Beta diversity P values

Hello :qiime2: :wave:

When looking at weighted/unweighted UniFrac,
which of P values is the correct one?
the one in the "overview" or at the " Pairwise permanova results" ?

I was comparing two groups, and the statistical tests remained the same but the P values are not the same.

so I was wondering, what is the difference between the P value in the "overview" table and at the bottom table.

Thanks!

Hello @Tina_Khone :wave:

Actually, both of them are correct!

PERMANOVA can give you slightly different p-values each time you are runing it. But 0.05 is an artificial threshold, not an absolute one as well. So if one you got 0.049 and second time 0.051 you still can report is as a significant result, or report everything > 0.05 < 0.1 as a trend towards significance.

In the "overview", you can see a p-value of the test, which was performed on multiple groups. And if the number of groups > 2, all you can say is that they are or there are not significantly distant between groups. To learn, which groups are different, you need a pairwise comparison as a post hoc (other p-values). Pairwise comparisons also should be corrected for multiple testing (q-values).
Usually, if overall p-value is high, no need for pairwise comparisons since it is an indicator that there is no significant variation among tested groups.
If overall p-value <= 0.05 or a little bit higher, but still close enough you can report q-values from pairwise tests.

PS. For only two groups, you can report "overall" p-value.

4 Likes

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.