Classical multivariate analysis is based on the assumption that the data come from a multivariate normal distribution. The tests of multinormality have therefore received very much attention. Several tests for assessing multinormality, among them Mardia’s popular multivariate skewness and kurtosis statistics, are based on standardized third and fourth moments. In Mardia’s construction of the affine invariant test statistics, the data vectors are first standardized using the sample mean vector and the sample covariance matrix. In this paper we investigate whether, in the test construction, it is advantageous to replace the regular sample mean vector and sample covariance matrix by their affine equivariant robust competitors. Limiting distributions of the standardized third and fourth moments and the resulting test statistics are derived under the null hypothesis and are shown to be strongly dependent on the choice of the location vector and scatter matrix estimate. Finally, the effects of the modification on the limiting and finite-sample efficiencies are illustrated by simple examples in the case of testing for the bivariate normality. In the cases studied, the modifications seem to increase the power of the tests.
[1]
C. Croux,et al.
Principal Component Analysis Based on Robust Estimators of the Covariance or Correlation Matrix: Influence Functions and Efficiencies
,
2000
.
[2]
H. Oja,et al.
A scatter matrix estimate based on the zonotope
,
2003
.
[3]
K. Mardia.
9 Tests of unvariate and multivariate normality
,
1980
.
[4]
H. Oja.
Descriptive Statistics for Multivariate Distributions
,
1983
.
[5]
Hannu Oja,et al.
Multivariate spatial sign and rank methods
,
1995
.
[6]
K. Mardia.
Measures of multivariate skewness and kurtosis with applications
,
1970
.
[7]
Anil K. Bera,et al.
Tests for multivariate normality with Pearson alternatives
,
1983
.
[8]
James A. Koziol,et al.
Probability plots for assessing multivariate normality
,
1993
.
[9]
Hannu Oja,et al.
The affine equivariant sign covariance matrix: asymptotic behavior and efficiencies
,
2003
.
[10]
Christophe Croux,et al.
Testing the information matrix equality with robust estimators
,
2006
.
[11]
H. Oja,et al.
Sign and rank covariance matrices
,
2000
.
[12]
Werner A. Stahel,et al.
Robust Statistics: The Approach Based on Influence Functions
,
1987
.