Test rademu #34
Test rademu #34
Conversation
|
|
||
| Identify taxa for which you want robust score tests: | ||
|
|
||
| For instance, which taxa at depth 0 - 10 are more likely to occur or not occur in Natuurgrasland compared to the reference category Akker: |
There was a problem hiding this comment.
Not related to rademu, so probably not the right place to put this, but I forgot we also have to interpret the sccomp data like this? If I remember correctly, in order to do that, we need the model evaluation plots? Or can we somehow export the data from the sccomp model evaluation plots?
There was a problem hiding this comment.
Yes. I think sccomp has a contrasts option to test for such specific effects. See https://mangiolalaboratory.github.io/sccomp/articles/introduction.html#contrasts
| m_refit$coef$pval[taxa_to_test] | ||
| ``` | ||
|
|
||
| Visualise only the significant (p-value <= 0.05) taxa according to the score test: |
There was a problem hiding this comment.
Not specific to rademu, but this sounds like a good idea in general? Can it be applied to the sccomp data?
There was a problem hiding this comment.
Yes, but keep in mind that we have land-use interacting with depth. So the first thing to do is define the specific contrasts we are interested in. Gather for which taxa they are significant and then (optionally) visualise these results (can also be tabularized).
A test of the
radEmupackage as an alternative tosccomp. In short: I'm not convinced that the approach is statistically more valid. So my current advice is that we stick to thesccompframework. This PR is therefore only to document that we have tested this.