One-sided and multi-sided rank tests against trend in the case of concordant observers (Q1066589)

Assume that to m observers (independent) samples from k populations are available in the form \(X_{\alpha,i,l}=\phi_{\alpha}(X_{i,l})\), \(\alpha =1,...,m\); \(i=1,...,k\); \(1\leq l\leq n_{\alpha,i}\), where \(\phi_{\alpha}\) is an unknown strictly increasing function. Let \(F_ i\) be the common distribution function of the \(X_{i,l}\), \(1\leq l\leq n_{\alpha,i}.\) The author investigates certain rank-tests for \(H_ 0:F_ 1=...=F_ k\) versus certain location and regression type (local) alternatives. The results constitute extensions of the classical case \(m=1\), \(\phi_{\alpha}=id\), which, e.g., is dealt with in detail in the book by \textit{J. Hájek} and \textit{Z. Šidák}, Theory of rank tests. (1967; Zbl 0161.381).