Skip to contents

Bartlett Test for Homogeneity of Variances of Z-Residual

Source: R/test.var.bartl.R
test.var.bartl.Rd

Performs Bartlett's test to assess whether the variance of Z-residuals differs across levels of a covariate. Numeric covariates with many distinct values are binned, and small or empty bins are removed before testing.

Usage

test.var.bartl(Zresidual, fitted.value, k.bl = 10)

Arguments

Zresidual

Numeric vector of Z-residuals.

fitted.value

Numeric or factor covariate to test against.

k.bl

Integer; the number of bins to discretize a numeric covariate (default 10).

Value

Numeric p-value from Bartlett's test for homogeneity of variances.

Details

The function handles covariates as follows:

  • If fitted.value is a factor or has fewer than k.bl unique values, it is treated as categorical.

  • Otherwise, numeric covariates are binned into k.bl bins.

  • Bins with fewer than 3 observations are removed.

  • If insufficient bins remain, the covariate is log-transformed and binned again.

Bartlett's test is then applied to the Z-residuals grouped by the factor or binned covariate.

Examples

if (FALSE) { # \dontrun{
Zres <- rnorm(100)
x <- runif(100)
test.var.bartl(Zres, x, k.bl = 5)
} # }