vows (version 0.4)

permF.mp: Permutation F-tests for massively parallel linear models

Description

Performs permutation F-tests for parallel linear models with a common design matrix. Currently restricted to testing with the intercept-only model as the null hypothesis. The permutation method controls the familywise error rate (FWER) at a desired level; see Details.

Usage

permF.mp(formula, nperm = 499, alpha = 0.05, report.every = 50)

Arguments

formula

a formula such as "Y ~ X", where Y is an $n \times V$ response matrix and X is an $n \times p$ design matrix common to all $V$ models.

nperm

number of permutations.

alpha

level at which to control the FWER.

report.every

parameter controlling how often to report the number of permutations performed; by default, every 50.

Value

maxF.perm

maximal F-statistics obtained from each of the permuted data sets.

F.obs

the observed F-statistics.

threshold

critical value obtained from the permutations.

pvalue

adjusted (familywise error rate-controlling) p-values.

Details

The observed F-statistics are referred to a permutation distribution of the maximum F-statistic over all $V$ tests. This is a standard approach to FWER control in neuroimaging (Nichols and Holmes, 2001).

References

Nichols, T. E., and Holmes, A. P. (2001). Nonparametric permutation tests for functional neuroimaging: a primer with examples. Human Brain Mapping, 15, 1--25.

See Also

F.mp

Examples

Y = matrix(rnorm(6000), nrow=20)
X = rnorm(20)
t3 = permF.mp(Y~X)