Adherence and concordance among Q‐Graders in the sensory analysis of coffees

Abstract

AbstractIn any sensory analysis, it is important to investigate whether the tasters agree with each other, because they can assign scores in a correlated, but not the concordant way. This work investigated the adherence and the concordance among tasters of the coffee sensorial analysis of the Arabica and Conilon varieties, and their impacts on the final quality ratings of these varieties. The samples were obtained from regions of the Espirito Santo state, Brazil, and were evaluated by six tasters, with five repetitions and four treatments, totaling 120 observations. Nonparametric statistical techniques were used to compare the distribution of grades, and Lin's concordance coefficient was used to analyze the concordance among tasters. The Factor Analysis by Principal Components revealed that different distributions and disagreements among tasters affect, mainly, the sensorial attributes body, overall, flavor and acidity, thus changing the final result of the evaluation and, consequently, the final quality ratings of the products. These results allow new approaches in conducting the coffee sensorial analysis.Practical applicationsThe production of specialty coffees brings an excellent opportunity to add value, especially for small producers, or family producers on properties where the relief and climate make harvesting difficult, such as the mountain regions of the state of Espírito Santo, Brazil. This new perspective allows small producers to participate in contests to promote their product, which can be sold at different prices in the national and international markets. However, if the quality assessment methods suffer some kind of interference, the final evaluation may be unfair, or at least, incompatible with the reality of each product. The objective of this study is not to disqualify the current evaluation methods, but to justify improvements in these methods, which must be appreciated and improved by evaluators in the area.