A method for obtaining asymptotic expansions under alternatives based on the properties of the likelihood ratio (Q1269923)

We obtain asymptotic expansions under alternatives for distributions of statistics ``close'' to optimal in the problem of testing hypotheses. Sums of independent random variables, linear combinations of order statistics, and linear rank statistics are considered. We base the method for proving the main theorem on the theory due to \textit{L. Le Cam} [see Univ. California Publ. Stat. 3, 37-98 (1960; Zbl 0104.12701)]. A general lemma is proved which is specific to linear combinations of order statistics and linear rank statistics. The proof of the theorem is presented in J. Math. Sci., New York 76, No. 2, 2222-2240 (1995; Zbl 0852.60028). Asymptotic expansions (a.e.) under alternatives play an important role in the asymptotic theory of testing hypotheses. Many authors used a.e. for the investigation of asymptotic efficiency and deficiencies of tests. Besides improving the accuracy of numerical calculations, these expansions may be used for obtaining a.e. the power functions of tests that have the same limit powers, since in this case to compare these tests one should invoke higher-order terms.