Post hoc test for beta diversity?

User Support
1

Hi All,

So I calculated the beta diversity for my experiment. The p-values tell me that all are significantly different. My question is: would it be possible to apply a post-hoc test to rank the treatments based on the distance value so that I can see which has the highest distance and which the lowest distance? Or perhaps this is not possible?

Thank you for your support.

Cheers,
Pablo

Screenshot
Screenshot1280×1024 91.8 KB

2

This visualizer :point_down: will compare all groups report which pairs are significantly different:
https://docs.qiime2.org/2019.10/plugins/available/diversity/beta-group-significance/

You can sort by distance using your the means you have in that boxplot. Is that what your asking about?

Colin

3

Hi Colin,

Thanks a lot for your response! I have indeed used the bega group significance plug-in and I obtained pairwise comparisons.

So what you are saying is that I can just use the median distance of each box plot and then rank them from highest to lowest? Just trying to follow your reasoning.

Thanks for the support.

Cheers,
Pablo

1 Like
4

Yeah, that was first thought. Sorting by means or medians seems reasonable... but I'm not a statistician! You should probably get their advice and see if there is an established test for this method.

I just googled "post-hoc test rank by distance" and found this paper. With >500 citations, that could be a good thought.
https://cran.r-project.org/web/packages/PMCMR/vignettes/PMCMR.pdf

I know @mortonjt had stats experience. Maybe he can point us in the right direction.

Colin

7

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

8

@Pablo_V, I'm not sure this question has been answered or not, but is there a reason why the standard PERMANOVA test cannot be applied here? That would provide F-statistics for all pairwise comparisons, which you can then sort by effect size (i.e. the F-statistic).

2 Likes