HOW TO USE TESTS FOR UNIVARIATE NORMALITY TO ASSESS MULTIVARIATE NORMALITY

The assumption of multivariate normality (MVN) underlies many importan t techniques in multivariate analysis. In the past 50 years, over 50 t ests of this assumption have been proposed. However, for various reaso ns, practitioners are often reluctant to address the MVN issue. In thi s article, several techniques for assessing MVN based on well-known te sts for univariate normality are described and suggestions are offered for their practical application. The techniques are illustrated using two previously published sets of real-life data. In one of the exampl es it is shown that simply testing each of the marginal distributions for univariate normality can lead to a mistaken conclusion.