Computer-intensive methods for sensory data analysis, exemplified by Durbin’s rank test

Durbin’s rank test, a Friedman-type test in balanced incomplete block designs (BIBDs), has wide applications in sensory analysis. The chi-square approximation, which is commonly used for Durbin’s rank test, is not adequate for small BIBDs. This paper discusses some computer-intensive methods including Monte Carlo test and permutation (randomization) test for Durbin’s rank test. S-Plus programs for the procedures are developed and are available upon request.