R help - Oct 2009 - Plotting 1 covariate, 3 factors

I'm interested in plotting a y with an x factor as the combination of 2
factors and colour with respect to a third, which the code below does with
interaction.plot().  However, this is because I redefine the x to be 1
factor.  Is there a way of getting it to plot without redefining it, and
ideally to not join up the lines BETWEEN levels a and b, but just join those
between after and before for one level of f3.  I figure this could be done
by manually drawing over blank lines using ?lines but am not sure what the
coordinates would be and figured there is probably an easier way where
someone has dealt with this before.  Any thoughts greatly appreciated,

Paul

#####

y<-rnorm(36)
f1<-rep(c("after","before"), 18)
f2<-rep(1:3,12)
f3<-rep(1:2, each=18)

## Define new factor to be f1 and f3 for x axis - clumsy code, but gets its
done;

ff<-numeric(length(y))
for (i in 1:length(y))
{if (f1[i]=="a" & f3[i]==1) ff[i]<-"1, a"
else if(f1[i]=="a" & f3[i]==2) ff[i]<-"2, a"
else if(f1[i]=="b" & f3[i]==1) ff[i]<-"1, b"
else ff[i]<-"2, b"}

## Plot of interest;

interaction.plot(ff,f2,y)
-- 
View this message in context:
http://www.nabble.com/Plotting-1-covariate%2C-3-factors-tp25789442p25789442.html
Sent from the R help mailing list archive at Nabble.com.

## Paul

## I think you are looking for interaction2wt

y <- rnorm(36)
f1 <- rep(c("after", "before"), 18)
f2 <- rep(1:3, 12)
f3 <- rep(1:2, each=18)

## your definition of ff was faulty.  It gave a constant.
f3.f1 <- interaction(f3, f1)
interaction.plot(f3.f1, f2, y)

f2 <- factor(f2)
f3 <- factor(f3)

## lattice would be better
xyplot(y ~ f2 | f3, groups=f2)


## if you don't have HH, you will need to install it and its dependencies
## install.packages("HH")
## You may need to close R and restart R

require(HH)
interaction2wt(y ~ f1 + f2 + f3)

position(f1) <- c(1.5, 2.5)
position(f3) <- c(1.5, 2.5)
interaction2wt(y ~ f1 + f2 + f3, box.ratio=.6)

## Rich

I'm not sure if this is exactly what you are looking for but I would
generally create an interaction plot using the lattice 'dotplot' with
type = c("p","a") so I get both the original data and the
lines
joining the averages for the different factor levels.  I also prefer
the horizontal orientation to the vertical orientation.  Combining all
these variations produces something like

dotplot(f2 ~ y | f1, groups = f3, aspect = 0.2, layout = c(1,2), type
= c("p","a"), pch = 21, strip = FALSE, strip.left = TRUE,
auto.key list(columns = 2, lines = TRUE))

On Wed, Oct 7, 2009 at 11:00 AM, Paul Chatfield <p.s.chatfield at
rdg.ac.uk> wrote:
>
> I'm interested in plotting a y with an x factor as the combination of 2
> factors and colour with respect to a third, which the code below does with
> interaction.plot(). ?However, this is because I redefine the x to be 1
> factor. ?Is there a way of getting it to plot without redefining it, and
> ideally to not join up the lines BETWEEN levels a and b, but just join
those
> between after and before for one level of f3. ?I figure this could be done
> by manually drawing over blank lines using ?lines but am not sure what the
> coordinates would be and figured there is probably an easier way where
> someone has dealt with this before. ?Any thoughts greatly appreciated,
>
> Paul
>
> #####
>
> y<-rnorm(36)
> f1<-rep(c("after","before"), 18)
> f2<-rep(1:3,12)
> f3<-rep(1:2, each=18)
>
> ## Define new factor to be f1 and f3 for x axis - clumsy code, but gets its
> done;
>
> ff<-numeric(length(y))
> for (i in 1:length(y))
> {if (f1[i]=="a" & f3[i]==1) ff[i]<-"1, a"
> else if(f1[i]=="a" & f3[i]==2) ff[i]<-"2, a"
> else if(f1[i]=="b" & f3[i]==1) ff[i]<-"1, b"
> else ff[i]<-"2, b"}
>
> ## Plot of interest;
>
> interaction.plot(ff,f2,y)
> --
> View this message in context:
http://www.nabble.com/Plotting-1-covariate%2C-3-factors-tp25789442p25789442.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>