[R] homogenity inside groups

[Petr PIKAL]

Yes, that is what I meant. It is not a species but some products and I 
have various parameters measured for each product. But basically I thought 
that ecological data are quite similar.

So I would be glad to be able to try your code.

Thank you

Petr Pikal
petr.pikal at precheza.cz

Gavin Simpson <gavin.simpson at ucl.ac.uk> napsal dne 15.11.2007 18:32:08:

> On Thu, 2007-11-15 at 16:07 +0100, Petr PIKAL wrote:
> > Dear all
> > 
> > I would like to show my audience that some variables are homogenous 
inside 
> > groups but different outside. I can use by with summary for all 
variables
> > 
> >  by(iris[,1:4],  iris$Species, summary)
> > 
> > what can be quite messy in case of more than few variables and about 8 

> > groups
> > 
> > or densityplot for one variable
> > 
> > densityplot(~Petal.Length | Species, iris)
> > 
> > I have two questions:
> > 
> > 1.      Is there any other plot to show all variables at once? 
Something 
> > like
> > 
> > densityplot(~iris[,1:4] | Species, iris)
> > 
> > 2.      Is it possible to evaluate homogenity of many (20-30) 
variables 
> > inside groups by some other function/table/graph?
> 
> Hi Petr,
> 
> I haven't replied-all by the way, in case I've misunderstood, but...
> 
> If you mean that you have a data set with say 10 samples split into 2
> groups, and for each sample you have measured many variables (say
> species in a quadrat or lots of morphological parameters on individual
> plants), then one way might be to look at the work of Marti Anderson
> [*]. She has developed a method that calculates the multivariate
> distance between each sample in a group and that group's multivariate
> centroid. You then take these distances to group centroid and do an
> ANOVA on them, the general point being that this is a multivariate
> analogue of something like a Levene's test and if groups variances are
> heterogeneous then one or more groups will have a higher/lower mean
> distance to centroid than the other groups.
> 
> groups <- something.that.gives.groups.as.a.factor()
> dis <- something.that.gives.distances.centroids(my_data, groups)
> anova(lm(dis ~ groups))
> 
> If this is the case, then I have some code that I'm currently working on
> which does this, which works (!) and which I can send to you. Marti
> tests for homogeneity using a permutation test. I have that as well, but
> currently it doesn't give the same results as Marti's PERMDISP2
> programme (standalone Fortran, source not available), though I can't see
> what I'm doing wrong, if anything - and my permutation p-value closely
> matches the ANOVA p-value for tests where the data don't violate ANOVA
> assumptions too grossly - Levene's test is quite robust in this regard.
> 
> Let me know if this is what you meant and if the code will be useful and
> I'll send a reply to the list for the archives and send you my code.
> 
> All the best,
> 
> G
> 
> [*] Anderson, M.J. (2006) Distance-based tests for homogeneity of
> multivariate dispersions. Biometrics 62, 245--253
> 
> http://www.stat.auckland.ac.nz/~mja/Programs.htm
> 
> > 
> > Thank you
> > 
> > Petr Pikal
> > petr.pikal at precheza.cz
> > 
> > ______________________________________________
> > 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.
> -- 
> %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
>  Dr. Gavin Simpson             [t] +44 (0)20 7679 0522
>  ECRC, UCL Geography,          [f] +44 (0)20 7679 0565
>  Pearson Building,             [e] gavin.simpsonATNOSPAMucl.ac.uk
>  Gower Street, London          [w] http://www.ucl.ac.uk/~ucfagls/
>  UK. WC1E 6BT.                 [w] http://www.freshwaters.org.uk
> %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
>