hc.test: Higher Criticism

In ThomasTaus/omnibus: Omnibus Test of Global Null Hypothesis

Description

hc-test computes p-values under the global null hypothesis using Higher Criticism.

Usage

hc.test(p, tuning = c("half", "halfmin", "all"), N.sim = 10000)

Arguments

p	numeric vector or matrix of input p-values. In the case of a matrix, rows correspond to individual hypotheses.
tuning	character string specifying the tuning parameter (see 'Details'). Possible values are 'half', 'halfmin' or 'all'.
N.sim	numeric indicating the number of simulations that should be performed under H0.

Details

After sorting p-values in ascending order the Higher Criticism statistic is computed depending on the tuning parameter defined by tuning:

half

only the most extreme half of p-values are used.

halfmin

similar to 'half' but also excluding the most extreme p-value.

all

all p-values are used.

Value

A numeric vector of p-values under the global null hypothesis, with length of 1 if p is a vector or length equal ncol(p) if p is a matrix.

Author(s)

Thomas Taus and Andreas Futschik

References

Tukey J.W.: T13: N the higher criticism. Course nodes, Statistics 411, Princeton University 1976.

Donoho D. and Jin J.: Higher criticism for detecting sparse heterogeneous mixtures, Ann. Stat. 2004, 32:962-994.

Donoh D. and Jin J.: Higher criticism for large-scale inference: especially for rare and weak effects, Stat. Sci. 2015, 30:1-25.

See Also

omnibus.test.

ThomasTaus/omnibus documentation built on May 30, 2019, 3:01 p.m.