Hypothesis testing for sphericity covariance structure without a normality condition

i

This article has been updated. A Correction to this article was published on .

i

The Original Article was published on

i

The Original Article was published on

i

This article has been updated

i

This article was retracted on

i

This article has been updated

Abstract

This paper investigated the sphericity test for high-dimensional covariance matrices, which is a fundamental problem in probability and statistics courses. While several existing powerful tests in textbooks and literature are available, they often rely on either the normality assumption or an unknown fourth moment. This work aimed to relax these restrictions, serving as a complement and extension of traditional course content on this subject. Specifically, we constructed a consistent estimator related to the fourth moment and proposed a new test procedure. The proposed test statistic depends on the second, third, and fourth arithmetic means of the eigenvalues of the sample covariance matrix, but does not require the normality assumption. Numerical simulations illustrated that the proposed test maintains satisfactory empirical size, and achieves power comparable to, or in some cases superior to, the existing tests.

Keywords

hypothesis test / sphericity / covariance matrix / non-normality / high-dimensional data

CHANGE HISTORY

A Correction to this paper has been published:
https://doi.org

The Retraction Note was updated to reflect reception of statements of disagreement from all authors.

This article has been retracted.Please see the Retraction Notice for more detail:
https://doi.org

References

[1] T. Pavlenko, A. Björkström, A. Tillander, Covariance structure approximation via gLasso in high-dimensional supervised classification, J. Appl. Stat., 39 (2012), 1643-1666. https://doi.org/10.1080/02664763.2012.663346

[2] L. Huang, Y. Xiao, K. Liu, H. So, J. Zhang, Bayesian Information criterion for source enumeration in large-scale adaptive antenna array, IEEE Trans. Veh. Technol., 65 (2016), 3018-3032. https://doi.org/10.1109/TVT.2015.2436060

[3] M. Tang, Y. Rong, A. Maio, C. Chen, J. Zhou, Adaptive radar detection in gaussian disturbance with structured covariance matrix via invariance theory, IEEE Trans. Signal Proces., 67 (2019), 5671-5685. https://doi.org/10.1109/TSP.2019.2941119

[4] Y. Rong, A. Aubry, A. Maio, M. Tang, Adaptive radar detection in low-rank heterogeneous clutter via invariance theory, IEEE Trans. Signal Proces., 69 (2021), 1492-1506. https://doi.org/10.1109/TSP.2021.3058447

[5] C. Chen, M. Tang, Y. Rong, Detection of a rare multichannel gaussian signal via higher criticism, IEEE Signal Process. Lett., 29 (2022), 2063-2067. https://doi.org/10.1109/LSP.2022.3212631

[6] X. Wang, B. Zhang, Target selection in shrinkage estimation of covariance matrix: A structural similarity approach, Stat. Probab. Lett., 208 (2024), 110048. https://doi.org/10.1016/j.spl.2023.110048

[7] T. Anderson, An Introduction to Multivariate Statistical Analysis, 3rd Edition, WILEY, 2003.

[8] S. John, Some optimal multivariate tests, Biometrika, 58 (1971), 123-127. https://doi.org/10.1093/BIOMET/58.1.123

[9] S. John, The distribution of a statistic used for testing sphericity of normal distributions, Biometrika, 59 (1972), 169-173. https://doi.org/10.1093/biomet/59.1.169

[10] O. Ledoit, M. Wolf, Some hypothesis tests for the covariance matrix when the dimension is large compared to the sample size, Ann. Stat., 4 (2002), 1081-1102. https://doi.org/10.1214/aos/1031652392

[11] M. Birke, H. Dette, A note on testing the covariance matrix for large dimension, Stat. Probab. Lett., 74 (2005), 281-289. https://doi.org/10.17877/DE290R-15067

[12] M. Srivastava, Some tests concerning the covariance matrix in high dimensional data, J. Jpn. Stat. Soc., 35 (2005), 251-272. https://doi.org/10.14490/jjss.35.251

[13] T. Fisher, X. Sun, C. Gallagher, A new test for sphericity of the covariance matrix for high dimensional data, J. Multivar. Anal., 101 (2010), 2554-2570. https://doi.org/10.1016/J.JMVA.2010.07.004

[14] M. Srivastava, H. Yanagihara, T. Kubokawa, Tests for covariance matrices in high dimension with less sample size, J. Multivar. Anal., 130 (2014), 289-309. https://doi.org/10.1016/j.jmva.2014.06.003

[15] S. Chen, L. Zhang, P. Zhong, Tests for high-dimensional covariance matrices, J. Am. Stat. Assoc., 105 (2010), 810-819. https://doi.org/10.1198/jasa.2010.tm09560

[16] Q. Wang, J. Yao, On the sphericity test with large-dimensional observations, Electron. J. Stat., 7 (2013), 2164-2192. https://doi.org/10.1214/13-EJS842

[17] X. Tian, Y. Lu, W. Li, A robust test for sphericity of high-dimensional covariance matrices, J. Multivar. Anal., 141 (2015), 217-227. https://doi.org/10.1016/j.jmva.2015.07.010

[18] S. Yuan, J. Zhou, J. Pan, J. Shen, Sphericity and identity test for high-dimensional covariance matrix using random matrix theory, Acta Math. Appl. Sin. E., 37 (2021), 214-231. https://doi.org/10.1007/s10255-021-1004-1

[19] T. Fisher, On testing for an identity covariance matrix when the dimensionality equals or exceeds the sample size, J. Stat. Plann. Inference, 142 (2012), 312-326. https://doi.org/10.1016/j.jspi.2011.07.019

[20] K. Xu, Testing diagonality of high-dimensional covariance matrix under non-normality, J. Stat. Comput. Simul., 87 (2017), 3208-3224. https://doi.org/10.1080/00949655.2017.1322630

[21] T. Himeno, T. Yamada, Estimations for some functions of covariance matrix in high dimension under non-normality and its applications, J. Multivar. Anal., 130 (2014), 27-44. https://doi.org/10.1016/j.jmva.2014.04.014

[22] M. Lopes, A. Blandino, A. Aue, Bootstrapping spectral statistics in high dimensions, Biometrika, 106 (2019), 781-801. https://doi.org/10.1093/biomet/asz031

[23] N. Dörnemann, H. Dette, Sequential eigenvalue statistics for change-point detection in covariance matrices, preprint, arXiv:2409.15588. https://doi.org/10.48550/arXiv.2409.15588