Dear Simone,
You could do a likelihood ratio test on the lambda-fitted model vs. the
no-lambda model. You could also use the AICc values returned for each
model by geiger and calculate AICc weights. Brian O'Meara has a nice
summary of this on his website:
Hi Andrew, Simone, et al.,
It's probably worth adding the very minor distinction that the results of
Andrew's suggested test and of phytools' phylosig test have slightly
different interpretations. Assuming lambda 1, a significant LRT when
comparing the fit of a lambda model to a non-lambda, BM
Hi Simone,
It may be worth noting that likelihood ratios are not expected to be
chi-square when testing nested models in which a parameter is forced to
it's boundary (i.e. lambda = 0, since lambda 0 is not defined; or lambda
= 1, depending on how you have defined that boundary.) So some care is