Presented at the 2013 Pangborn Symposium in Rio de Janeiro, Brazil.
Sensory professionals continue to face the challenge of quantifying and accounting for differences in individual panelist performance within their difference testing programs. In this presentation we discuss how recent developments in Thurstonian modeling can be used to track panelist performance over a series of (potentially non-replicated) difference tests within the same product category. In particular, we show how Thurstonian modeling can assign individual sensitivity parameters to each panelist and can update estimates of these parameters as difference tests are conducted over time. This approach both monitors differences in panel members and provides more precise estimates of product differences. The ideas in this presentation build on recent advances in the use of Thurstonian modeling to either analyze results from several experiments or to model data from replicated testing, and constitute an additional practical contribution of Thurstonian modeling to industry.
Colleagues can download the poster here:
Not a Colleague? Click here to join for free!