pm.wilcox.test: Wilcoxon test for Partially Matched Two Sample Data

Description

Performs rank-based two sample test for partially matched two sample data by combining information from matched and unmatched data

Usage

pm.wilcox.test(Xpaired, Ypaired, Xextra = NULL, Yextra = NULL,
 alternative = c("two.sided", "less", "greater"),
 method = c("SR-MW", "MW-MW", "all"), mode = c("test",
 "var", "power.study"), useC = FALSE, correct = NULL,
 verbose=FALSE)

Value

An htest object.

Arguments

Xpaired: Xpaired
Ypaired: Ypaired
Xextra: Xextra
Yextra: Yextra
alternative: alternative
method: String. SR-MW is recommended, all is for development only.
mode: String. Do not change it to var, for developpment only.
useC: Boolean. Do not set it to TRUE, for development only.
verbose: verbose
correct: Continuity correction.

Details

If Xpaired and Ypaired have NAs, the corresponding unpaired data in Ypaired and Xpaired will be combined with Yextra and Xextra.

Examples

Run this code


set.seed(1)
z=rnorm(20, sd=0.5) # induces correlation between X and Y
X=rnorm(20)+z
Y=rnorm(20,mean=0.8)+z
X[1:10]=NA
boxplot(X,Y,names=c("X","Y"))

pm.wilcox.test(X,Y)
# for comparison
wilcox.test(X,Y,paired=TRUE)
wilcox.test(X,Y,paired=FALSE)# often a conservative test due to the correlation

# no paired data
Y1=Y
Y1[11:20]=NA
pm.wilcox.test(X,Y1)
# should match the following
wilcox.test(X,Y1,paired=FALSE)

# only 1 pair of matched data
Y1=Y
Y1[12:20]=NA
pm.wilcox.test(X,Y1)

Run the code above in your browser using DataLab