Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · math/0701003

To how many simultaneous hypothesis tests can normal, Student's t or bootstrap calibration be applied?

J. Fan·P. Hall·Q. Yao
View original source ↗Read source document ↗Export metadata ↓

Abstract

In the analysis of microarray data, and in some other contemporary statistical problems, it is not uncommon to apply hypothesis tests in a highly simultaneous way. The number, $ν$ say, of tests used can be much larger than the sample sizes, $n$, to which the tests are applied, yet we wish to calibrate the tests so that the overall level of the simultaneous test is accurate. Often the sampling distribution is quite different for each test, so there may not be an opportunity for combining data across samples. In this setting, how large can $ν$ be, as a function of $n$, before level accuracy becomes poor? In the present paper we answer this question in cases where the statistic under test is of Student's $t$ type. We show that if either Normal or Student's $t$ distribution is used for calibration then the level of the simultaneous test is accurate provided $\logν$ increases at a strictly slower rate than $n^{1/3}$ as $n$ diverges. On the other hand, if bootstrap methods are used for calibration then we may choose $\logν$ almost as large as $n\half$ and still achieve asymptotic level accuracy. The implications of these results are explored both theoretically and numerically.

Explore related subjects

math.STstat.TH

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

J. Fan, P. Hall, Q. Yao. 2006-12-29. To how many simultaneous hypothesis tests can normal, Student's t or bootstrap calibration be applied?. https://arxiv.org/abs/math/0701003

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Second-order PACF asymptotics and discrimination between fractional Gaussian noise and $\FARIMA(0,d,0)$

Fractional Gaussian noise and $\FARIMA(0,d,0)$ have the same long-memory pole $|θ|^{-2d}$ and hence the same leading PACF law $α(n)\sim d/n$. We show that this agreement breaks at the first non-universal order. For $0<d<1/2$, the pure fGn PACF satisfies $$ α_{\fGn}(n)=\frac d n+\frac{C_{\fGn}(d)}{n^2}+o(n^{-2}), \qquad C_{\fGn}(d)<d^2, $$ The proof uses the Bingham--Inoue--Kasahara representation, a phase-coefficient expansion for fGn, and a Hankel-operator perturbation argument. Thus the fGn spectral envelope is invisible at first order but visible in second-order finite prediction, explaining why short-memory order selection can differ when fGn data are fitted by FARIMA-type models.

math.ST↗
arXiv2026

U-centering as subset ANOVA: edge regression and higher-order theory

Unbiased squared distance covariance and the Hilbert-Schmidt independence criterion (HSIC) are fourth-order U-statistics, yet U-centering evaluates them from pairwise arrays in $O(n^2)$ operations. We show that U-centering is exactly least-squares residualization after fitting additive endpoint effects. This analysis of variance (ANOVA) interpretation explains the zero row sums and the normalization through residual degrees of freedom. The same geometry gives regression representations of dependence and two-sample statistics, exact permutation variances, and relations among pooled and regression mean squares. For arrays indexed by $r$-element subsets and $n\ge2r$, higher-order U-centering removes all lower-order subset effects and projects onto a space of dimension $\binom{n}{r}-\binom{n}{r-1}$. For fixed symmetric square-integrable kernels evaluated on independent and identically distributed (i.i.d.) paired observations, we derive the residual covariance matrix, exact mean-squared errors for estimating the highest-order Hoeffding components, unbiased positive semidefinite (PSD) covariance estimates, and moment formulas under independence and permutation. The normalized residual inner product equals a direct unbiased estimator involving up to $2r$ observations, but can be evaluated in $O(n^r)$ operations for fixed $r$. For four-argument matching kernels, three matrix products yield exactly unbiased estimators of a fourth spectral moment of kernel cross-covariance in $O(n^3)$ arithmetic and $O(n^2)$ storage. A numerical illustration distinguishes dependence structures with equal HSIC and different spectral concentration.

math.ST↗
arXiv2026

A Partial-Matching Proof of the Neyman--Pearson Lemma

Testing seeks to distinguish distributions, whereas matching seeks to identify their common mass. I show that this apparent opposition provides a direct proof of the Neyman--Pearson lemma. Maximal partial matching between the alternative distribution and a scaled null distribution separates them into a common component and two mutually singular residuals. An elementary dual problem then turns this separation into a testing rule: complementary slackness forces rejection on the alternative residual and non-rejection on the scaled-null residual, leaving freedom to randomize at the likelihood-ratio threshold. Thus the optimal test emerges from maximal matching rather than being proposed in advance. The matching profile also determines the optimal power envelope.

math.ST↗