Thurstonian-derived models, covariates, and consumer relevance
Sensory discrimination test methods are widely used by industry to guide decision-making. Interpretation increasingly relies on Thurstonian-derived models, which use mathematics to encode psychological decision-making rules, and map method-dependent results onto a putative method-independent discriminable distance (d′). It is also possible to estimate the response bias, or tau (τ), in some test methods, such as the same-different test method. Rousseau (2015) exploited the ability to interrelate results across test methods to devise a strategy for estimating the size of a consumer-relevant discriminable distance within a particular product category. However, all published studies thus far are based on simulated data.
This talk has several aims. The first aim is to demystify Thurstonian-derived models. It will be addressed by showing the derivation of a psychometric function for illustrative purposes, then discussing some of the assumptions that underlie the model, and predictions that would be expected to be satisfied if the model holds true. The second aim is to review the opportunities afforded by analysis of sensory discrimination test data using generalized linear models with a psychometric link function, which can be used to estimate not only d′, but also the effects of potential covariates. The third aim is to review Rousseau’s strategy for estimating what discriminal distance has “consumer relevance”, informed by empirical data collected in August 2016 at Compusense Inc. (Guelph, Ontario, Canada). Specifically, the advantages, challenges, and limitations with implementing Rousseau’s strategy will be reviewed based of empirical data.
Rousseau, B. (2015). Sensory discrimination testing and consumer relevance. Food Quality and Preference, 43, 122-125.
Castura, J. C., King, S. K., & Findlay, C. J. (2017). Thurstonian-derived models, covariates, and consumer relevance. In: Applied Use of Discrimination Testing: Industry-Current Basic Practices, Trends, and Advancements (presented by Rothman, L., Castura, J. C., Kuesten, C., & Goldman, A.). Institute of Food Technologists Annual Meeting and Food Expo. 25-28 June. Las Vegas, NV, USA.