Tukey's Method of Multiple Comparison in the Randomized Blocks Model

Abstract

In Kempthorne's randomization model for the randomized blocks design, the errors are correlated and are not assumed to have common variance. Under simple conditions, it is shown in this article that the exact level of confidence of Tukey's method of multiple comparison approaches the nominal level of confidence as the number of blocks increases. Monte Carlo results that illustrate this convergence are also presented.

Key Words:

• Tukey's method of multiple comparisons
• Multiple comparisons
• Randomized blocks design
• Randomization model