[R] Testing significance in a design with unequal but proportional sample sizes
Prof Brian Ripley
ripley at stats.ox.ac.uk
Fri Mar 5 17:40:49 CET 2004
Those are contrasts treating the interaction as a set of four treatments,
not an interaction between two factors. So you would need to set
contrasts on the factor ab <- a:b (outside a formula), with formula
a+b+ab.
On Fri, 5 Mar 2004, Christophe Pallier wrote:
>
>
> Prof Brian Ripley wrote:
>
> >On Fri, 5 Mar 2004, pallier wrote:
> >
> >...
> >
> >
> >
> >>Actually, the different types of main effects defined above just
> >>correspond to different
> >>contrasts on the cell means. So if there is an easy solution to compute
> >>arbitrary contrasts
> >>on the cell means in a factorial design, this could an approach to this
> >>question. (Anyone?)
> >>
> >>
> >
> >There are at least three such ways. ?contrasts (for the assignment
> >function contrasts<-) and ?C, as well as the contrasts= argument to aov
> >(the function you were discussing ...).
> >
> >
> Thanks.
> I know the existence of 'contrasts' and I read the section about
> contrasts matrix in your book (MASS 3rd edition), as well as
> in the R online documentation, but I probably do not understand them
> well: It still escapes me how to proceed to compute
> "arbitrary" contrasts, such as, say:
>
> a1b1 a1b2 a2b1 a2b2
> 1 1 -1 -1
>
> a1b1 a1b2 a2b1 a2b2
> .5 .5 -1 0
>
> in a model "x~ a * b" where a and b are two binary factors.
>
> (the contrasts should be on the cell means, ignoring the sample size of
> subgroups. I know how to compute the size of the contrasts from the
> table of means returned by tapply, but I whould also need the associated
> MSE).
>
> Sorry if the solution is obvious.
>
> Christophe Pallier
>
>
>
>
>
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
More information about the R-help
mailing list