preference. Does this have any application in product testing of cosmetics? THE LECTURER: Sequential testing is very useful for small tests involving labora- tory personnel. However, for tests when products are sent outside to panelists it is usually simpler to estimate the expected degree of difference, and to calculate the panel size to give this difference the required significance. All the products can then be sent out at the same time and all the answers will come back at approximately the same time.