A special study consisting of a survey among 326 housewives, has shown that the friendliness effect in product tests is minimised if the method below is used: The interviewer shows three packs with the same contents, but with different code numbers/letters. The housewife is given the suggestion that two of the packs contain a now product and the third pack is the product she is always using, without identifying the packs. The housewife herself selects the pack she will test, presumably under the impression that she has a chance of one third to get the product she is always using. A control group, where all packs contain the product she is always using, interviewed with the same questionnaire, provides the level of the bias which always remains. This level is the zero level for the results of the product test for the new product. With this system the friendliness-bias can be reduced considerably. In this study the reduction of the bias was about 50%.