[R] how to analyze the following data?--anxious for the result

[Prof Brian Ripley]

Asserting `y isn't noramal' only eliminates one possibility.  Is it 
binary, a count, long-tailed, short-tailed, ...?

This is really a statistical consulting problem (and apparently about 
SAS): see the R posting guide for where to seek advice on such problems.
As the guide says, had we had sufficient information we might have been 
able to help, but as you need to ask again, please do so in a more 
appropriate place.

On Thu, 8 Jun 2006, zhijie zhang wrote:

> Dear friends,
> I have a dataset: response var--y, class var-group, and the third variable-x.
> I want to test whether there is statistical significance bewteen
> group for y with the controlled x. First, i want to use analysis of
> covariance in SAS, but i found that y isn't noramal and can't become
> normal through transformation.
> Under that condition, what should i do using R / SAS?
> Any suggestions are great appreciate!
>

-- 
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