pair.wmw.test: WMW test for paired data

In robustrank: Robust Rank-Based Tests

Description

Performs a WMW-type test of the strong null for paired data.

Usage

pair.wmw.test(X, Y, alternative = c("two.sided", "less", "greater"),
 correct = TRUE, perm = NULL, mc.rep = 10000, method =
 c("exact.2", "large.0", "large", "exact", "exact.0",
 "exact.1", "exact.3"), verbose = FALSE, mode =
 c("test", "var"), p.method = NULL, useC = TRUE)

Arguments

X	Sample 1.
Y	Sample 2.
alternative	Alternative hypothesis.
correct	Whether to apply continuity correction.
perm	Whether to use permutation distribution or normal approximation to find p-value. See details.
mc.rep	Number of Monte Carlo replicates for permutation test.
method	Choices of test statistics.
verbose	Print debug message when positive.
mode	For development used.
useC	For development used.
p.method	Method for obtaining p values.

Details

When perm is NULL, if (min(m,n)>=20) normal approximatino is used to find p value, otherwise permutation test is used. When permutation test is used, if the number of possible permutations is less than mc.rep, a test statistic is computed for all permutations; otherwise, Monte Carlo is done.

Value

P value for now.

References

Under prep.

Examples

dat=sim.partially.matched(m=15,n.x=0,n.y=20,distr="mixnormal",params=c(p.1=0.3,p.2=0.3),seed=1)
X=dat$X; Y=dat$Y
pair.wmw.test(X, Y, perm=TRUE,  method="large.0", verbose=1)
pair.wmw.test(X, Y, perm=FALSE, method="large.0", verbose=1)

robustrank documentation built on July 15, 2020, 5:08 p.m.