A multivariate two-sample mean test for small sample size and missing data

Yujun Wu; Marc G. Genton; Leonard A. Stefanski

doi:10.1111/j.1541-0420.2006.00533.x

A multivariate two-sample mean test for small sample size and missing data

Yujun Wu^*, Marc G. Genton, Leonard A. Stefanski

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

30 Scopus citations

Abstract

We develop a new statistic for testing the equality of two multivariate mean vectors. A scaled chi-squared distribution is proposed as an approximating null distribution. Because the test statistic is based on componentwise statistics, it has the advantage over Hotelling's T² test of being applicable to the case where the dimension of an observation exceeds the number of observations. An appealing feature of the new test is its ability to handle missing data by relying on only componentwise sample moments. Monte Carlo studies indicate good power compared to Hotelling's T² and a recently proposed test by Srivastava (2004, Technical Report, University of Toronto). The test is applied to drug discovery data.

Original language	English (US)
Pages (from-to)	877-885
Number of pages	9
Journal	Biometrics
Volume	62
Issue number	3
DOIs	https://doi.org/10.1111/j.1541-0420.2006.00533.x
State	Published - Sep 2006
Externally published	Yes

Keywords

Drug discovery
High-dimensional data
Small n large p

ASJC Scopus subject areas

General Immunology and Microbiology
Applied Mathematics
General Biochemistry, Genetics and Molecular Biology
General Agricultural and Biological Sciences
Statistics and Probability

Access to Document

10.1111/j.1541-0420.2006.00533.x

Cite this

@article{b3ccfe000e4b40129998c58eaa3b1e98,

title = "A multivariate two-sample mean test for small sample size and missing data",

abstract = "We develop a new statistic for testing the equality of two multivariate mean vectors. A scaled chi-squared distribution is proposed as an approximating null distribution. Because the test statistic is based on componentwise statistics, it has the advantage over Hotelling's T2 test of being applicable to the case where the dimension of an observation exceeds the number of observations. An appealing feature of the new test is its ability to handle missing data by relying on only componentwise sample moments. Monte Carlo studies indicate good power compared to Hotelling's T2 and a recently proposed test by Srivastava (2004, Technical Report, University of Toronto). The test is applied to drug discovery data.",

keywords = "Drug discovery, High-dimensional data, Small n large p",

author = "Yujun Wu and Genton, {Marc G.} and Stefanski, {Leonard A.}",

year = "2006",

month = sep,

doi = "10.1111/j.1541-0420.2006.00533.x",

language = "English (US)",

volume = "62",

pages = "877--885",

journal = "Biometrics",

issn = "0006-341X",

publisher = "Oxford University Press",

number = "3",

}

TY - JOUR

T1 - A multivariate two-sample mean test for small sample size and missing data

AU - Wu, Yujun

AU - Genton, Marc G.

AU - Stefanski, Leonard A.

PY - 2006/9

Y1 - 2006/9

N2 - We develop a new statistic for testing the equality of two multivariate mean vectors. A scaled chi-squared distribution is proposed as an approximating null distribution. Because the test statistic is based on componentwise statistics, it has the advantage over Hotelling's T2 test of being applicable to the case where the dimension of an observation exceeds the number of observations. An appealing feature of the new test is its ability to handle missing data by relying on only componentwise sample moments. Monte Carlo studies indicate good power compared to Hotelling's T2 and a recently proposed test by Srivastava (2004, Technical Report, University of Toronto). The test is applied to drug discovery data.

AB - We develop a new statistic for testing the equality of two multivariate mean vectors. A scaled chi-squared distribution is proposed as an approximating null distribution. Because the test statistic is based on componentwise statistics, it has the advantage over Hotelling's T2 test of being applicable to the case where the dimension of an observation exceeds the number of observations. An appealing feature of the new test is its ability to handle missing data by relying on only componentwise sample moments. Monte Carlo studies indicate good power compared to Hotelling's T2 and a recently proposed test by Srivastava (2004, Technical Report, University of Toronto). The test is applied to drug discovery data.

KW - Drug discovery

KW - High-dimensional data

KW - Small n large p

UR - http://www.scopus.com/inward/record.url?scp=33748763822&partnerID=8YFLogxK

U2 - 10.1111/j.1541-0420.2006.00533.x

DO - 10.1111/j.1541-0420.2006.00533.x

M3 - Article

C2 - 16984331

AN - SCOPUS:33748763822

SN - 0006-341X

VL - 62

SP - 877

EP - 885

JO - Biometrics

JF - Biometrics

IS - 3

ER -

A multivariate two-sample mean test for small sample size and missing data

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this