Skip to contents

Angular Randomisation Test of Homogeneity

Source: R/stat_tests.R
ar_test.Rd

Performs Angular Randomisation Test for homogeneity on two samples of circular data after Ruxton et al. (2023). Test whether samples of directions share the mean direction P-values are estimated using permutation.

Usage

ar_test(x, y, n_perm = 1000L, axial = FALSE, alpha = NULL)

Arguments

x, y

numeric vectors. Angles in degrees

n_perm

integer. Number of permutations

axial

logical. Whether the data are axial, i.e. \(\pi\)-periodical (TRUE) or directional, i.e. \(2 \pi\)-periodical (FALSE, the default).

alpha

(optional) numeric. Significance level of the test (values between 0 and 1).

Value

A list with class "htest" containing the following components:

statistic

the value of the G statistic

parameter

the number of data in x and y, and the number of permutations

p.value

the p-value for the test

alternative

a character string describing the alternative hypothesis.

method

a character string indicating what type of test was performed.

alpha

the significance level

reject

logical. whether to reject the null-hypothesis based on the significance level.

data.name

a character string giving the name(s) of the data.

Details

Null Hypothesis (\(H_0\)): The two circular samples share an identical underlying probability distribution.

Alternative Hypothesis (\(H_{1}\)): The two samples come from different distributions.

Interpretation

  • Small p-value (\(p < \alpha\), e.g., <0.05): Reject the null hypothesis. This indicates strong evidence that the two samples come from different circular distributions (differing in central tendency/mean direction or shape).

  • Large p-value (\(p \ge \alpha\)): Fail to reject the null hypothesis; there is insufficient evidence to claim the two circular samples differ.

Note

Concentration Differences: The test can suffer from markedly lower statistical power if the underlying unimodal distributions differ by concentration (dispersion/spread) rather than location—especially with small, uneven sample sizes where the smaller sample comes from the more concentrated distribution.

Axial/Multimodal Data: ART performs poorly and loses power when applied to axially symmetric or symmetrically multimodal distributions.

References

Ruxton, G.D., Malkemper, E.P. & Landler, L. Evaluating the power of a recent method for comparing two circular distributions: an alternative to the Watson U2 test. Sci Rep 13, 10007 (2023). https://doi.org/10.1038/s41598-023-36960-1

See also

Other Two-Sample-Tests: equal_kappa_test(), uniform_scores_test(), walraff_test(), watson_two_sample, watson_wheeler_test_perm(), watson_williams_test()

Examples

set.seed(20250411)
x1 <- c(35, 45, 50, 55, 60, 70, 85, 95, 105, 120)
x2 <- c(75, 80, 90, 100, 110, 130, 135, 140, 150, 160, 165)
ar_test(x1, x2)
#> 
#> 	Angular Randomisation Test of Homogeneity
#> 
#> data:  x1 and x2
#> G = 104.37, nx = 10, ny = 11, n_perm = 1000, p-value = 0.003996
#> alternative hypothesis: Samples come from different distributions
#> 

# San Andreas Fault Data:
data(san_andreas)
data("nuvel1")
PoR <- subset(nuvel1, nuvel1$plate.rot == "na")
sa.por <- PoR_shmax(san_andreas, PoR, "right")
ar_test(sa.por$azi.PoR, rvm(100, 135, 10), axial = TRUE, alpha = 0.05)
#> 
#> 	Angular Randomisation Test of Homogeneity
#> 
#> data:  sa.por$azi.PoR and rvm(100, 135, 10)
#> G = 92652, nx = 1126, ny = 100, n_perm = 1000, p-value = 0.988
#> alternative hypothesis: Samples come from different distributions
#>