A. L. ROBINSON1, E. Chambers, IV1, and G. Milliken2. (1) Sensory Analysis Center, Kansas State Univ., Dept. of Human Nutrition, 124 Justin Hall, Manhattan, KS 66506, (2) Dept. of Statistics, Kansas State Univ., 102D Dickens Hall, Manhattan, KS 66506
Just-About-Right (JAR) scales are commonly used to assess consumer expectations of a product attribute to provide directional information to product developers. Studies have shown that JAR scales did not always provide an appropriate estimation of liking. Disparity between the scales was found. Thus, the results could be misleading to the development process. Research was conducted to compare optimal amounts of specific sensory attributes using two scales: a JAR and a hedonic scale. Four studies were conducted. Three tests explored a particular type of expected bias: (1) a health bias (butter in mashed potatoes), (2) positive attribute bias (cocoa in chocolate cake), and (3) negative attribute bias (sourness in salad dressing). No bias was expected in the fourth test (pizza sauce on pizza), which was used as a control. Consumers (100+ for each study) were asked to indicate their impression of the strength of attributes (that is, butteriness, chocolate flavor, sourness, sauce amount) using a 7-point JAR scale (descriptors=too much, just about right, and not enough). Consumers also evaluated the degree of liking of the same attribute using a 7-point hedonic scale (like extremely to dislike extremely). In each study, a quadratic regression model was used with individual consumer scores to predict optimal values of each attribute with each scale. Individual predictions from each scale type were averaged to find the predicted overall optimum level of the attribute for the consumers. Results demonstrated that the amounts predicted by the JAR scale were statistically significantly different (P < 0.05) than those predicted by the hedonic scale. Additionally, for the three tests in which bias was expected, the results indicated meaningful differences larger than 10%. Thus, the existence of bias is evident and additional research is needed to determine the extent to which the bias can be attributed to specific effects.
Session 99F, Sensory Evaluation: General