ace.test: ACE permutation test of independence

In acepack: ACE and AVAS for Selecting Multiple Regression Transformations

ACE permutation test of independence

Description

Performs a permutation test of independence or association. The alternative hypothesis is that x and y are dependent.

Code authored by Bernhard Klar, Shawn Garbett.

Usage

acetest(x, y = NULL, nperm = 999, ...)

## S3 method for class 'acetest'
summary(object, ..., digits)

## S3 method for class 'acetest'
print(x, ...)

## S3 method for class 'acetest'
plot(
  x,
  acol = "blue",
  xlim = c(min(x$tp), max(c(x$tp, ceiling(x$ace * 10)/10))),
  col = "black",
  breaks = 100,
  main = "ACE Correlation Permutations",
  xlab = bquote(rho(.(x$xname), .(x$yname))),
  lwd = 2,
  ...
)

Arguments

x	a numeric vector, or a matrix or data frame with two columns. The first column is the 'y' and the second column is the 'x' when calling ace.
y	a vector with same length as x. Default is NULL.
nperm	number of permutations. Default is 999.
...	additional arguments to pass to cor.
object	S3 object of test results to dispatch.
digits	Number of significant digits to round for summary.
acol	for plot; color of the point estimate of correlation
xlim	for plot;xlimit of histogram
col	for plot;color of histogram bars
breaks	for plot;number of breaks. Default to 100.
main	for plot; main title of plot
xlab	for plot; x-axis label
lwd	for plot; line width of point estimate

Value

a list containing the following:

• ace The value of the test statistic.

• pval The *p*-value of the test.

References

Holzmann, H., Klar, B. 2025. "Lancaster correlation - a new dependence measure linked to maximum correlation". Scandinavian Journal of Statistics. 52(1):145-169 <doi:10.1111/sjos.12733>

See Also

cor

Examples

n <- 200
z <- matrix(rnorm(2*n), n) / sqrt(rchisq(n, 2)/2)
x <- z[,2]; y <- z[,1]
cor.test(x, y, method="spearman")
acetest(x, y)

plot(acetest(z))

acepack documentation built on April 4, 2025, 2:06 a.m.