Hi,
I am a little bit confused about how AMOVA and HOMOVA differ. I understand that AMOVA tells you whether there is a significant difference between the centroids of groups but I’m struggling to understand what HOMOVA tells you.
When would you choose to use one test over the other?
Cheers,
Laura
My analogy is that if you wanted to know whether men and women had different heights, you’d want to know whether the mean height of men is different than the mean heigh of women given the variation in their heights. You’d probably use a T-test or by analogy, AMOVA. Say you didn’t care about their mean heights, but were rather interested in whether there was a significant amount of variation in their heights. Does the standard deviation of men’s heights differ from the SD of women’s heights? Here you’d probably use Bartlett’s test for homogeneity of variance. For microbiome data we use HOMOVA to see whether the spread in the data is different between groups. We’ve used this type of test to look at differences in stability.
Hope this helps…
Pat
Hello,
I use these two metrics for my beta-diversity analysis, and I have a question. Based on the explanation here and in your 2008 paper ( Evaluating different approaches that test whether microbial communities have the same structure | The ISME Journal | Oxford Academic (oup.com), what would be the difference between doing a statistical test with alpha diversity between 2 groups (is this even correct?) VS. HOMOVA analysis on beta diversity? Thank you for your help!
Regards,
Elliston Vallarino
Hi Elliston,
Well the alpha diversity approach would use a metric like richness or shannon diversity for each sample whereas the beta diversity approach would use something like Bray-Curtis distances between samples. Those are pretty different approaches and answer different questions. Alpha diversity looks at each sample separately whereas beta diversity compares the composition and abundance of OTUs between pairs of samples. You could do a Kruskall-Wallis test between shannon diversities and AMOVA between Bray-Curtis distances.
Pat