Currently, confint applied to a phylolm object does not throw any warning, but falls back to confint.default, which uses the quantiles of the Gaussian distribution.
With the proposed change, confint would call confint.lm, and gives the (broader) intervals obtained from a Student distribution with n-d degrees of freedom.
See the added tests for an example.
I added a line in the doc and NAMESPACE, checks pass locally, but I hope I did not forget anything while defining the method ?
Hi,
Currently,
confintapplied to aphylolmobject does not throw any warning, but falls back toconfint.default, which uses the quantiles of the Gaussian distribution.With the proposed change,
confintwould callconfint.lm, and gives the (broader) intervals obtained from a Student distribution with n-d degrees of freedom.See the added tests for an example.
I added a line in the doc and NAMESPACE, checks pass locally, but I hope I did not forget anything while defining the method ?
Thanks, Paul