A new weighted rank coefficient of concordance

There are many situations where n objects are ranked by b>2 independent sources or observers and in which the interest is focused on agreement on the top rankings. Kendall's coefficient of concordance [10 M. Kendall and B. Smith, The problem of m rankings, Ann. Math. Stat. 10 (1939), pp. 275–287. doi: 10.1214/aoms/1177732186[Crossref] , [Google Scholar]] assigns equal weights to all rankings. In this paper, a new coefficient of concordance is introduced which is more sensitive to agreement on the top rankings. The limiting distribution of the new concordance coefficient under the null hypothesis of no association among the rankings is presented, and a summary of the exact and approximate quantiles for this coefficient is provided. A simulation study is carried out to compare the performance of Kendall's, the top-down and the new concordance coefficients in detecting the agreement on the top rankings. Finally, examples are given for illustration purposes, including a real data set from financial market indices.     

Concordance coefficients to measure the agreement among several sets of ranks

In this paper, two measures of agreement among several sets of ranks, Kendall's concordance coefficient and top-down concordance coefficient, are reviewed. In order to illustrate the utility of these measures, two examples, in the fields of health and sports, are presented. A Monte Carlo simulation study was carried out to compare the performance of Kendall's and top-down concordance coefficients in detecting several types and magnitudes of agreements. The data generation scheme was developed in order to induce an agreement with different intensities among m (m>2) sets of ranks in non-directional and directional rank agreement scenarios. The performance of each coefficient was estimated by the proportion of rejected null hypotheses, assessed at 5% significance level, when testing whether the underlying population concordance coefficient is sufficiently greater than zero. For the directional rank agreement scenario, the top-down concordance coefficient allowed to achieve a percentage of significant concordances that was higher than the one achieved by Kendall's concordance coefficient. Mainly, when the degree of agreement was small, the results of the simulation study pointed to the advantage of using a weighted rank concordance, namely the top-down concordance coefficient, simultaneously with Kendall's concordance coefficient, enabling the detection of agreement (in a top-down sense) in situations not detected by Kendall's concordance coefficient.     

Testing for concordance between several criteria

The problem of whether the rankings of some objects given by a set of criteria (or judges) show any agreement or are more or less independent is addressed. The most familiar measure for concordance is the Kendall W coefficient. Classical tests for concordance are the Friedman test and the F test. Legendre [Species associations: the Kendall coefficient of concordance revisited. J. Agric. Biol. Environ. Stat. 2005;10(2):226–245] compared via simulation the Friedman test and its permutation version. Unfortunately, the simulation study of Legendre was very limited because it considered neither the copula aspect nor the F test. Kendall W is a rank-based correlation measure, and therefore it is not affected by the marginal distributions of the underlying variables, but only by the copula of the multivariate distribution. In this article, the simulation study of Legendre is deeply extended by considering the copula aspect as well as the F test. It is shown that the Friedman test is too conservative and less powerful than both the F test and the permutation test for concordance which always have a correct size and behave alike. The F test should be preferred because it is computationally much easier. Surprisingly, the power function of the tests is not much affected by the type of copula.     

KENDALL'S TAU AND CONTINGENCY TABLES

Kendall's tau is a coefficient of concordance between two rankings of n objects. Its definition and large sample normal approximation are easily extended to the case where one of the rankings contains ties. In this paper, definition and normal approximation are extended further to the case where both rankings contain ties. The results are applied to give a fully distribution-free test for two-way contingency tables with ordered categories.     

Test for the equality of several correlation coefficients

We derive two C(α) statistics and the likelihood-ratio statistic for testing the equality of several correlation coefficients, from k ≥ 2 independent random samples from bivariate normal populations. The asymptotic relationship of the C(α) tests, the likelihood-ratio test, and a statistic based on the normality assumption of Fisher's Z-transform of the sample correlation coefficient is established. A comparative performance study, in terms of size and power, is then conducted by Monte Carlo simulations. The likelihood-ratio statistic is often too liberal, and the statistic based on Fisher's Z-transform is conservative. The performance of the two C(α) statistics is identical. They maintain significance level well and have almost the same power as the other statistics when empirically calculated critical values of the same size are used. The C(α) statistic based on a noniterative estimate of the common correlation coefficient (based on Fisher's Z-transform) is recommended.     

Detecting a Change in the Correlation Coefficient in a Sequence of Bivariate Normal Variables

We consider the problem of testing the hypothesis that the correlation coefficient is stable in a sequence of n observations of independent, bivariate normal random variables against the alternative that the correlation coefficient changes after an unknown point t(t < n). We propose an estimate of the changepoint t and report on power comparisons between the commonly used test for this problem and our proposed test. Some applications to finance are discussed.     

