Note on Rank-Biserial Correlation when There Are Ties

The objective of this article is to demonstrate with examples that the two-sided tie correction does not work well. This was developed by Cureton so Kendall’s tau-type and Spearman’s rho-type formulas for rank-biserial correlation yield same result when ties are present. However, a based on bracket achieves desired goal, which demonstrated algebraically checked three examples. On one hand, 10-element random sample given Cureton, in performs well, taken up. other two given, 7-element clinical 31 participants, work, but new does. It concluded corrected coincide Goodman-Kruskal’s gamma as compared Glass’ formula matches Somers’ dY|X or asymmetric measure association Y ranking respect X dichotomy. use underreported coefficient suggested, very easy calculate from its equivalence Kruskal-Wallis’ dY|X.