A New Coefficient of Correlation

Abstract

Is it possible to define a coefficient of correlation which is (a) as simple as the classical coefficients like Pearson's correlation or Spearman's correlation, and yet (b) consistently estimates some simple and interpretable measure of the degree of dependence between the variables, which is 0 if and only if the variables are independent and 1 if and only if one is a measurable function of the other, and (c) has a simple asymptotic theory under the hypothesis of independence, like the classical coefficients? This article answers this question in the affirmative, by producing such a coefficient. No assumptions are needed on the distributions of the variables. There are several coefficients in the literature that converge to 0 if and only if the variables are independent, but none that satisfy any of the other properties mentioned above.2010 Mathematics Subject Classification. 62H20, 62H15.

Measuring Complexity in Financial Data

“…First, the results indicate that the information content of the mutual information matrix and that of the correlation matrix become quite similar after the year 2000, so a non-linear measure, such as mutual information is not very useful. There are, however, some new measures of association, with different asymptotic theory (e.g., [46]), that could be explored in future work. Second, an open problem relating to the construction of the Granger causal matrix from pairwise regression is that it does not test for joint significance and there can be type I error due to multiple testing, leading to false discovery of edges [47].…”

Does divergence from normal patterns of integration increase as chromosomal fusions increase in number? A test on a house mouse hybrid zone

“…Furthermore, to more quantitatively asses this potential relationship, we compute and test for significance the correlation between the deviation from the PLS model computed on the standard race (expressed in Procrustes distance) and diploid number. To allow for both linear and non-linear relationships, we employ both the usual Pearson product moment correlation coefficient, which is best suited for linear relationships, and a recently proposed correlation coefficient Xi ( Chatterjee 2020 ) which is asymptotically comprised between 0 and 1 and is useful for non-linear and non-monotonic relationships. As Xi is not a symmetric coefficient ( Chatterjee 2020 ), we computed it in the R package XICOR ( Chatterjee 2020 ) considering Procrustes distance from standard integration as dependent on diploid number (as per our biological hypothesis).…”

Testing for Endogeneity of Covid-19 Patient Assignments

“…We test graphically the null hypothesis by verifying if the asymptotically valid confidence interval at level 95% contains 0, which is a test for absence of endogeneity at date t . 4 Thus, we do not average the conditional dependence measure with respect to the threshold as proposed in Chatterjee (2020) .…”