Skip to content

anova table

4 messages · George A. Dowding, Kaspar Pflugshaupt, Douglas Bates

Original
George A. Dowding
# 
I am taking a stats class and need to do a single factor anova table 
with four treatments for an assignment.  The instructor showed the 
class how to do it with minitab, and I am unsure how to proceede in
R. 
 
Here is the data: 
 
I.   .8 .2 1.4 .9 .9 1.0 1.0 1.1 1.4 1.2 1.0 
II.  1.6 1.1 .2 1.3 1.5 .7 .5 1.3 1.3 .4 
III  1.8 .3 1.1 2.0 2.8 1.5 1.3 2.9 1.7 .8 
IV.  2.4 2.6 2.82.9 1.7 2.6 2.7 2.1 2.1 2.3 
 
I have been reading the various help files and can't quite finde the 
right example to get started from.
Kaspar Pflugshaupt
#
On Wednesday 04 April 2001 07:20, George A. Dowding wrote:
Try help(aov). Basically, you will have to  bring your data into a form like

treatment	response
1		.8
1		.2
2		1.6
2		1.1
...		...

and then use something like

model <- aov(response~treatment)
summary(model)

This will give you a standard ANOVA table.

Cheers (and good luck with the assignment)

Kaspar
Douglas Bates
# 
Kaspar Pflugshaupt <pflugshaupt at geobot.umnw.ethz.ch> writes:
If the data are in the form of separate rows for each group, you could
read the data in as one column and use the gl function to generate the
treatment column.  If the data are stored as four columns in a table
then you could use read.table to read the data into a data frame and
the stack function to create one long response vector and the
corresponding treatment column.  (The name of the stack function is
taken from the similarly named command in Minitab.)

The simplest way of doing this is to bring in each level's data as a
separate vector and use stack.  That would look like
Error in data.frame(values = unlist(unname(x)), ind = factor(rep(names(x),  : 
	arguments imply differing number of rows: 41, 0
values ind
1     0.8  I.
2     0.2  I.
3     1.4  I.
4     0.9  I.
5     0.9  I.
6     1.0  I.
7     1.0  I.
8     1.1  I.
9     1.4  I.
10    1.2  I.
Call:
   aov(formula = values ~ ind, data = dat)

Terms:
                     ind Residuals
Sum of Squares  14.05771  10.57009
Deg. of Freedom        3        37

Residual standard error: 0.5344887 
Estimated effects may be unbalanced

Notice that you can save the result of the aov function
and do things like plot it to get diagnostic plots.
If you need to do multiple comparisons of means, use the Tukey
function from the Devore5 package.  That package provides the data and
sample analyses from an introductory text for Engineering Statistics.

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
George A. Dowding
# 
First, thanks, that worked perfectly.  I entered the data and read it
from a file using read.table.  You read my mind with the suggestion of
using Tukey's procedure. We are actually using the  book by Devore.
I am also supposed to do a contrast using a "method attributed to
Scheff\'e." to quote Devore.

Douglas Bates writes:
 > Kaspar Pflugshaupt <pflugshaupt at geobot.umnw.ethz.ch> writes:
 >
> > On Wednesday 04 April 2001 07:20, George A. Dowding wrote:
> > > I am taking a stats class and need to do a single factor anova table
 > > > with four treatments for an assignment.  The instructor showed the
 > > > class how to do it with minitab, and I am unsure how to proceede in
 > > > R.
 > > >
 > > > Here is the data:
 > > >
 > > > I.   .8 .2 1.4 .9 .9 1.0 1.0 1.1 1.4 1.2 1.0
 > > > II.  1.6 1.1 .2 1.3 1.5 .7 .5 1.3 1.3 .4
 > > > III  1.8 .3 1.1 2.0 2.8 1.5 1.3 2.9 1.7 .8
 > > > IV.  2.4 2.6 2.82.9 1.7 2.6 2.7 2.1 2.1 2.3
 > > >
 > > > I have been reading the various help files and can't quite finde the
 > > > right example to get started from.
 > > 
 > > Try help(aov). Basically, you will have to  bring your data into a form like
 > > 
 > > treatment	response
 > > 1		.8
 > > 1		.2
 > > 2		1.6
 > > 2		1.1
 > > ...		...
 > > 
 > > and then use something like
 > > 
 > > model <- aov(response~treatment)
 > > summary(model)
 > > 
 > > This will give you a standard ANOVA table.
 > 
 > If the data are in the form of separate rows for each group, you could
 > read the data in as one column and use the gl function to generate the
 > treatment column.  If the data are stored as four columns in a table
 > then you could use read.table to read the data into a data frame and
 > the stack function to create one long response vector and the
 > corresponding treatment column.  (The name of the stack function is
 > taken from the similarly named command in Minitab.)
 > 
 > The simplest way of doing this is to bring in each level's data as a
 > separate vector and use stack.  That would look like
 > 
 > > grp1 <- c(.8, .2, 1.4, .9, .9, 1.0, 1.0, 1.1, 1.4, 1.2, 1.0)
 > > grp2 <- c(1.6, 1.1, .2, 1.3, 1.5, .7, .5, 1.3, 1.3, .4)
 > > grp3 <- c(1.8, .3, 1.1, 2.0, 2.8, 1.5, 1.3, 2.9, 1.7, .8)
 > > grp4 <- c(2.4, 2.6, 2.8, 2.9, 1.7, 2.6, 2.7, 2.1, 2.1, 2.3)
 > > dat <- stack(list(grp1, grp2, grp3, grp4))
 > Error in data.frame(values = unlist(unname(x)), ind = factor(rep(names(x),  : 
 > 	arguments imply differing number of rows: 41, 0
 > > dat <- stack(list(I. = grp1, II. = grp2, III. = grp3, IV. = grp4))
 > > dat[1:10, ]
 >    values ind
 > 1     0.8  I.
 > 2     0.2  I.
 > 3     1.4  I.
 > 4     0.9  I.
 > 5     0.9  I.
 > 6     1.0  I.
 > 7     1.0  I.
 > 8     1.1  I.
 > 9     1.4  I.
 > 10    1.2  I.
 > > aov(values ~ ind, data = dat)
 > Call:
 >    aov(formula = values ~ ind, data = dat)
 > 
 > Terms:
 >                      ind Residuals
 > Sum of Squares  14.05771  10.57009
 > Deg. of Freedom        3        37
 > 
 > Residual standard error: 0.5344887 
 > Estimated effects may be unbalanced
 > 
 > Notice that you can save the result of the aov function 
 > > fm1 <- aov(values ~ ind, data = dat)
 > and do things like plot it to get diagnostic plots.
 > > plot(fm1)
 > 
 > If you need to do multiple comparisons of means, use the Tukey
 > function from the Devore5 package.  That package provides the data and
 > sample analyses from an introductory text for Engineering Statistics.
 > 
 > -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
 > r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
 > Send "info", "help", or "[un]subscribe"
 > (in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
 > _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._