[R] adonis(), design

[Gavin Simpson]

On Thu, 2010-03-04 at 01:30 -0800, Kay Cichini wrote:
> ....inclusion of repeated measures of course should gain power, but here I
> guess one would have to restrict permutations and that's what may reduce
> power drastically if sample size is small, at least that's how I understood
> it.
> I have also dug out a thread where someone asked for random factors in
> adonis() and that's quite the same topic, I think. Gavin answered that
> function permuted.Index2() would serve - but still I don't know how to do it
> practically..

You can't - without ripping these particular guts out of vegan,
inserting the code from package 'permute', and then hacking adonis to
use the new code. That's quite a lot of user-intervention. Not to
mention that 'permute' is pre-alpha at the moment and full of bugs...

> There is one of my postings in the r forge vegan forum that handles the same
> problem with sampling design for beta.disper() and if it turns out that in
> both cases the key is permuted.Index2() this thread may be redundant.

For the moment, there isn't really a permutation-based solution that
works for your problem within vegan and adonis. We are planning on a
major update to vegan over the summer to interface vegan's ordination
functions with the permutation code in 'permute'. Of course, that is
conditional upon me getting 'permute' working well first.

G

> 
> Greetings,
> Kay
-- 
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
 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
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%