scientific article; zbMATH DE number 3976162

From MaRDI portal
Publication:3740867





Let \(\{F_ 0(\cdot;\theta)\), \(\theta\in \Theta \}\) be a family of distribution functions parameterized by a p-vector \(\theta\). Let \(X_ 1,...,X_ n\) be a random sample from a distribution F. Suppose it is desired to test the null hypothesis \(H_ 0: F\in \{F_ 0(\cdot;\theta)\), \(\theta\in \Theta \}\) versus some general alternative hypothesis. For example, the general alternative hypothesis may be that F has monotone hazard rate. Let \(c_ n(t)=n^{-1}\sum^{n}_{j=1}\exp \{itX_ j\}\) and \(c_ 0(t;\theta)=\int^{\infty}_{-\infty}e^{itx} dF_ 0(x;\theta)\) be, respectively, the characteristic functions of the empirical and of the hypothesized distributions. The authors consider a test statistics, for testing \(H_ 0\), of the form \[ C({\hat \theta})=\int^{\infty}_{-\infty}D\{c_ n(t),\quad c_ 0(t;{\hat \theta})\}dW(t) \] where D is some distance function, W is a complex- valued weighting function and \({\hat \theta}\) the maximum likelihood estimator of \(\theta\). The case where \(F_ 0\) is the exponential distribution is studied in detail. Some asymptotic properties are derived. A comparison with other tests for exponentiality is given.




Cited in
(27)








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 Q3740867)