On Tue, 2 Apr 2002 16:24:30 +0200, "Laurence" <[EMAIL PROTECTED]>
wrote:

> Hi,
> 
> Residuals coming from regression.
> 
> I look for a method explaining ho how to make a chi-square on residuals
> (Without ploting anything) to prove or not the normality of the
> distribution.

Theory works in just the one direction: 
there might exist a test like you want 
that gives evidence *against*  the normality of residuals.
You probably have to be more specific about what you want.

If you want a *test* , which is 'fixed' by 
size (alpha) or chi-square value, 
then it is almost assured of 'rejecting normality'
when  the N gets large enough.  Is that okay?

> 
> Programs or algorithms are welcome.
> 

It is my impression that the efficient ones 
will be application-dependent.

You might look up 'standardized residuals'
and review values in your old sets of data. 

Find the single values or sets of values 
that concern you, and then find a rule of thumb
that picks them out (depending on N, if N varies).

-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to