STATISTICAL ANALYSIS OF PAIRED SAMPLE DATA BY RANKS
I. C. A. Oyeka, Edith Uzoma Umeh · 2012
This paper developed a rank based test statistic for testing the equality of two population medians. Instead of taking the differences between pairs of observations and using these differences with the signs test or taking the absolute values of these differences assigning them ranks and applying the Wilcoxon signed sum rank test, one may first assign ranks to members of the paired observations then used them to develop a rank – based test statistic for testing the equality of two populations. From the above result, the proposed method is observed to be more efficient than the Wilcoxon Signed rank sum test which in turn is more efficient than the ordinary sign test since it uses all available information quite unlike the ordinary Sign test and Wilcoxon signed rank sum test that ignores zero differences. Also it was observed that it had an added advantage over these ones since it enables the statistical comparison of the medians of two related populations that are measurements on as low as the ordinal scale.