In multivariate analysis it is generally assumed that the observations are normally distributed. It was Mardia ([1] to [5]), who first introduced measures of multivariate skewness and kurtosis; these statistics are affine invariant and can be used for testing multivariate normality. Skewness and kurtosis tests remain among the most powerful, general and easy to implement. In this paper we show some properties of these statistics when population distribution is singular.
@article{urn:eudml:doc:38386,
title = {Multivariate skewness and kurtosis for singular distributions.},
journal = {Extracta Mathematicae},
volume = {8},
year = {1993},
pages = {98-101},
zbl = {1184.62082},
mrnumber = {MR1285732},
language = {en},
url = {http://dml.mathdoc.fr/item/urn:eudml:doc:38386}
}
Ardanuy, Ramón; Sánchez, José Manuel. Multivariate skewness and kurtosis for singular distributions.. Extracta Mathematicae, Tome 8 (1993) pp. 98-101. http://gdmltest.u-ga.fr/item/urn:eudml:doc:38386/