Relating chemical/physical and sensory data in food acceptance studies

Abstract

To understand food acceptance requires the integration of objective descriptive sensory data with consumer acceptance information and chemical/physical data. The paper briefly discusses sensory and chemical/physical information in the context of food acceptance. It stresses the need that such information should be as meaningful as possible and provides an example of the potential benefits gained from applying generalised Procrustes analysis to rationalise the sensory data. Such treatment can provide a more reliable representation of the inter-relationships between products, on the basis of their sensory properties, independent of descriptive terminology used. A number of the mathematical techniques available for investigating how such data relate to the complex stimuli found in foods are also discussed. By viewing the data as a vector space of variables salient features are highlighted; attention is drawn to the limitations of the more conventional regression and correlation approaches and the concepts behind alternatives, such as partial least squares regression. Reasons are given as to why the latter approach would appear to be the more suitable for relating sensory and chemical/physical data. The methods are illustrated using an example taken from work on wine.