A comparative appraisal of two equivalence tests for multiple standardized effects

Area of Science:

• Biostatistics
• Statistical Inference
• Clinical Trial Design

Background:

• Equivalence testing is preferred over difference-based methods for demonstrating treatment comparability.
• Extensions of equivalence tests for multiple groups are less explored than for two groups.

Purpose of the Study:

• To compare ANOVA F and studentized range tests for evaluating comparability of several standardized effects.
• To assess power and sample size requirements for these distinct equivalence testing approaches.

Main Methods:

• Detailed comparison of ANOVA F and studentized range tests for multiple group equivalence.
• Power and sample size evaluations based on standardized mean range and fixed standard deviation.
• Utilized SAS and R computer codes for critical values, p-values, power, and sample size calculations.

Main Results:

• Neither ANOVA F nor studentized range test is uniformly more powerful for equivalence testing.
• Studentized range test shows a clear advantage in sample size requirements for achieving desired power.
• This advantage is particularly notable when effect configurations approach the minimum difference for equivalence.

Conclusions:

• The studentized range test is more sample-size efficient for multi-group equivalence testing under specific conditions.
• Both methods are valuable, but the studentized range test offers practical benefits for study planning.
• Available computational codes facilitate the application and planning of equivalence studies.