RRPP:: pairwise comparison of unique allometries

[dave.a...@colby.edu]

Hopefully no one will object if I have a question related to RRPP, rather than geomorph itself!

I have a "shape" dataset consisting of 14 linear measurements from 207 specimens in 8 species. I constructed a list including the data table, and metadata for use with `lm.rrpp`. Body size and species both have a strong influence on "shape" (Y ~ size + species). Next, I wanted to test whether species have unique allometries, examining the strength of the interaction, size:species (from a model of the form Y ~ size * species). The interaction term is significant and has a large effect size (Z = 3.8, compared to 5.6 and 11.6 for size and species). But where I'm not certain is the post hoc comparison looking at differences in the allometric slopes for each species.

Should it be in the form:

```

pairwise(fit       = Y ~ size * species,

         fit.null  = Y ~ size + species,

         groups    = species,

         covariate = size)

```

Or is having the interaction term in the model and specifying a `covariate` argument redundant? After looking at the docs for `pairwise` I wasn't sure quite has the covariate argument works.

Thanks for your help here!

-Dave

[Mike Collyer]

Hi Dave,

You set it up correctly.  You null model has parallel slopes but allows intercepts to vary, which is the appropriate null model.  The reason to have a covariate declared is so the function knows to compare slopes instead of LS means.

Cheers!

Mike

--
You received this message because you are subscribed to the Google Groups "geomorph R package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to geomorph-r-pack...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/geomorph-r-package/b807a701-17e3-48fe-893d-8c6edfd8c9c6n%40googlegroups.com.

[dave.a...@colby.edu]

Excellent. Thanks!