Concordance between two linear orders: The Spearman and Kendall coefficients revisited

This paper discusses the two classic measures of concordance between two linear orders L and L′, the Kendall tau and the Spearman rho, equivalently, the Kendall and Spearman distances between such orders. We give an expression for ρ(L,L′)−τ(L,L′) as a function of the parameters of the partial order L∪L′, which allows the determination of extremal values for this difference and an investigation of when tau and rho are equal. This expression for ρ(L,L′)−τ(L,L′) is derived from a relation between the Kendall and Spearman distances between linear orders that is equivalent to both the Guilbaud (1980) formula linking rho, tau, and a third coefficient sigma, and Daniels’s (1950) inequality. We also prove an apparently new monotonicity property of rho. In the conclusion we point out possible extensions and add general historical comments.

[1]  H. Daniels A property of rank correlations , 1948 .

[2]  I. R. Savage Contributions to the Theory of Rank Order Statistics: Applications of Lattice Theory , 1964 .

[3]  R. Möhring Algorithmic Aspects of Comparability Graphs and Interval Graphs , 1985 .

[4]  Rolf H. Möhring,et al.  Computationally Tractable Classes of Ordered Sets , 1989 .

[5]  Donald E. Knuth,et al.  The Art of Computer Programming: Volume 3: Sorting and Searching , 1998 .

[6]  H. E. Daniels,et al.  Rank Correlation and Population Models , 1950 .

[7]  Peter C. Fishburn,et al.  Interval orders and interval graphs : a study of partially ordered sets , 1985 .

[8]  C. Barbut Sur les Graphes Admettant le Nombre Maximum de Sous-Graphes à Trois Sommets et Deux Arêtes, et les Paires d'Ordres Totaux qui Maximisent |Rho — Tau| , 1988 .

[9]  J. M. Bevan,et al.  Rank Correlation Methods , 1949 .

[10]  Monotonicity of Rank Statistics in some Non–parametric: Testing Problems , 1987 .

[11]  M. Kendall Rank Correlation Methods , 1949 .

[12]  W. Kruskal Ordinal Measures of Association , 1958 .

[13]  V. Giakoumakis,et al.  Coefficients d'accord entre deux préordres totaux , 1987 .

[14]  H. Griffin,et al.  Graphic Computation of Tau as a Coefficient of Disarray , 1958 .

[15]  G. Guilbaud Relation entre les deux coefficients de corrélation de rangs , 1980 .

[16]  James Durbin,et al.  Inversions and Rank Correlation Coefficients , 1951 .

[17]  D. Critchlow Metric Methods for Analyzing Partially Ranked Data , 1986 .

[18]  P. Moran Recent Developments in Ranking Theory , 1950 .

[19]  Joseph S. Verducci,et al.  Probability Models and Statistical Analyses for Ranking Data , 1992 .

[20]  I. R. Savage,et al.  Contributions to the theory of rank order statistics , 1954 .

[21]  H. Young Condorcet's Theory of Voting , 1988, American Political Science Review.

[22]  P. Diaconis Group representations in probability and statistics , 1988 .

[23]  Bernard Monjardet Sur diverses formes de la \regle de Condorcet , 1990 .

[24]  W. Trotter,et al.  Combinatorics and Partially Ordered Sets: Dimension Theory , 1992 .

[25]  Bernard Monjardet,et al.  The median procedure in cluster analysis and social choice theory , 1981, Math. Soc. Sci..

[26]  Bernard Monjardet,et al.  A use for frequently rediscovering a concept , 1985 .

[27]  B. Monjardet,et al.  Concordance et consensus d'ordres totaux: Les coefficients K et W , 1985 .

[28]  R. Graham,et al.  Spearman's Footrule as a Measure of Disarray , 1977 .

[29]  S. Merrill A Comparison of Efficiency of Multicandidate Electoral Systems , 1984 .