This article investigates the efficiency of repeated measures designs versus completely randomized designs in terms of both ad hoc multiple comparisons and a priori pairwise multiple comparisons. Criteria are derived for determining when a repeated measures design is preferable to a completely rancomized design for a one-way classification problem. Numerical examples are given.
- correlation matrix
- simultaneous confidence intervals
- uniform efficency
ASJC Scopus subject areas
- Statistics and Probability