COMPARISON BETWEEN THE CONDITIONAL LIKELIHOOD RATIO AND THE USUAL LIKELIHOOD RATIO TESTS

This paper compares, under contiguous alternatives, the conditional likelihood ratio test of Cox and Reid with the usual likelihood ratio test on the basis of second-order power. It is seen that although the power of the former is identical, up to the second order, with that of the likelihood ratio test with known nuisance parameter, it does not hold with the latter. This provides a justification for the use of the conditional likelihood ratio test, rather than the usual likelihood ratio test, in the presence of a nuisance parameter.