LRTTBoot: The likelihood ratio parametric bootstrap with trace test...

In TVMM: Multivariate Tests for the Vector of Means

Description

An parametric bootstrap alternative test version of the Likelihood ratio test (LRT) to test the hypotheses about a vector of population averages. This test has the advantage of being valid for high dimension data (n <p). This test should be preferred by the user, as it controlled the type I error and had greater power in all scenarios evaluated.

Usage

LRTTBoot(X, mu0, B)

Arguments

X	a matrix n x p containing n observations and p variables. It should not contain missing values (NA).
mu0	a vector containing the mean population to be tested.
B	the number of resamples bootstrap parametric which must be at least equal to 2000.

Value

the numerical value and the p-value of the test statistic.

Examples

set.seed(0)
library(MASS)
n <- 30
p <- 2
rho <- 0.9
delta <- 0.9
mu <- rep(0, times  = p)
Sigma <- (1 - rho) * diag(p) + rho * matrix(1, p, p)
mu0 <- rep(0.3271,times = p)
B <- 200
X <- mvrnorm(n, mu, Sigma)
LRTTBoot(X=X, mu0=mu0, B=B)

TVMM documentation built on Dec. 15, 2020, 5:33 p.m.