Those with more formal statistical backgrounds may provide better
advice, but in my own informal training I've come to wonder why
parametric stats persist in the face of modern computing power. As I
understand it, Fisher developed ANOVA as low-computation method of
approximating the Randomization Test (a.k.a. exhaustive permutation
test). Where computation power has grown exponentially since Fisher's
time, these days it is feasible to compute the full R-Test in many
cases, and for those cases where the full R-Test is not feasible,
non-exhaustive permutation variants usually satisfy. Indeed, it has
been shown (http://brm.psychonomic-journals.org/content/37/3/426.short)
that the R-Test is more powerful than the F-Test in the face of skewed
distributions.
My advice would thus be to abandon parametrics and simply code a
randomization test variant of the ANOVA you want.
Mike
On Tue, Mar 31, 2009 at 10:29 AM, Sarti Maurizio <sarti.m at irea.cnr.it>
wrote:> Dear R-Helpers,
>
> Parametric statistics are statistics where the population is assumed to fit
any
> parametrized distributions (most typically the normal distribution).
> My problem is:
> 1) if my polulation is no normal
> 2) if the sample data of all replications and treatments were well fitted
from
> the Weibull distribution (shape and scale parameters).
>
> Can be The shape and scale parameters compared between treatments by using
the
> canonical analysis of the variance ANOVA?
>
> Many thanks for your help with these questions.
>
> Maurizio
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
--
Mike Lawrence
Graduate Student
Department of Psychology
Dalhousie University
Looking to arrange a meeting? Check my public calendar:
http://tinyurl.com/mikes-public-calendar
~ Certainty is folly... I think. ~