As far as I know, the results of adonis() can be interpreted the same
way as a univariate ANOVA would. The exception being that differences in
(multivariate) location are tested in the space of principal coordinates
associated with a particular distance measure. Consequently, the
distance measure has a strong effect on the results. Also, p-values are
obtained by permutations, such that the only assumption is that the
units are exchangeable under the null hypothesis - no assumption about
normality etc.
Date: Sun, 22 Nov 2009 22:06:13 +0100
From: Roman Lu?trik <roman.lustrik at gmail.com>
Subject: [R-sig-eco] interpreting adonis()
To: r-sig-ecology at r-project.org
Message-ID:
<63a206010911221306s314e6412r905ebddfcb14387b at mail.gmail.com>
Content-Type: text/plain
Can someone nudge me into the right direction on how to interpret
adonis()?
anosim() is fairly simple though its barplot with notches, but mr.
Oksanen
recommends adonis(), which is less intuitive. I hope this is not too
trivial
issue for this list.
Best wishes,
Roman