perMANOVA output

[Jessica Lu]

I've successfully ran perMANOVA through PC-ORD with euclidean distances and understand most of the output.
I have some quick questions about the f-value and the multiple comparisons.

1) is the f value reported in permanova a "pseudo-f" value? similarly in permanovas ran through R and Primer? Do I adjust it in any way, or just report it as a pseudo f?

2) my overall p value wasn't significant, but some of my pairwise comparisons have significant p-value (p<0.05), with the note: "p-values are not corrected for multiple comparisons." does that mean my "significant" p-values in the pairwise comparisons section are not valid?

Thanks in advance for any help!

[Bruce McCune]

1. The "pseudo-f" from perMANOVA is the same in interpretation as an
F ration from regular ANOVA. You can report it as simply F. It needs
no adjustment.

2. This is a harder one, and the answer depends on the way in which
you pose your problem and what kind of answer you are seeking. You
are right to be on guard about this -- it is easy to get a
non-significant overall F, but to have significant pairwise Fs --
this is the classical problem with doing a bunch of post-hoc comparisons.

That quote you gave doesn't mean your pairwise comparisons are not
valid, but you might consider using a Bonferroni procedure to
calculate a new cutoff value for p, based on an experimentwise error
rate. The tradeoff is that as you make it less likely to make a type
I error, you are making it more likely to make a type II error.
Depending on what you are doing, one kind of error or the other might
be less desirable.

Others should feel free to chip in on this. There are lots of
viewpoints out there!

-Bruce McCune