Premium
TWO MODELS FOR ESTIMATING THE DISCRIMINABILITY OF FOODS AND BEVERAGES
Author(s) -
HAUTUS MICHAEL J.,
IRWIN R. JOHN
Publication year - 1995
Publication title -
journal of sensory studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 53
eISSN - 1745-459X
pISSN - 0887-8250
DOI - 10.1111/j.1745-459x.1995.tb00014.x
Subject(s) - statistics , mathematics , variance (accounting) , observer (physics) , task (project management) , sensitivity (control systems) , econometrics , pattern recognition (psychology) , computer science , artificial intelligence , physics , accounting , management , quantum mechanics , economics , business , electronic engineering , engineering
In order to evaluate the suitability of signal detection theory methods for assessing the discriminability of foods and beverages, the discriminability of two dairy milk products that differed in fat content was measured with two detection‐theoretic methods: the single‐interval rating method, and the same‐different method. The nominal fat contents of the milk products were 0.1 and 1.6%. Measures of discriminability for three observers were derived by fitting receiver operating characteristics (ROCs) based on equal‐variance normal models to the ratings of each observer with a procedure that combined jackknifing and maximum‐likelihood estimation. The fitted ROCs provided a good fit to the data indicating that the equal‐variance models were appropriate for these tasks. The best‐fitting estimates of d′ obtained for each task were not significantly different, demonstrating that d′ is a measure of sensitivity that is largely independent of the task from which it is determined. However, estimates of proportion correct obtained for each task were shown to be significantly different.