Distribution of Fisher's combination statistic when the tests are dependent
Author:
- 9 January 2009
- journal article
- research article
- Published by Taylor & Francis in Journal of Statistical Computation and Simulation
- Vol. 80 (1) , 1-12
Abstract
Many questions in multivariate analysis involve the situation when the number of variables is greater than the available sample size. Combination test statistics provides one method to deal with this situation. However, the properties of this approach have not been fully investigated. In this article, an approximation to the null distribution of Fisher's method of combining one-sided test statistics was investigated using the pairwise correlations. The unbiased variance estimate of the combination test statistic is derived. A Monte Carlo study indicates that the new approximation provides a closer type-I error rate than other existing methods.Keywords
This publication has 13 references indexed in Scilit:
- OligoSpawn: a software tool for the design of overgo probes from large unigene datasetsBMC Bioinformatics, 2006
- Combining dependent P-valuesStatistics & Probability Letters, 2002
- Pros and cons of permutation tests in clinical trialsStatistics in Medicine, 2000
- Recent Advances in the Central Limit Theorem and Its Weak Invariance Principle for Mixing Sequences of Random Variables (A Survey)1986
- Estimation of a Single Effect Size: Parametric and Nonparametric Methods1985
- 400: A Method for Combining Non-Independent, One-Sided Tests of Significance1975
- Asymptotic Optimality of Fisher's Method of Combining Independent Tests IIJournal of the American Statistical Association, 1973
- Asymptotic Optimality of Fisher's Method of Combining Independent TestsJournal of the American Statistical Association, 1971
- Unbiased Estimation of Certain Correlation CoefficientsThe Annals of Mathematical Statistics, 1958
- The central limit theorem for dependent random variablesDuke Mathematical Journal, 1948