Re: Bayesian analysis for morphological data!!!

[David Marjanovic <david.marjanovic@gmx.at>]

> Er, all that seems a lot complicated to me. I suppose a comparison
> between trees based on this "new" method and trees based on the "old"
> method is available somewhere so as to give one a more precise idea?

Looking at the trees wouldn't help. A tree is a tree.

Basically, Bayesian methods are likelihood methods with an inbuilt 
alternative to bootstrapping -- to learn how well supported the clades in 
your analysis are, you don't need to do an extremely time-consuming 
analysis after the analysis; instead support values already come attached 
to the tree.

Likelihood methods are parsimony plus a model of evolution -- parsimony 
alone, the old method, assumes that characters don't evolve except when 
the data say it must have, while likelihood methods, the new ones, 
presuppose that every character will evolve over time, or rather, over 
branch length (not necessarily all at the same speed). In other words, 
convergence is expected. As a result, likelihood methods are more robust 
to long-branch attraction.

-- 
Lust, ein paar Euro nebenbei zu verdienen? Ohne Kosten, ohne Risiko!
Satte Provisionen für GMX Partner: http://www.gmx.net/de/go/partner