Ordered Alternatives and the 1 1/2-Tail Test

Under proper conditions, two independent tests of the null hypothesis of homogeneity of means are provided by a set of sample averages. One test, with tail probability P ₁, relates to the variation between the sample averages, while the other, with tail probability P ₂, relates to the concordance of the rankings of the sample averages with the anticipated rankings under an alternative hypothesis. The quantity G = P ₁ P ₂ is considered as the combined test statistic and, except for the discreteness in the null distribution of P ₂, would correspond to the Fisher statistic for combining probabilities. Illustration is made, for the case of four means, on how to get critical values of G or critical values of P ₁ for each possible value of P ₂, taking discreteness into account. Alternative measures of concordance considered are Spearman's ρ and Kendall's τ. The concept results, in the case of two averages, in assigning two-thirds of the test size to the concordant tail, one-third to the discordant tail.     

A simple generalization of the Box–Muller method for obtaining a pair of correlated standard normal variables

In the simple and widely used method of Box–Muller [G. Box and M. Muller, A note on the generation of random normal deviates, Ann. Math. Statist. 29 (1958), pp. 610–611], from a pair of uniform and independent random variables in (0,1), a pair of standard and independent normal variables is obtained. In this article, we present a very simple and elegant generalization of this method to obtain a pair of correlated standard normal variables with a given coefficient of correlation. This generalized method, which is computationally very easy, is interpreted in geometric terms, considering a translation of the uniform interval (0,1) and a rotation of a defined angle, both related to the coefficient of correlation. Some numerical results are simulated and statistically analysed, proving that the generalization is extremely simple and powerful.     

General scores statistics on ranks in the analysis of unbalanced designs

The authors consider the situation of incomplete rankings in which n judges independently rank k_i ∈ {2, …, t} objects. They wish to test the null hypothesis that each judge picks the ranking at random from the space of k_i! permutations of the integers 1, …, k_i. The statistic considered is a generalization of the Friedman test in which the ranks assigned by each judge are replaced by real‐valued functions a(j, k_i), 1 ≤ j ≤ k_i ≤ t of the ranks. The authors define a measure of pairwise similarity between complete rankings based on such functions, and use averages of such similarities to construct measures of the level of concordance of the judges' rankings. In the complete ranking case, the resulting statistics coincide with those defined by Hájek & ?idák (1967, p. 118), and Sen (1968). These measures of similarity are extended to the situation of incomplete rankings. A statistic is derived in this more general situation and its properties are investigated.     

Testing for the Presence of a Maverick Judge

Abstract

The presence of a maverick judge, one whose rankings differ greatly from the other members of a panel, can result in incorrect rankings and a sense of unfairness among contestants. We develop and explore the properties of a likelihood ratio test, assuming a Mallows type distribution, for the presence of a maverick judge when each judge selects his or her best k out of n objects, k ≤ n. Detection of a maverick judge, who may be viewed as a multivariate outlier, turns out to be very difficult unless the judges are very consistent and there are repeat observations on the panel.     

Analysis of multiple sets of incomplete rankings

The L statistic is extended to allow incomplete or partial rankings within groups. As a special case of these extensions, tests for ofered alternatives analogous to the page test are provided for other than complete rankings. A two group illustration is provided by an alternative analysis of the Bradley and Terry taste-testing experiment on pork-roasts.     

On the correlation coefficient between the first and the r-th smallest order statistics based on n independent exponential random variables

It is proved that the correlation coefficient between the first and the r-th smallest order statistics based on n independent, exponentially distributed random variables is less than or equal to the corresponding correlation coefficient when the n indpendent random variables are identically and exponentially distributed with expectation equal to unity     

Testing symmetry based on empirical likelihood

In this paper, we propose a general kth correlation coefficient between the density function and distribution function of a continuous variable as a measure of symmetry and asymmetry. We first propose a root-n moment-based estimator of the kth correlation coefficient and present its asymptotic results. Next, we consider statistical inference of the kth correlation coefficient by using the empirical likelihood (EL) method. The EL statistic is shown to be asymptotically a standard chi-squared distribution. Last, we propose a residual-based estimator of the kth correlation coefficient for a parametric regression model to test whether the density function of the true model error is symmetric or not. We present the asymptotic results of the residual-based kth correlation coefficient estimator and also construct its EL-based confidence intervals. Simulation studies are conducted to examine the performance of the proposed estimators, and we also use our proposed estimators to analyze the air quality dataset.     

