References & Citations
Statistics > Methodology
Title: A New Non-parametric Test for Multivariate Paired Data and Pair Matching
(Submitted on 3 Jul 2020 (v1), last revised 19 Sep 2021 (this version, v2))
Abstract: In paired design studies, it is common to have multiple measurements taken for the same set of subjects under different conditions. In observational studies, it is many times of interest to conduct pair matching on multiple covariates between a treatment group and a control group, and to test the treatment effect represented by multiple response variables on well pair-matched data. However, there is a lack of an effective test on multivariate paired data. The multivariate paired Hotelling's $T^2$ test can sometimes be used, but its power decreases fast as the dimension increases. Existing methods for assessing the balance of multiple covariates in matched observational studies usually ignore the paired structure and thus they do not perform well under some settings. In this work, we propose a new non-parametric test for paired data, which exhibits a substantial power improvement over existing methods under a wide range of situations. We also derive the asymptotic distribution of the new test and the approximate $p$-value is reasonably accurate under finite samples through simulation studies even when the dimension is larger than the sample size, making the new test an easy-off-the-shelf tool for real applications. The proposed test is illustrated through an analysis of a real data set on the Alzheimer's disease research.
Submission history
From: Hao Chen [view email][v1] Fri, 3 Jul 2020 04:59:26 GMT (141kb,D)
[v2] Sun, 19 Sep 2021 19:27:37 GMT (187kb,D)
Link back to: arXiv, form interface, contact.