Hi R-Sig-Phylo,
We are using the PGLS function in caper and would like to calculate
marginal R^2 values for the factors in the model. The model takes the form:
m1 <- pgls(y ~ a + b + c, data = comp_data, lambda = "ML", delta = "ML",
kappa = "ML")
For normal GLMs we'd run the full model, then run a model (m1.1) dropping
out factor c. We'd next extract the R^2 value of each and subtract them to
yield the marginal R^2 of factor c. We would then do the same procedure
excluding factors a and b to get their margin R^2s.
Our questions are:
1. Does this procedure violate any assumptions of PGLS?
2. Should we should set lambda, delta, and kappa to "1" to make sure that
the trees are scaled the same for each model comparison?
3. Is it more appropriate to use the adjusted R^2
[summary(m1.1)$adj.r.squared] as opposed to the un-penalized value
[summary(m1.1)$r.squared]?
Many thanks for your thoughts!
Anthony
[[alternative HTML version deleted]]
_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
