An algorithm for construction of multiple hypothesis testing (Q1861597)

It is defined that \(X=(X_1,\dots,X_n)\) has a multivariate \(r\) th order conditionally independent distribution (\(MCI_r\)) if the pdf of \(X\) can be represented in the form \[ \int\prod_{i=1}^n f(x_i,z_1,\dots,z_r)h_1(z)\dots h_r(z_r)\prod_{j=1}^rdz_j. \] The author derives an algorithm for the evaluation of the probabilities \(\Pr\{X_{1:n}<b_1,\dots,X_{n:n}<b_n\}\), \(X_{i:n}\) being the order statistics of some \(MCI_r\) vector. The algorithm is used to calculate the critical values for Simes-type procedures of multiple hypothesis testing.