Consistency of a Trained Taste Panel

Abstract

A means of determining the reliability of a trained taste panel is to test their consistency of judgments on duplicate samples. Fourteen hundred and twentythree duplicate sets of' research milk samples were tested by 19 judges, for identification and measurement of intensity of off-flavor. The samples were scored on a 10-point scale ranging from 31 to 40. Each of the 17 flavor criticisms is based on a 6-point range. Analysis of data indicates an acceptable degree of consistency in panel performance, even though there is considerable variation among judges. The average absolute differences determined on a monthly basis ranged from 0.10 to 0.39 for a mean of 0.26 for the panel as a whole, whereas, for individual judges the range was 0.41 to 0.80 for a mean of 0.65.