SOURCES OF ERROR AND THE NO‐PREFERENCE OPTION IN DAIRY PRODUCT TESTING

Abstract

ABSTRACT When products are physically identical, consumers will choose a no‐preference option in a paired preference test only about 30% of the time, giving a nonintuitive result. We evaluated the no‐preference false alarm rate in dairy products and the potential sources of error in panelist behavior, including discrimination problems and poor repeatability of judgments. Paired preference tests were conducted with milk (2% versus nonfat or skim), with cottage cheese (4% versus nonfat) and with identical samples. Triangle, dual standard and same/different discrimination tests were also conducted. The no‐preference option was chosen by about 30% of the consumers when samples were identical, and less frequently when samples were different. Discrimination performance, although statistically significant, was poorer than expected with about 30% discriminators (d′ of 1.7) in most tests. Even in the dual standard tests on milk (visually obvious and with references present), performance was only 88% correct (76% discriminators). Repeatability in preference tests was also poorer than expected, with only 56% choosing the same preference option for different samples on repeat trials although the group preferences were stable. Same/different tests indicated that over 50% of the consumers called identical samples different, although this fell short of explaining the 70% average false alarm rate in preferences. Even with apparently discriminable consumer products, accurate discrimination of the samples and repeatability of preferences should not be assumed. Random responding may contribute to error variance and produce nonsensical results such as the expression of choice between identical samples.