Generalized Linear Models  (GLMs) can be fitted to data with correlated error structures using the method of Generalized Estimating Equations (GEE).

Applications to phylogenetic comparative data are unusual in that there is only one cluster (the phylogeny) whereas the degrees of freedom for significance tests are based on having multiple clusters.

The current solution to this problem has been to calculate the “phylogenetic degrees of freedom” from the tree. There is no statistical justification for this procedure.

A more principled approach is to construct a phylogenetically “correct” variance estimator and use that for the construction of significance tests.

This project will attempt to derive such an estimator and test it by simulation and on real data.

Supervisor: Dr Simon Blomberg