Estimating the correlation coefficient in the presence of correlated observations from a bivariate normal population

Estimation of the correlation coefficient between two variates (p) in the presence of correlated observations from a bivar iate normal population is considered The estimated maximum likelihood estimator (EMLE), an estimate based on the maximum likelihood estimator (MLE), is proposed and studied for the estimation of p For the large sample case , approximate expressions foi the variance and the bias of the Pearson estimate of the correlation coefficient are derived. These expressions suggests that the Pearson’s estimator possesses high mean square error (MSE) in estimating ρ in comparison to the MLE The MSE is particularly high when the observations within clusters aie highly correlated. The Pearson’s estimate, the MLE, and the EMLE aie evaluated in a simulation study This study shows that the proposed EMLE pefoims bettei than the Pearson’s correlation coefficient except when the number of clusters is small.     

A test of concordance based on Gini’s mean difference

A new rank correlation index, which can be used to measure the extent of concordance or discordance between two rankings, is proposed. This index is based on Gini’s mean difference computed on the totals ranks corresponding to each unit and it turns out to be a special case of a more general measure of the agreement of m rankings. The proposed index can be used in a test for the independence of two criteria used to rank the units of a sample, against their concordance/discordance. It can then be regarded as a competitor of other classical methods, such as Kendall’s tau. The exact distribution of the proposed test-statistic under the null hypothesis of independence is studied and its expectation and variance are determined; moreover, the asymptotic distribution of the test-statistic is derived. Finally, the implementation of the proposed test and its performance are discussed. Both the authors contributed equally to this work; however, the actual writing of the paper was as follows: Sects. 2 and 3 are due to C. G. Borroni, Sects. 1 and 4 are due to M. Zenga.     

Estimators of the multiple correlation coefficient: Local robustness and confidence intervals

Many robust regression estimators are defined by minimizing a measure of spread of the residuals. An accompanying R ²-measure, or multiple correlation coefficient, is then easily obtained. In this paper, local robustness properties of these robust R ²-coefficients are investigated. It is also shown how confidence intervals for the population multiple correlation coefficient can be constructed in the case of multivariate normality.     

New weighted rank correlation coefficients sensitive to agreement on top and bottom rankings

Three new weighted rank correlation coefficients are proposed which are sensitive to both agreement on top and bottom rankings. The first one is based on the weighted rank correlation coefficient proposed by Maturi and Abdelfattah [13 T.A. Maturi and E.H. Abdelfattah, A new weighted rank correlation, J. Math. Stat. 4 (2008), pp. 226–230. doi: 10.3844/jmssp.2008.226.230[Crossref] , [Google Scholar]], the second and the third are based on the order statistics and the quantiles of the Laplace distribution, respectively. The limiting distributions of the new correlation coefficients under the null hypothesis of no association between the rankings are presented, and a summary of the exact and approximate quantiles for these coefficients is provided. A simulation study is performed to compare the performance of Kendall's tau, Spearman's rho, and the new weighted rank correlation coefficients in detecting the agreement on the top and the bottom rankings simultaneously. Finally, examples are given for illustration purposes, including a real data set from financial market indices.     

Evaluating the Significance Test When the Correlation Coefficient is Different from Zero in the Test of Hypothesis

Sample size and correlation coefficient of populations are the most important factors which influence the statistical significance of the sample correlation coefficient. It is observed that for evaluating the hypothesis when the observed value of the correlation coefficient's r is different from zero, Fisher's Z transformation may be incorrect for small samples especially when population correlation coefficient ρ has big values. In this study, a simulation program has been generated for to illustrate how the bias in the Fisher transformation of the correlation coefficient affects estimate precision when sample size is small and ρ has big value. By the simulation results, 90 and 95% confidence intervals of correlation coefficients have been created and tabled. As a result, it is suggested that especially when ρ is greater than 0.2 and sample sizes of 18 or less, Tables 1 and 2 can be used for the significance test in correlations.     

A WEIGHTED RANK MEASURE OF CORRELATION

Spearman's rank correlation coefficient is not entirely suitable for measuring the correlation between two rankings in some applications because it treats all ranks equally. In 2000, Blest proposed an alternative measure of correlation that gives more importance to higher ranks but has some drawbacks. This paper proposes a weighted rank measure of correlation that weights the distance between two ranks using a linear function of those ranks, giving more importance to higher ranks than lower ones. It analyses its distribution and provides a table of critical values to test whether a given value of the coefficient is significantly different from zero. The paper also summarizes a number of applications for which the new measure is more suitable than Spearman's.