A Rankit Analysis of Paired Comparisons for Measuring the Effect of Sprays on Flavor

Publisher Website
Google Scholar
Abstract
Paired comparisons were used to determine subject preferences for sauce made from apples sprayed in a 2 x 2 factorial experiment with one of two insecticides, lead arsenate or parathion, and one of two fungicides, thiram or sulphur. Six pairs of samples, representing all possible treatment combinations, were tasted on each of six test days by the same 25 subjects, each subject recording both his choice within each pair and the degree of his preference as suggested by Sheffe. Three complementary analyses were computed in terms of rankits, defined as the mean deviate for the ranked items in a sample from a normal population with mean zero and unit standard deviation. Two analyses were based upon the frequency of choices for one treatment in each pair: (1) with the replicate as the unit and the rankits for a series of 26 as the variate, and (2) with the taster as the unit and the rankits for a series of seven as the variate. A third analysis was based upon the degree of preference within each test pair, converted to rankits. In each analysis the preferences were satisfactorily additive, the order of tasting within each pair had no significant effect upon preference, and the results on six test days were mutually consistent. Sauce from apples sprayed with lead arsenate was preferred to parathion, and thiram to sulphur, the flavors associated with the insecticide and fungicide being independent of each other. Individual tasters differed significantly in their flavor preferences for the direct treatment contrast especially when the degree of preference was reported. An analysis of the same data by four alternative statistical methods gave consistent results.

This publication has 1 reference indexed in Scilit: