A Mixture-Model Approach for Parallel Testing for Unequal Variances : Statistical Applications in Genetics and Molecular Biology

www.degruyter.com uses cookies, tags, and tracking settings to store information that help give you the very best browsing experience.
To understand more about cookies, tags, and tracking, see our Privacy Statement
I accept all cookies for the De Gruyter Online site

Jump to ContentJump to Main Navigation

Statistical Applications in Genetics and Molecular Biology

Editor-in-Chief: Stumpf, Michael P.H.

IMPACT FACTOR increased in 2014: 1.127
5-year IMPACT FACTOR: 1.537
Rank 47 out of 122 in category Statistics & Probability in the 2014 Thomson Reuters Journal Citation Report/Science Edition

SCImago Journal Rank (SJR) 2014: 0.740
Source Normalized Impact per Paper (SNIP) 2014: 0.470
Impact per Publication (IPP) 2014: 0.926

Mathematical Citation Quotient (MCQ) 2014: 0.17


30,00 € / $42.00 / £23.00

Get Access to Full Text

A Mixture-Model Approach for Parallel Testing for Unequal Variances

Haim Y. Bar1 / James G. Booth2 / Martin T. Wells3

1Cornell University

2Cornell University

3Cornell University

Citation Information: Statistical Applications in Genetics and Molecular Biology. Volume 11, Issue 1, Pages 1–21, ISSN (Online) 1544-6115, DOI: 10.2202/1544-6115.1762, January 2012

Publication History

Published Online:

Testing for unequal variances is usually performed in order to check the validity of the assumptions that underlie standard tests for differences between means (the t-test and anova). However, existing methods for testing for unequal variances (Levene's test and Bartlett's test) are notoriously non-robust to normality assumptions, especially for small sample sizes. Moreover, although these methods were designed to deal with one hypothesis at a time, modern applications (such as to microarrays and fMRI experiments) often involve parallel testing over a large number of levels (genes or voxels). Moreover, in these settings a shift in variance may be biologically relevant, perhaps even more so than a change in the mean. This paper proposes a parsimonious model for parallel testing of the equal variance hypothesis. It is designed to work well when the number of tests is large; typically much larger than the sample sizes. The tests are implemented using an empirical Bayes estimation procedure which `borrows information' across levels. The method is shown to be quite robust to deviations from normality, and to substantially increase the power to detect differences in variance over the more traditional approaches even when the normality assumption is valid.

Keywords: empirical Bayes; EM algorithm; shrinkage estimation; false discovery rate; mixture model; simultaneous tests

Citing Articles

Here you can find all Crossref-listed publications in which this article is cited. If you would like to receive automatic email messages as soon as this article is cited in other publications, simply activate the “Citation Alert” on the top of this page.

Babak Shahbaba and Wesley O. Johnson
Statistics in Medicine, 2013, Volume 32, Number 12, Page 2114

Comments (0)

Please log in or register to comment.