Characterizing parameters of multivariate elliptical distributions*
Maia Berkane, Peter M. Bentler
Abstract
Maia Berkane, Peter M. Bentler
Abstract
This paper defines new parameters characterizing multivariate elliptical distributions. Mardia's coefficient of multivariate kurtosis is shown to be essentially one of these parameters. A simple relation is established between centered multivariate product moments and second moments of the variables. The general results are verified on the contaminated normal distribution as an example.
OpenAlex reports 13 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This paper defines new parameters characterizing multivariate elliptical distributions. Mardia's coefficient of multivariate kurtosis is shown to be essentially one of these parameters. A simple relation is established between centered multivariate product moments and second moments of the variables. The general results are verified on the contaminated normal distribution as an example.
Key concepts: Multivariate statistics, Kurtosis, Elliptical distribution, Mathematics, Multivariate analysis, Multivariate stable distribution, Matrix t-distribution, Simple (philosophy)