Poster: When are two products close enough to be equivalent?

Presented at the 2015 Pangborn Symposium in Gothenburg, Sweden.

When comparing a gold standard and an alternative formulation for a “match”, a difference will always be found given the sample size is large enough. The question is thus not whether products are different but whether they are “close enough”.

In order to answer and set up suitable specifications, a scientist needs to establish the size of the sensory difference below which the products can be deemed as equivalent. In this research, we investigated two approaches to set this threshold value called δR: a paired preference test (PP) and a same-different test with sureness judgments (SD). The SD allows the estimation of consumers’ psychological (τ) criterion representing the size of the sensory difference above which two products will be called “different”.

The investigation involved 256 consumers (126 M, 130 F, average 24.8 yrs.) The stimuli were organic apple and orange juices. Within each flavor, the original concentration was diluted to create two sets of small sensory differences. Each consumer performed one PP and one SD for each flavor. This resulted in 128 trials for each method and flavor/size of difference.

For the SD test d' values were significantly different among the product pairs, but the size of the τ were consistent across the flavors and differences. Based on these results, a δR value of 0.92 (Thurstonian δ) was found.

For the PP test, graphing the SD d' values against the proportion of preferences, a value similar to that of the SD was found (δR = 0.82) for a preference threshold of 55/45.

These consistent results indicate that when consumers start calling samples “different” they also seem to start preferring one sample over the other at the 55/45 preference ratio level. This value can then be used as δR and assist a scientist to set consistent and suitable discrimination program specifications.

Colleagues can download the poster here:

When are two products close enough to be equivalent?

Not a Colleague? Click here to join for free!