[banner]

An R Companion for the Handbook of Biological Statistics

Salvatore S. Mangiafico

Mann–Whitney and Two-sample Permutation Test

The Mann–Whitney U-test is a nonparametric test, also called the Mann–Whitney–Wilcoxon test.  It tests for a difference in central tendency of two groups, or, with certain assumptions, for the difference in medians.  It is conducted with the wilcox.test function in the native stats package.  It can be used with continuous or ordinal measurements.

 

As another non-parametric alternative to t-tests, a permutation test can be used.  An example is shown in the “Permutation test for independent samples” section of this chapter.

 

Mann–Whitney U-test

 

### --------------------------------------------------------------
### Mann–Whitney U-test, biological data analysis class, pp. 128–129
### --------------------------------------------------------------

Input =("
Group Value
2pm    69   
2pm    70   
2pm    66   
2pm    63   
2pm    68   
2pm    70   
2pm    69   
2pm    67   
2pm    62   
2pm    63   
2pm    76   
2pm    59   
2pm    62   
2pm    62   
2pm    75   
2pm    62   
2pm    72    
2pm    63   
5pm    68
5pm    62
5pm    67
5pm    68
5pm    69
5pm    67
5pm    61
5pm    59
5pm    62
5pm    61
5pm    69
5pm    66
5pm    62
5pm    62
5pm    61
5pm    70
")

Data = read.table(textConnection(Input),header=TRUE)

 

Box plots

 

boxplot(Value ~ Group,
        data = Data,
        names=c("2 pm","5 pm"),
        ylab="Value")

 

 

 

 

wilcox.test(Value ~ Group, data=Data)

 

Wilcoxon rank sum test with continuity correction

 

W = 186, p-value = 0.1485

 

#     #     #

 

 

Permutation test for independent samples

Permutation tests are nonparametric tests, and can be performed with the coin package.  The permutation test compares values across groups, and can also be used to compare ranks or counts.  This test is analogous to a nonparametric t-test.   Normality is not assumed but the test may require that distributions have similar variance or shape to be interpreted as a test of means.

 

 

### --------------------------------------------------------------
### Two-sample permutation test, biological data analysis class,
###   pp. 128–129
### --------------------------------------------------------------

Input =("
Group Value
2pm    69   
2pm    70   
2pm    66   
2pm    63   
2pm    68   
2pm    70   
2pm    69   
2pm    67   
2pm    62   
2pm    63   
2pm    76    
2pm    59   
2pm    62   
2pm    62   
2pm    75   
2pm    62   
2pm    72    
2pm    63   
5pm    68
5pm    62
5pm    67
5pm    68
5pm    69
5pm    67
5pm    61
5pm    59
5pm    62
5pm    61
5pm    69
5pm    66
5pm    62
5pm    62
5pm    61
5pm    70
")

Data = read.table(textConnection(Input),header=TRUE)

library(coin)

independence_test(Value ~ Group,
                  data = Data)

 

Asymptotic General Independence Test

 

Z = 1.2761, p-value = 0.2019

 

#     #     #