Transitioning from Proportion of Discriminators to Thurstonian d

Rousseau, B., Ennis, D. M., and Ennis, J. M. (2013). IFPress, 16(2) 3-4.



In a previous technical report we discussed a commonly used measure of sensory difference, the proportion of discriminators (Pd), and explained that this measure is method specific. We concluded that:

“The concept of the proportion of discriminators is intuitive and appealing. Unfortunately, it is also method-specific and one can achieve very different values using different methods, which reduces its appeal as a management criterion. The use of a method–independent index, such as the Thurstonian d′, can provide more stable and accurate information that may be valuable in a decision that involves the consideration of consumers’ ability to discriminate between alternate product formulations.”


Many companies have implemented sensory discrimination programs using Pd as a parameter for power calculations. Since this measure has shortcomings, there is a need to consider a unit that is method-independent such as the Thurstonian d. Fortunately, tools exist that permit the translation of a method-specific Pd into the corresponding value. In this report we discuss how these tools can be used in practice.


Figure 4. Relationship of to proportion correct and proportion discriminators for the triangle and tetrad tests; example of a Pd of 25%.