[R] A regression problem using dummy variables

Bert Gunter gunter.berton at gene.com
Thu Jul 3 16:21:04 CEST 2008


Contrasts for categorical variables are a technical issue in linear models.
They are best understood (imo) as the choice of bases for the vector space
of the model matrix. So how good is your linear algebra? A terse but I think
very clear treatment of these issues can be found in the relevant chapter of
V&R's MASS book.

FWIW, I have found that even statisticians get confused by these issues
(certainly I have!).

-- Bert Gunter

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On
Behalf Of Ron Michael
Sent: Thursday, July 03, 2008 2:19 AM
To: rlearner309; Peter Dalgaard
Cc: r-help at r-project.org
Subject: Re: [R] A regression problem using dummy variables

"which group contrasts you want to look at" can you clarify me on that
statement?

--- On Thu, 3/7/08, Peter Dalgaard <p.dalgaard at biostat.ku.dk> wrote:



More information about the R-help mailing list