Test of independence in the Farlie-Gumbel-Morgenstern distribution

Guven B.

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS, vol.32, no.9, pp.1753-1765, 2003 (SCI-Expanded) identifier identifier


We consider the hypotheses; H-0 : theta = 0 vs. H-1 : theta greater than or equal to eta where theta is the dependence parameter of the Farlie-Gumbel-Morgenstren distribution and eta is an element of (0, 1]. A test, which maximizes the minimum power over the alternative hypothesis, is given for these hypotheses. The power function of this test is monotone increasing over the alternative hypothesis. Furthermore, the asymptotic distribution and the approximate power of the test are presented.