Generate exchangeable (equicorrelated) Gaussian null data

Draws an n by p matrix of multivariate Gaussian observations with a common mean mu and an exchangeable (equicorrelation) covariance: variance sig^2 on the diagonal and sig^2 * rho off-diagonal. Under this model all observations share the same mean, so it serves as a null for cluster-mean comparisons.

Usage

generate_exchangeable_null_data(n, p, rho = 0, mu = 0, sig = 1)

Arguments

n

Number of observations (rows).

p

Number of variables (columns).

rho

Equicorrelation parameter (off-diagonal correlation).

mu

Common mean added to every entry.

sig

Common standard deviation.

Value

A list with elements dat (the n by p data matrix) and Sig (the p by p covariance matrix used).

Examples

set.seed(1)
out <- generate_exchangeable_null_data(30, 2, rho = 0.1)
dim(out$dat)
#> [1] 30  2