[R] Model matrix using dummy regressors or deviation regressors

[John Fox]

Dear Peng Yu,

If this is indeed Peng Yu (as the email address seems to imply), I believe
that you have a copy of my R and S-PLUS Companion to Applied Regression,
where the answer is on p. 152, which describes the contrasts argument to lm;
the contrasts argument to aov is similar. Hints: (1) the argument is named
'contrasts' and not 'contrasts.args' (as it is for model.matrix, the help
page for which incidentally has a clear explanation with examples); (2) the
argument takes a named list of contrasts for particular factors not a vector
of contrasts with two elements.

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: pengyu.ut at gmail.com [mailto:pengyu.ut at gmail.com] On Behalf Of
> bluesky315 at gmail.com
> Sent: February-09-10 11:44 PM
> To: John Fox
> Cc: r-help at stat.math.ethz.ch
> Subject: Re: [R] Model matrix using dummy regressors or deviation
regressors
> 
> On Tue, Feb 9, 2010 at 6:09 PM, John Fox <jfox at mcmaster.ca> wrote:
> > Dear bluesky315,
> >
> > There are several ways in R to determine regressors associated with
> factors.
> > One way is to set the global contrasts option. To get "deviation"
> > regressors, use options(contrasts=c("contr.sum", "contr.poly")), and see
> > ?options and ?contrasts for details. Also see Section 11.1.1 of the
> > Introduction to R manual that comes with R.
> 
> Instead of change the option globally, it is cleaner to pass the
> contrast choice as an argument to aov(). It seems that 'contrasts.arg'
> of model.matrix() can be used for this purpose.
> 
> I tried something like contrasts.args=c('contr.sum', 'contr.poly').
> But it is not working. Would you please show me how to specify this
> argument for the example that I gave in the original email?
> 
> > I hope this helps,
> >  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 bluesky315 at gmail.com
> >> Sent: February-09-10 6:33 PM
> >> To: r-help at stat.math.ethz.ch
> >> Subject: [R] Model matrix using dummy regressors or deviation
regressors
> >>
> >> The model matrix for the code at the end the email is shown below.
> >> Since the model matrix doesn't have -1, I think that it is made of
> >> dummy regressors rather than deviation regressors. I'm wondering how
> >> to make a model matrix using deviation regressors. Could somebody let
> >> me know?
> >>
> >> > model.matrix(aaov)
> >>    (Intercept) A2 B2 B3 A2:B2 A2:B3
> >> 1            1  0  0  0     0     0
> >> 2            1  0  0  0     0     0
> >> 3            1  0  0  0     0     0
> >> 4            1  0  0  0     0     0
> >> 5            1  1  0  0     0     0
> >> 6            1  1  0  0     0     0
> >> 7            1  1  0  0     0     0
> >> 8            1  1  0  0     0     0
> >> 9            1  0  1  0     0     0
> >> 10           1  0  1  0     0     0
> >> 11           1  0  1  0     0     0
> >> 12           1  0  1  0     0     0
> >> 13           1  1  1  0     1     0
> >> 14           1  1  1  0     1     0
> >> 15           1  1  1  0     1     0
> >> 16           1  1  1  0     1     0
> >> 17           1  0  0  1     0     0
> >> 18           1  0  0  1     0     0
> >> 19           1  0  0  1     0     0
> >> 20           1  0  0  1     0     0
> >> 21           1  1  0  1     0     1
> >> 22           1  1  0  1     0     1
> >> 23           1  1  0  1     0     1
> >> 24           1  1  0  1     0     1
> >> attr(,"assign")
> >> [1] 0 1 2 2 3 3
> >> attr(,"contrasts")
> >> attr(,"contrasts")$A
> >> [1] "contr.treatment"
> >>
> >> attr(,"contrasts")$B
> >> [1] "contr.treatment"
> >>
> >>
> >>
> >> #############
> >> a=2
> >> b=3
> >> n=4
> >> A = rep(sapply(1:a,function(x){rep(x,n)}),b)
> >> B = as.vector(sapply(sapply(1:b, function(x){rep(x,n)}),
> >> function(x){rep(x,a)}))
> >> Y = A + B + rnorm(a*b*n)
> >>
> >> fr = data.frame(Y=Y,A=as.factor(A),B=as.factor(B))
> >> aaov=aov(Y ~ A * B,fr)
> >> summary(aaov)
> >> model.matrix(aaov)
> >>
> >> ______________________________________________
> >> 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.
> >
> >
> >