Abstract
It it is known for a linear hypothesis testing problem that the alternative region is constrained by some restrictions, then besides of the usual F-test there offer other criteriae for testing the hypothesis. In this paper it is assumed that all possible parameters are contained in a circular cone. The null-distribution of the likelihood ratio statistic is evaluated and various monotonicity properties of the power function are proved. It follows for some special cases that on the alternative the likelihood ratio criterion is uniformly more powe ful than the F. test. At the end is given a short survey on other tests known in the literature.
1The paper is partially based on [10].
1The paper is partially based on [10].
Notes
1The paper is partially based on [10].