[R] categorical variable in scatterplot (car)

[John Fox]

Hi Peter,

Thanks for the example. It was previously pointed out to me (in a private
message) that the legend wasn't printed by scatterplot() when the groups
variable isn't numeric. In fixing that, I also took care of the problem that
you noticed. The fix is (or actually will be shortly) in the development
version of the car package on R-Forge, which will be released sometime this
spring or summer.

Thanks again,
 John


> -----Original Message-----
> From: Peter Ehlers [mailto:ehlers at ucalgary.ca]
> Sent: April-25-10 1:20 PM
> To: John Fox
> Cc: 'Anthony Lopez'; R-help at r-project.org
> Subject: Re: [R] categorical variable in scatterplot (car)
> 
> Hi John,
> 
> The problem seems to be with the order in which the 'levels' of
> the conditioning variable appear. Here's a reproducible example:
> 
>   Prestige$tp<- with(Prestige, ifelse(type == "prof", 0, 1))
>   scatterplot(prestige ~ income | tp, data=Prestige)
> 
> Note that I've just switched the 0/1 from your example.
> 
> A quick look at scatterplot.formula suggests that wrapping
> the 'X[, 3]' in this line
> 
>     scatterplot(X[, 2], X[, 1], groups = X[, 3], xlab = xlab,
> 
> inside an as.factor() would solve the problem.
> 
>   -Peter
> 
> 
> On 2010-04-25 8:40, John Fox wrote:
> > Dear Peter and Anthony,
> >
> > Thanks, Peter, for answering the question, but scatterplot() should work
> > even if z is not a factor, and does for me in the following example:
> >
> >> library(car)
> >> Prestige$tp<- with(Prestige, ifelse(type == "prof", 1, 0))
> >> scatterplot(prestige ~ income | tp, data=Prestige)
> >
> > So, Anthony, the usual advice about providing a reproducible example
seems
> > applicable here.
> >
> > Regards,
> >   John
> >
> > --------------------------------
> > John Fox
> > Senator William McMaster
> >    Professor of Social Statistics
> > Department of Sociology
> > McMaster University
> > Hamilton, Ontario, Canada
> > web: socserv.mcmaster.ca/jfox
> >
> >
> >> -----Original Message-----
> >> From: r-help-bounces at r-project.org
[mailto:r-help-bounces at r-project.org]
> > On
> >> Behalf Of Peter Ehlers
> >> Sent: April-24-10 11:57 PM
> >> To: Anthony Lopez
> >> Cc: R-help at r-project.org
> >> Subject: Re: [R] categorical variable in scatterplot (car)
> >>
> >> On 2010-04-24 21:30, Anthony Lopez wrote:
> >>> Hello R folks,
> >>>
> >>> I am encountering a problem with the following scatterplot function
from
> >> the
> >>> car package:
> >>>
> >>>> scatterplot(y~x|z)
> >>>
> >>> where y and x are continuous (interval) random variables and z is a
> >>> categorical variable.  When z is a categorical variable coded 1 or 2,
I
> >>> (appropriately) get a scatterplot of y by x, coded by z.  Similarly,
> > when z
> >>> is a categorical variable coded 1, 2, or 3, there is again, no
problem.
> >>>    However, when z is a categorical variable coded 0 or 1, the
> > scatterplot
> >>>
> >>>> scatterplot(y~x|z)
> >>>
> >>> is exactly identical to the one generated by
> >>>
> >>>> scatterplot(y~x)
> >>>
> >>> It is not possible that this is due to the fact that there is no
> > difference
> >>> between the categories.  It is as if R doesn't "see" that I want it
> > coded
> >> by
> >>> z.  But this only happens when one of the categories of z is coded "0"
> >> (i.e.
> >>> zero).  Any ideas why this is so, or how I can fix this without
recoding
> > my
> >>> variable?
> >>
> >> Make z a factor (which it really should be anyway).
> >>
> >>    -Peter Ehlers
> >>
> >>>
> >>> Thank you!
> >>>
> >>> Anthony
> >>>
> >>> 	[[alternative HTML version deleted]]
> >>>
> >>> ______________________________________________
> >>> 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.
> >>>
> >>>
> >>
> >> --
> >> Peter Ehlers
> >> University of Calgary
> >>
> >> ______________________________________________
> >> 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.
> >
> >
> >
> 
> --
> Peter Ehlers
> University of Calgary