In this paper we investigate performances of the test of multinormality introduced by Malkovich and Afifi. An approximation formula of the power of the test against elliptically symmetric distributions is derived. Examples which illustrate the present results are also discussed.

In this paper, first we make a maximal extension of the well known gauss-Markov Theorem (GMT) in its linear framework. In particular, the maximal class of distributions of error term for which teh GMT holds is derived.Second, we establish a nonlinear version of the maximal GMT and describe some...