Rank tests for matched pair experiments with censored data (Q1118286)

We consider the problem of testing bivariate symmetry in matched pair experiments where \((X_ 1,X_ 2)\) are time measurements such as failure or survival times. The observations are subject to random right censoring so that what is observed is \(Y_ j=\min (X_ j,Z_ j)\) and \(\delta_ j=I(X_ j=Y_ j)\), \(j=1,2\), where \((Z_ 1,Z_ 2)\) is a pair of censoring times independent of \((X_ 1,X_ 2)\). Tests that generalize the conditional Wilcoxon and the log-rank tests are considered as well as general linear rank statistics. It is shown that suitably standardized versions of these statistics are asymptotically normal under fixed and converging alternatives and they are consistent against the alternative of ordered hazards.