pdf("g06.pdf");options(width=64)
#setwd("C:\\Users\\kolassa\\Class555")
setwd("~/Taught1/960-555/Data")
<pre>
#********************************************************/
#* Yarn strength data from Example Q of Cox & Snell     */
#* (1981).  Variables represent strength of two types of*/
#* yarn collected from six different bobbins.           */
#********************************************************/
yarn<-as.data.frame(scan("yarn.dat",what=
    list(strength=0, bobbin=0,type="")))
#<a name=block1> Block 1</a>
yarna<-yarn[yarn$type=="A",]
cat('\nMultiple Comparisons for Yarn with No Correction\n')
pairwise.wilcox.test(yarna$strength,yarna$bobbin,exact=F,
  p.adjust.method="none")
#<a name=block2> Block 2</a>
library(MultNonParam)#For higgins.fisher.kruskal.test
higgins.fisher.kruskal.test(yarna$strength,yarna$bobbin)
#<a name=block3> Block 3</a>
# pairwise.wilcox.test corrects for multiple comparisons
# methods using only on p-values.  This requirement 
# excludes methods of Tukey and Scheffe.
cat('\nBonferroni Comparisons for Yarn Type A Data\n')
pairwise.wilcox.test(yarna$strength,yarna$bobbin,
  exact=F,p.ajust.method="bonferroni")
#<a name=block4> Block 4</a>
library(MultNonParam)#For tukey.kruskal.test
tukey.kruskal.test(yarna$strength,yarna$bobbin)
# The data at HTTP://lib.stat.cmu.edu/datasets/Andrews/T58.1
# represent corn (maize) yields resulting from various fertilizer 
# treatments.  Test the null hypothesis that corn weights 
# associated with various fertilizer combinations have the same 
# distribution, vs. the alternative hypothesis that a measure of 
# location varies among these groups.  Treatment is a three-digit 
# string representing three fertilizer components.  Fields in this 
# file are separated by space.  The first three fields are example, 
# table, and observation number.  The fourth and following fields 
# are location, block, plot, treatment, ears of corn, and corn weight.
# Location TEAN has no ties; restrict attention to that location.
# The yield for one of the original observations 36 was missing 
# (denoted by -9999 in the file), and is omitted in this analysis.
# We calculate the Kruskal-Wallis test, with 12 groups, and hence 
maize<-as.data.frame(scan("T58.1",what=list(exno=0,tabno=0, 
   lineno=0,loc="",block="",plot=0,trt="",ears=0, wght=0)))
maize$wght[maize$wght==-9999]<-NA
maize$nitrogen<-as.numeric(substring(maize$trt,1,1))
#<a name=block5> Block 5</a>
library(clinfun)# For jonckheere.test
maize$wght[maize$wght==-9999]<-NA
#Location TEAN has no tied values.  R treats ranks of
#missing values nonintuitively.  Remove missing values.
maize$nitrogen<-as.numeric(substring(maize$trt,1,1))
tean<-maize[(maize$loc=="TEAN")&(!is.na(maize$wght)),]
# Perform the Jonckheere-Terpstra test
jonckheere.test(tean$wght,tean$nitrogen)
cat('\n K-W Test for Maize, to compare with JT  \n')
kruskal.test(tean$wght,tean$nitrogen)
#<a name=block6> Block 6</a>
library(MultNonParam)#For terpstrapower, kwpower, kwsamplesize, kweffectsize
terpstrapower(rep(20,3),(0:2)/2,"normal")
#<a name=block7> Block 7</a>
kwpower(rep(20,3),(0:2)/2,"normal")
#<a name=block8> Block 8</a>
kwsamplesize((0:2)/2,"normal")
#<a name=block9> Block 9</a>
kwsamplesize((0:2)/2,"normal",taylor=TRUE)
#<a name=block10> Block 10</a>
kweffectsize(60,(0:2)/2,"normal")
#<a name=block11> Block 11</a>
library(MultNonParam)#For powerplot
#powerplot uses Monte Carlo, and the default
#sample size is large enough to make it slow.
powerplot()
</pre>