Testing of Two Samples

VARIOUS methods, parametric and non-parametric, are used for examining the significance of the difference between two given samples. The most common of these is the parametric one, the t -test 1 , which is based on the assumption that the samples are drawn from a normal population. There is Pitman's 2 w -test which has been built up by considering all the possible samples that can be drawn by pooling together the two samples. Besides these, Wald and Wolfowitz 3 have used the run theory for deciding the significance of the difference between two samples. This method has been extended for k samples by myself 4 . Recently, tests have been constructed by Wallis 5 , Kruskall 6 and Rijkoort 7 by using rank numbers for each of the samples.