Consumer sensory perception of cheese depends on context: A study using comment analysis and linear mixed models

This study examines the effects of information and personal context on consumers' hedonic liking and sensory description of Vermont artisan cheeses. Four representative Vermont cheeses (unripened goat's cheese, bloomy-rind, cheddar, and blue) were presented to 182 different consumers in 2 different information conditions: socially and contextually relevant production information (informed) or generic information about the cheese category from dairy science manuals (blind). Consumers reported their liking using a hedonic line scale and, subsequently, provided free-text comments describing what they liked or disliked about the cheeses. Consumers also reported their familiarity with each cheese and completed the food involvement scale (FIS). Hedonic liking was modeled as a function of cheese type, information condition, familiarity, and FIS score using linear mixed-effect modeling; positive and negative consumer comments were tabulated using textual analysis and descriptive maps were generated using correspondence analysis (CA). The linear mixed-effect models demonstrated that consumer liking was significantly affected by cheese type, information, familiarity, and the consumer's FIS score. Analysis of consumer comments by CA produced empirically sensible, consumer-derived sensory profiles of each cheese; in addition, stable and significant sensory differences were found between sensory descriptions in the informed and blind conditions. Thus, this research demonstrates that information and extrinsic consumer attributes are important for understanding both consumer liking and consumer sensory perception. In addition, textual analysis by CA of free-text consumer comments is presented as an expeditious method for producing sensory profiles, and the advantages of this method are discussed. (C) 2013 Elsevier Ltd. All rights reserved.