Hodges-Lehmann and Chernoff efficiencies of linear rank statistics (Q1193962)

The two-sample problem is considered. To check the hypothesis \(H: F_ 1=F_ 2\) versus the alternative \(A: F_ 1=G(x,0)\), \(F_ 2=G(x,\theta)\), \(\theta>0\), on the base of two independent samples \(X_ 1,\dots,X_ m\) and \(Y_ 1,\dots,Y_ n\), the linear rank statistics of the form \[ S_ N=N^{-1}\sum^ m_{i=1}J(R_ i/(N+1)) \] are considered. Here \(R_ i\), \(i=1,\dots,m\), is the rank of the observation \(X_ i\) in the combined sample, \(N=m+n\), \(J\) is a score function. The main result of the paper consists in the observation that local Hodges- Lehmann, Chernoff and Bahadur efficiencies coincide with the intermediate and Pitman efficiencies.