R help - Apr 2007 - Difficulties Using glht.mmc to Calculate Tukey Intervals for Means

Greetings,

 

In the following one-way ANOVA I am attempting to calculate the means of
each treatment along with their 95% Tukey confidence intervals for the data
shown below using a routine from the HH package.

 

library(HH)

options(digits=10)

 

# load data

 


treat

voltage


1

130


1

74


1

155


1

180


2

150


2

159


2

188


2

126


3

138


3

168


3

110


3

160


4

34


4

80


4

40


4

75


5

136


5

106


5

122


5

115


6

174


6

150


6

120


6

139


7

20


7

82


7

70


7

58


8

25


8

58


8

70


8

45


9

96


9

82


9

104


9

60

 

treat <- factor(treat)

 

fit <- aov(voltage ~ treat)

summary(fit)

 

fit.mmc <- glht.mmc(fit,linfct = mcp(treat = "Tukey"))

fit.mmc

 

The problem I am having is that the confidence intervals are different every
time I run the glht.mmc routine. Is this behavior the result of something I
am doing wrong?

 

Thanks

 

Tom La Bone

 

 

 

 

 


	[[alternative HTML version deleted]]

The behavior you are seeing is part of the design of the
summary.glht function in the multcomp package.
There is some randomness in the adjustment for multiplicity.
See
library(multcomp)
?summary.glht
?adjusted