Comparison of preference clustering outcomes from replicated consumer tests—A case study with mandarins

Abstract

AbstractCommercial and experimental mandarins consisting of 15 varieties were evaluated through a combination of consumer testing, sensory descriptive analysis, and metabolomic profiling over the course of 2 years. The objectives of the research were to compare repeated consumer clustering solutions and to evaluate the potential market success of new mandarin cultivars. Consumers rated their liking of various sensory modalities, such as overall liking paired with Just‐About‐Right (JAR) scaling for sweetness, sourness, firmness, and juiciness and Check‐All‐That‐Apply attributes. Repeated preference clustering of the consumers across several tastings consistently revealed the presence of two clusters who differed in their liking of the sensory and chemical profiles of the fruit. The first cluster of consumers showed significant negative correlations with perceived sourness and concentration of citric acid and a left‐skewed distribution of JAR sourness ratings as compared with the second cluster of consumers. The experimental varieties of “DaisySL,” “KinnowLS,” “Shasta Gold,” and a commercially produced clementine sample had the highest overall liking rating of all 29 mandarins that were evaluated by the consumers. This work demonstrates that repeating clustering studies with a paired internal metric, such as a JAR question, may help to validate cluster solutions.Practical ApplicationsThese results show that conducting multiple consumer trials using different products may help to confirm hypotheses regarding cluster preferences for sensory and chemical profiles. Throughout this study, the drivers of liking identified for the preference clusters were consistent with their JAR ratings of select attributes (i.e., sourness). We suggest that clustering experiments should be conducted multiple times, especially when ephemeral products like fruit are being studied, to ensure that solutions are genuine. Additionally, an internal metric, such as a Just‐About‐Right question or a Check‐All‐That‐Apply question may help to differentiate the consumer clusters.