Bot. Bull. Acad. Sin. (1995) 36: 121-133

Wu and Chen Sensory analysis of quality control

Sensory analysis in quality controlthe agreement among raters

Hong-Pang Wu1,3 and Lee-Shen Chen2

1Institute of Botany, Academia Sinica, Taipei, Taiwan 115, Republic of China

2Department of Statistics, Ming Chuan College, Taipei, Taiwan, Republic of China

(Received September 18, 1993; Accepted September 9, 1994)

Abstract. This paper considers cluster analysis, kappa methods, and log-linear association models, which are often used to study agreement in medical and psychology research but are rarely used in sensory evaluation, to evaluate the agreement of tea sensory data among panelists. We recorded sensory data six times during the period from 1989 to 1990. There were 8 to 10 well-trained panel members, giving 11 ordinal categories for each tea sample. We then combined the 11 categories into 5 and applied the three methods to measure the agreement among panelists, and we found that agreement among the panelists was not high enough, regardless of the proposed methods.

Keywords: Cluster analysis; Kappa coefficient (k); Log-linear association model; Panel; Sensory control.

3Corresponding authors.