A test, introduced by Gehan in 1965, as an alternative to the Mantel–Haenszel test for comparing two survivor functions (see hazard rate) using censored data (because, for example, some patients are still living). With two samples, of sizes m and n, the test statistic is W, given by The calculation of NA>B and NB>A requires examination of the mn pairs of observations (with one member of each pair from each sample). The quantity NA>B is the number of pairs in which the member chosen from sample A is certain to have a longer lifetime than the member chosen from sample B.