| ||Format||Pages||Price|| |
|PDF (296K)||18||$25||  ADD TO CART|
|Complete Source PDF (1.8M)||95||$55||  ADD TO CART|
Statistical modeling was applied to determine the influence of special consumer groups as a small subset (10 to 30%) of respondents on the outcomes of consumer tests. Tables are displayed for standard deviations ranging from 1.0 to 2.0 where a special consumer group displays a differing hedonic response pattern from the general consumer population. Without segmenting by the special consumer group in the design and analysis, situations are identified (i.e., number of respondents, type I error, standard deviations and proportion of special consumers) where a crossover interaction greatly increases the chances for a type II error (i.e., missing true differences). When special consumers are a minority segment accepting one product over another, a small proportion of special consumers can greatly increase the power to conclude significance. Without segmenting in the design and analysis, minority acceptance can increase the chances for a type I error, i.e., falsely concluding that the general consumer population accepts one product over the other. As the total number of respondents increases, the smaller the proportion of minority acceptors needed to draw a conclusion with a type I error.
Testing strategies are presented to segment on the basis of special consumer groups to guard against the chances for type I or II errors. In addition, if a special consumer group responds differently from the general population, methods are presented to determine how the special consumer group differed. Strategies are discussed for randomized complete block and balanced incomplete block test designs.
consumer, acceptance, power, segmentation, analysis of variance, experimental design, error, modeling, interaction
Manager, Sensory and Statistical Services, Brown-Forman Corporation, Louisville, KY