2026/01/14 by Adam Chismar, Christopher T. Simons · 1 voice
Agricultural and Biological Sciences · Neuroscience · Nursing · #Biochemical Analysis and Sensing Techniques #Olfactory and Sensory Function Studies #Sensory Analysis and Statistical Methods
paper · doi:10.1016/j.foodqual.2026.105864
openalex publication_date 2026/01/14 · openalex created_date 2026/01/15 · openalex updated_date 2026/07/14
Allowing subjects to select a product from a consumer test as a “takeaway” has been suggested as a way to operationalize preference testing. We investigated the relationship between overall liking, purchase preference, and takeaway metrics for three flavors of fruit chips across three weeks. Three takeaway methods were used: “Only One,” in which panelists took only one sample cup in a flavor of their choice, “Fixed Number,” in which panelists took exactly three sample cups while deciding the flavor for each, and “Broad Choice,” in which panelists took anywhere from one to nine sample cups with a maximum of three of each flavor. Results suggest that the three methods perform similarly with respect to preference data, but for liking data the fixed number method had higher correlations. Additionally, it was found that normalizing the overall liking data via Z -scores produced higher correlations for both the fixed number and broad choice methods. Changes in liking scores and preference rankings remained consistent between weeks. Together the data suggest that correlation between stated and operational preferences were maximized under the “Fixed Number” condition once liking data had been normalized via a Z -score approach. These improvements in correspondence between stated preferences and actual behavior may enable development of novel models that more accurately translate traditional liking benchmarks into real-world purchasing behavior, preventing well-liked but not frequently purchased products from reaching the shelf and reducing resource loss in the process. • Different versions of takeaway preference questions produce different behaviors. • Three different takeaway methods performed equally well for preference rankings. • A fixed number takeaway method performed best for overall liking correlations. • Takeaway and overall liking correlations improved when liking data was normalized. • Changes in liking and preference between weeks were consistent.