Application of Two-Way Ordinal Analysis of Variation for Comparison of Results of Examination of Organoleptic (Sensory) Properties of Drinking Water, Sausage or other Objects

Tamar Gadrich, Industrial Engineering and Management Department, Braude College of Engineering, Karmiel, Israel (tamarg@braude.ac.il)

Application of newly developed two-way ordinal analysis of variation (ORDANOVA) for comparison of results of examination of organoleptic properties of different objects is discussed. Sensory (organoleptic) responses to object quality properties are ordinal (categorical) values, classified by experts according to predetermined set of ordered exhaustive and disjoint categories/classes. Equal/unequal and greater/less than are the legitimate operations between any two ordinal variables, but addition and multiplication of ordinal data are not legitimate operations.

A case study of expert responses to intensity of odor and taste of drinking water, obtained in an interlaboratory comparison of 45 ecological laboratories, is one of examples of the application. The intensity was classified into the six categories from ‘imperceptible’ to ‘very strong’. A decomposition of the total variation of the ordinal data and simulation of the multinomial distribution of the data-relative frequencies in different categories allowed testing statistical significance of the difference between laboratories and other relevant parameters.

Another example is a study of sensory responses of three experienced experts to five quality properties of purchased samples of a sausage of 16 producers, available on the market. Each of the quality properties (appearance, consistency, color, taste, and smell) was classified into the five categories from ‘very bad’ to ‘excellent’. The results showed the statistical significance of the difference between the expert responses to properties of the samples of the 16 producers, and the insignificant difference between responses of the experts related to the same sample. 

---

Short Biography of Presenting Author

Tamar Gadrich is a Senior Lecturer at the Department of Industrial Engineering and Management, Braude College of Engineering in Karmiel, Israel. She holds a PhD in Statistics from the Technion – Israel Institute of Technology. Her current research interests are in applied probability and statistics.