kSamples: K-Sample Rank Tests and their Combinations

Compares k samples using the Anderson-Darling test, Kruskal-Wallis type tests with different rank score criteria, and Steel's multiple comparison test. It computes asymptotic, simulated or (limited) exact P-values, all valid under randomization, with or without ties, or conditionally under random sampling from populations, given the observed tie pattern. Except for Steel's test it also combines these tests across several blocks of samples. Also analyzed are 2 x t contingency tables and their blocked combinations using the Kruskal-Wallis criterion. Steel's test is inverted to provide simultaneous confidence bounds for shift parameters. A plotting function shows that asymptotic approximations for small sample sizes provide only very limited utility when comparing tail probabilities with those obtained via simulation or exact calculations.

Version: 1.0
Depends: SuppDists
Published: 2012-08-31
Author: Fritz Scholz and Angie Zhu
Maintainer: Fritz Scholz <fscholz at u.washington.edu>
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
NeedsCompilation: yes
CRAN checks: kSamples results

Downloads:

Reference manual: kSamples.pdf
Package source: kSamples_1.0.tar.gz
OS X binary: kSamples_1.0.tgz
Windows binary: kSamples_1.0.zip