Example R programs and commands
		 9. R commands for multiple-comparison tests

# All lines preceded by the "#" character are my comments.
# All other left-justified lines are my input.
# All other indented lines are the R program output.
#
# Tabulated data for input to R
# 
#       ORIGINAL FORMAT
# Skateboarding injuries (per 1000 board-hours.)
# NOTE: not the same data as in example 8.
#
#  Mar   Jun    Sep  
# 9.7    4.2    9.6 
# 9.8    6.7    9.4 
# 9.6    5.8    8.9 
# 9.5    7.5    9.4 
#        4.4    9.5    
#
#
mar <- c(9.7,9.8,9.6,9.5,NA)   # only 4 values
jun <- c(4.2,6.7,5.8,7.5,4.4)
sep <- c(9.6,9.4,8.9,9.4,9.5)
month <- gl(3,5,labels=c("Mar","Jun","Sep"))
injuries <- c(mar,jun,sep)

# Pairwise t-test:
pairwise.t.test(injuries,month)

	   Pairwise comparisons using t tests with pooled SD 

   data:  injuries and month 

       Mar     Jun    
   Jun 0.00011 -      
   Sep 0.63313 0.00011

   P value adjustment method: holm 



# Tukey multiple-comparison test:
TukeyHSD(aov(injuries~month))

     Tukey multiple comparisons of means
       95% family-wise confidence level

   Fit: aov(formula = injuries ~ month)

   $month
	    diff       lwr       upr     p adj
   Jun-Mar -3.93 -5.525460 -2.334540 0.0000978
   Sep-Mar -0.29 -1.885460  1.305460 0.8770268
   Sep-Jun  3.64  2.135786  5.144214 0.0001146

# Similar computation with the sample means reordered:
TukeyHSD(aov(injuries~month), ordered=TRUE)

     Tukey multiple comparisons of means
       95% family-wise confidence level
       factor levels have been ordered

   Fit: aov(formula = injuries ~ month)

   $month
	   diff       lwr      upr     p adj
   Sep-Jun 3.64  2.135786 5.144214 0.0001146
   Mar-Jun 3.93  2.334540 5.525460 0.0000978
   Mar-Sep 0.29 -1.305460 1.885460 0.8770268

# Nonparametric multiple comparison
pairwise.wilcox.test(injuries,month)

	   Pairwise comparisons using Wilcoxon rank sum test

   data:  injuries and month

       Mar   Jun
   Jun 0.036 -
   Sep 0.063 0.036

   P value adjustment method: holm
   Warning messages:
   1: cannot compute exact p-value with ties in: wilcox.test.default(xi, xj, ...)
   2: cannot compute exact p-value with ties in: wilcox.test.default(xi, xj, ...)