ON THE RELIABILITY OF MAIL QUESTIONNAIRES IN PRODUCT TESTS.

Evidently the mail survey, including 65 per cent of our sample, provided results that were representative of the entire sample. True, the proportion of response is high for a mail survey, but a high return is to be expected in a situation where interest in the subject matter has been demonstrated by acceptance of participation in the test. Time pressures frequently call for relatively short test periods. In testing of lipsticks and with the type of design used here, the one-week test period seems adequate. Since the test product is used daily and frequently within each day, the formulation of a reliable opinion is to be expected in a relatively short time--in the case of lipsticks, probably less than one week. Frequency of use of a product is, of course, an important factor in the determination of the length of a test period. Basic to these findings is the inclusion in the sample of only those most likely to have stable opinions on this product. Had the sample included a sizable number of women who had just changed lipstick brands or otherwise lacked stable opinions, it is conceivable that the mail-interview and personal-interview replies may have differed markedly. [ABSTRACT FROM AUTHOR]