Presented at the 2013 Pangborn Symposium in Rio de Janeiro, Brazil.
Discrimination testing is most often used in situations where a product’s sensory properties must be matched by a reformulated version responding for instance to cost reduction initiatives or a new governmental regulations. The decision of accepting or rejecting an alternative is typically made based on the significance level of the results. However, if no significant difference is found, does it mean that consumers indeed cannot discriminate between the samples, i.e., that such change would not yield a significant preference test outcome?
In this research we investigated a link between discrimination test results and consumer preferences. The study involved two discrimination protocols, the triangle and tetrad tests, and a preference test between the same samples. The stimuli were two sets of apple and orange juices, the two samples within a set differing in their concentrations.
In study 1, 456 consumers compared each set of samples under two conditions: with and without resampling allowed. In each of the four conditions, the tetrad test yielded a higher proportion of correct answers than the triangle test, confirming Thurstonian predictions.
Of particular interest was the smallest measured difference (apple juice, no resampling, d'=0.4), which did not produce a significant result with either protocol. The corresponding calculated proportion of discriminators was 5%. In study 2, 208 consumers compared the same stimuli in a preference test. Interestingly the condition for which no significant difference was previously found resulted in a highly significant preference (71/29, p<0.001).
While this could first appear like a discrepancy, we show how it is not the case and that the findings can be explained with the proper modeling approach. We conclude by emphasizing the critical need to establish consumer relevance in any discrimination testing program to ensure the suitability of the decision making process.
Colleagues can download the full presentation here:
Why establishing the consumer relevance of a sensory difference is critical
Not a Colleague? Click here to join for free!