A novel multisensory quality index of a food product: An analysis of a sausage properties

A novel multisensory quality index of a food product is proposed as the negative common logarithm of the probability of a joint event, when the product quality properties under control are characterized simultaneously by the specified set of categories of the assessor/expert responses. This index is related to the entropy (a measure of uncertainty) of the probability distribution of the responses. R-code for calculation of the index was developed and validated. As a case study, the index was evaluated for a sausage, produced according to the same standard by two different manufacturers in a two-year period. Each batch of sausage was tested at the manufactures' laboratories for its chemical composition and for five sensory properties: appearance, consistency, color, smell and taste. The sensory properties were examined by five experts employed by each manufacturer. A total of 650 expert responses to the quality properties of 26 sausage batches of one manufacturer, and 1350 responses to the properties of 54 batches of the such sausage of the second manufacturer were studied. Since sausage sensory data are influenced by the product's chemical composition, the hypotheses of homogeneity of the product composition (measured values of mass fractions of protein, fat, moisture and salt) between the two manufacturers were tested and rejected. Then, for each manufacturer separately, the hypothesis of homogeneity of the responses by the different experts on the same quality property was tested and not rejected. A correlation analysis of the responses to different properties was performed for the corresponding datasets. The data of the first manufacturer were assumed to be uncorrelated, while correlation between all pairs of quality properties of the second manufacturer was statistically significant. The influence of the correlation on the quality index values was assessed, and the sausage quality of the two manufacturers was compared.