...sorry, i was slow to grasp that. of course it is the same as for other
stratified sampling designs where i have to care for not to mix values of
possibly very different ranges when doing a permutation test..
if i've got it right, staying with the before-after design, this should give
2^n possible permutations.
concluding, given a large enough nr. of possible permutations, i suppose one
should be able to this kind of analysis with adonis - would you agree?
yours,
kay
-----
------------------------
Kay Cichini
Postgraduate student
Institute of Botany
Univ. of Innsbruck
------------------------