scientific article; zbMATH DE number 4121200

From MaRDI portal
Publication:4734599





[For the entire collection see Zbl 0638.00031.] This paper is devoted to a discussion of some problems that occur in connexion with the basic likelihood function as it is commonly used in statistics. The arguments presented are not related to measure-theoretic difficulties, but rather to problems of what should be regarded as x and what \(\theta\) in the likelihood function f(x\(| \theta).\) The authors argue that there is no unique way of deciding which random variables and parameters should explicitly enter the likelihood function, and no unique way of deciding which should be regarded as part of x and which as part of \(\theta\). Their general conclusion is that ``a satisfactory theory of statistical inference must be based on the full Bayesian approach in which a complete joint distribution is specified. The ultimate likelihood function is specified as \[ LF \propto f(y| y')= \begin{cases} 1 &\text{ if }y=y',\\ 0 &\text{ otherwise}.\end{cases} \]











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 Q4734599)