1
Master's student in statistics, University of Isfahan
2
Assistant Professor, Department of Statistics, University of Isfahan
Abstract
Suppose we have m random samples of n independent multivariate normal distribution and we want to test the hypothesis that the i nth sample is an outlier. Such a test can be useful in the first phase of multivariate statistical process control studies with group observations. It is now known that there is a beta statistic for such a test and that the F statistic can also be used through the relationship between the beta distribution and the F distribution. In statistical literature, a clear and precise proof for the distribution of the test statistic is not found, and in some cases, the proof is incorrect or incomplete. In this article, we present a clear and relatively clear proof and examine the strengths and weaknesses of the test through simulation.
