Rank-Sum Comparisons between Groups

A common and deceptively simple step in research is making group comparisons in terms of some ordered characteristic such as age, income, occupational prestige, and the like. Ideally these comparisons are best made graphically, with the entire frequency distribution presented for each group. However, summary measures are generally used because it is impractical to present a large number of graphs and because quantitative measures allow the results to be correlated with other attributes of the groups. Thus researchers usually turn to such well-known measures of central tendency as arithmetic means and medians or to such dis-