Abstract:

The cost and timing of product testing depends on the size of the experiment. The power of a difference test depends on the number of judgments, a component of the experiment’s size. Power is the probability of correctly concluding that a difference exists given a significance level, sample size and specified difference. Efforts to improve the power of difference tests are important because they are rewarded with the development of more resource-efficient methods. Theoretical and applied research on difference testing has shown that some methods may require as much as 100 times the sampling of other methods to detect the same difference with the same power. Some recent experimental work in taste research has shown that retasting during an experiment may improve test power. In order to exploit this finding, it is useful to understand why this might occur. In this report, we develop a possible explanation of this effect and from this model we can predict a broad range of experimental outcomes.