scientific article; zbMATH DE number 3998962

From MaRDI portal
Publication:4725480





Let \(x_ 1,...,x_ n\), \(n\geq 2\) be an independent sample from a population \(F(x-\theta)\) depending on a location parameter \(\theta\in {\mathbb{R}}\). Under general conditions of the regularity type it is proved that the Fisher information on \(\theta\) contained in the pair \((\bar x,S^ 2)\), \(\bar x\) and \(S^ 2\) being the sample mean and variance respectively, coincides with that contained in \(\bar x\) iff F is Gaussian. The result refines in a sense the well known theorem due to \textit{T. Kawata} and \textit{H. Sakamoto} [J. Math. Soc. Japan 1, No.2, 111-115 (1949; Zbl 0039.141)] and \textit{A. A. Zinger} [Usp. Mat. Nauk 6, No.5(45), 172-175 (1951; Zbl 0043.341)] characterizing - with no a priori condition - the Gaussian distribution by independence of \(\bar x\) and \(S^ 2\). A similar result characterizing the gamma distribution is proved for the scale parameter case.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4725480)