TWO MODELS FOR ESTIMATING THE DISCRIMINABILITY OF FOODS AND BEVERAGES

摘要: In order to evaluate the suitability of signal detection theoty methods for assessing discriminability foods and beverages, two dairy mik products that differed in fat content was measured with detection-theoretic methods: single-interval rating method, same-different method. The nominal contents milk were 0.1 1.6%. Measures three observers derived by fitting receiver operating characteristics (ROCs) based on equal-variance normal models ratings each observer a procedure combined jackknifing maximum-likelihood estimation. fitted ROCs provided good fit data indicating appropriate these tasks. best-fitting estimates d' obtained task not sign@cantly diyerent, demonstrating is measure sensitivity largely independent from which it determined. However, proportion correct shown be significantly different.