Comparison of triangle and tetrad discrimination methodology in an applied manner

摘要

Choosing a discrimination test can involve numerous factors, one of which is statistical power. The tetrad test has been shown to possess statistical advantages over the more traditional triangle method. However, these statistical advantages may not be present when the effect size decreases by more than 1/3. The frequency of large decreases in effect size, as well as other measures of test performance has not been characterized. In this study, over thirty products were tested using both triangles and tetrads in order to compare the two methods. The products tested ranged from canned vegetables and fresh fruits to deli meats and baked goods. After testing, inconsistencies were found within and across product categories. Only 16% of the tests were found to end with a different conclusion regarding a statistical difference (p & 0.05). In six of the experiments, the triangle test showed a significantly higher effect size than the tetrad. In eight of the experiments, the reduction in effect size for the tetrad led to no power advantage of the tetrad test. Participants also noted that the product being tested affected their impression of test difficulty in multiple experiments. This study creates a functional comparison of tetrad and triangle testing and quantifies the frequency in which the tetrad method effect size decreases by more than 1/3, leading to decreased statistical power.