[R] se.contrast

[Prof Brian Ripley]

On Wed, 27 Oct 2004, I M S White wrote:

> After a one-way anova, se.contrast computes the standard error of a
> contrast, but not the value of the contrast itself. Wouldn't this be
> useful? Am I missing something?

Compatibility with S, plus `does what it says on the box'.  It was written
to make some of the MASS code work under R.

If you want the contrast and its se, try estimable() in package gmodels
(in gregmisc).

se.contrast would be more useful if anyone ever wrote an aovlist method.
(But, 6 years later, I still have not, and rarely missed it.)

-- 
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