1) Re-coded some likelihoods to fix bugs. I attempted before to keep things in sync with the log-likelihoods for the linear case, which did not work as I had intended. Might simplify this in the future if possible.
2) Added code to retrieve the log-likelihood per species/site in TMB. This is to facilitate 3), but does not work (yet?) for the laplace approximation. Potentially, TMB leaves out constants, or provides the wrong likelihood for the laplace with report. It is a known issue (with TMB users), and I have not been able to find a work around.
3) Added an option to calculate anova species-specific. Here, p-values are adjusted for multiple testing (using base R's p.adjust). This is partially to facilitate comparisons between the linear and quadratic response models, as the number of parameters is often too large to compare the two models in their entirety. The default is settings provides whole model comparisons (as before).
1) Re-coded some likelihoods to fix bugs. I attempted before to keep things in sync with the log-likelihoods for the linear case, which did not work as I had intended. Might simplify this in the future if possible. 2) Added code to retrieve the log-likelihood per species/site in TMB. This is to facilitate 3), but does not work (yet?) for the laplace approximation. Potentially, TMB leaves out constants, or provides the wrong likelihood for the laplace with report. It is a known issue (with TMB users), and I have not been able to find a work around. 3) Added an option to calculate anova species-specific. Here, p-values are adjusted for multiple testing (using base R's
p.adjust
). This is partially to facilitate comparisons between the linear and quadratic response models, as the number of parameters is often too large to compare the two models in their entirety. The default is settings provides whole model comparisons (as before).