Beta divercity and p value

Hi all,

I have a problem with beta diversity analysis - looking at the obtained graphs, you can see large differences between the groups, but the p-value is high and the results are statistically insignificant. Is it even possible? Have I done something wrong?

bray_curtis_distance_matrix.qzv (373.7 KB)
bray_curtis_emperor.qzv (847.6 KB)

Any info would be much appreciated.

Thank you for any help you can offer!


Hi @Paulina_Srednicka,

I see a large overall difference between groups. However, as you get into the pairwise comparisons with penalties, your sample sizes are small and you're paying a high penalty for multiple hypothesis correction.

At 6 samples (3 per group) your permutations pace is limited to 6! = 720, which is actually less than the 999 permutations you specified in our overall model.

Very few techniques will work with fewer than 5 samples per group. Permanova (beta diversity) kruskal wallis (alpha diversity), and differential abundance with ANCOM are all based on tests that are recommended with at least 5 samples per group (more is ideal).



Thank you very much :slight_smile:

1 Like

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