ABSTRACT
The existing literature on k-sample equivalence testing procedures is comparatively sparse. A reason for this fact might be that, at least for parametric settings, there is a straightforward yet often overconservative solution to problems of establishing equivalence of k distributions as soon as a test for the corresponding two-sample problem is available. The rationale behind this approach going back to Berger (1982) is a result which can appropriately be termed intersection-union principle because of its straight duality with Roy’s (1953) well-known union-intersection approach to multiple testing and interval estimation. The scope of potential applications of the principle is fairly wide and goes far beyond the specific multisample settings discussed in the subsequent sections of this chapter.
