Journal of Statistics: Advances in Theory and ApplicationsAuthor's: Rie Enomoto
Pages: [37] - [54]
Received Date: December 10, 2012, Revised December 19, 2012
Submitted by:
Koizumi et al. [5] proposed the omnibus test statistics using the sample skewness and kurtosis defined by Mardia [7] and Srivastava [16]. However, none of these statistics account for the covariance between skewness and kurtosis, which is not negligible for small sample size. In this paper, we consider the multivariate normality tests based on the sample measures of multivariate skewness and kurtosis defined by Srivastava [16]. We propose some new test statistics, which are taken into consideration the covariance between skewness and kurtosis. In order to evaluate accuracy of proposed test statistics, the numerical results by Monte Carlo simulation for some selected values of parameters are presented.
multivariate normality, omnibus test, sample skewness, sample kurtosis.
