Exact Bahadur Slope for Combining Independent Tests In The Case of Conditional Laplace Distribution


Abstract


This paper compares four methods of combining n independent tests. The methods are Fisher, logistic, sum of p-values and inverse normal. It is as- sumed that n independent test statistics {T(i),i = 1,...,n} are available to combine the n independent tests. The four methods are compared, as n → ∞, via exact Bahadur slope under the assumption that the test statistics follow Conditional Laplace Distribution T(i)|ξi ∼ L(τξi,1), ξi [a,),a 0 where ξ1, ξ2, ... are distributed according to the distribution function (DF) Hξ . It is shown that Fisher’s method performs the best as the evidence against the null hypothesis


DOI Code: 10.1285/i20705948v17n2p334

Keywords: Meta-analysis , Conditional Laplace distribution , combining independent tests , Bahadur efficiency , Exact Bahadur Slope.

Full Text: pdf
کاغذ a4

Creative Commons License
This work is licensed under a Creative Commons Attribuzione - Non commerciale - Non opere derivate 3.0 Italia License.