Premium
An Omnibus Test for Univariate and Multivariate Normality *
Author(s) -
Doornik Jurgen A.,
Hansen Henrik
Publication year - 2008
Publication title -
oxford bulletin of economics and statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.131
H-Index - 73
eISSN - 1468-0084
pISSN - 0305-9049
DOI - 10.1111/j.1468-0084.2008.00537.x
Subject(s) - kurtosis , univariate , normality test , normality , statistics , bivariate analysis , multivariate statistics , skewness , multivariate normal distribution , mathematics , omnibus test , econometrics , multivariate analysis , statistical hypothesis testing
We suggest a convenient version of the omnibus test for normality, using skewness and kurtosis based on Shenton and Bowman [ Journal of the American Statistical Association (1977) Vol. 72, pp. 206–211], which controls well for size, for samples as low as 10 observations. A multivariate version is introduced. Size and power are investigated in comparison with four other tests for multivariate normality. The first power experiments consider the whole skewness–kurtosis plane; the second use a bivariate distribution which has normal marginals. It is concluded that the proposed test has the best size and power properties of the tests considered.