On 15 Jun 2003, Mohammad Ehsanul Karim wrote in part:
> And, is it possible to provide example (at least hypothetical or by
> using any software packages) to each of these specific assumptions?
> to which I had replied:
> > Not clear what you're asking. Do you want data for which a given
> > assumption is explictly met, or explicitly violated? Or, do you
> > want to know what software packages deal, explicitly or otherwise,
> > with a given assumption?
>
> Sorry for my unclear request. I meant how I might demonstrate the
> violation of each assumption by using any software with examples.
If I understand you correctly, this may be impossible. I construe your
request to mean that you would like to be able to deal with "violation
of assumption" in a very general way. But it is not possible to tell
whether an assumption is violated or not, in any empirical way. The
best anyone has been able to do has been to generate data in an
artificial manner ("Monte Carlo study"), where one or more specific
assumptions have been violated in one or more very specific ways.
(For example, the assumption of homogeneity of variance in ANOVA
contexts has been addressed by generating random data for which the
(known) ratios of variances in the several cells of the design have
been, say, 3 or 1 or 1/3, and for which the ratios of numbers of cases
have been similarly manipulated. But such studies do not address either
(a) violations of other assumptions (e.g., that the error distributions
are Gaussian) or (b) other levels of violation of the assumptions
actually considered. I very much doubt whether similar kinds of studies
have been undertaken in the context of multiple regression; but if I am
in error on this point, surely one or more colleagues on this listserv
will be kind enough to point it out!)
I any case, it appears to me unlikely in the extreme that the general
kind of treatment that I presume you are seeking exists.
-----------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
