[Rd] contr.sum() and contrast names
jfox at mcmaster.ca
Tue Oct 30 13:08:44 CET 2012
On Tue, 30 Oct 2012 10:25:56 +0100
Milan Bouchet-Valat <nalimilan at club.fr> wrote:
> Le samedi 27 octobre 2012 à 10:44 -0400, John Fox a écrit :
> > Hi Milan,
> > Take a look at the contr.Sum() and contr.Treatment() functions in the
> > car package.
> Yeah, this is the kind of function I had in mind. Just that I think we
> should have an equivalent in the base packages.
Given the history of the issue, I think that it's unlikely that this will happen. Your message hasn't generated a cascade of discussion. And, of course, even if you had provoked a discussion, R Core would have to agree that this kind of change is desirable.
> > (I recall, BTW, the sometimes acrimonious previous discussion of this
> > issue.)
> I could not find this discussion, do you have any pointer?
Sorry. I don't recall when the discussion took place and failed to find it just now when I tried to locate it in the R email list archives.
> Anyways, going deeper and deeper into the archives, I've found you made
> the same proposal almost exactly 10 years ago! Oddly enough, it only
> prompted one reply at the time - has the rest of the discussion been
> removed from the archives because it was too rude? ;-)
I think that you know that the answer to that is "no." And to be clear, the nasty comments that I recall weren't directed at me, and the rudeness was bidirectional (at least insofar as I remember the exchange accurately, which I may not).
> Should we reopen
> the debate now that some time has been spent without actions being taken
> on that front?
You *have* reopened the issue but haven't elicited a response.
A nice characteristic of R is that if you don't like how something is implemented, you can offer your own version, which is what I did with contr.Treatment(), contr.Sum(), and contr.Helmert() in the car package, and similarly with Anova() in the car package. Sometimes these kinds of changes get into the standard R distribution, but more commonly, people who agree with you are free to use the alternatives that you provide.
> 1: http://tolstoy.newcastle.edu.au/R/devel/02b/0878.html
> R-devel at r-project.org mailing list
More information about the R-devel