Faith in parsimony

[Richard.Zander at MOBOT.ORG]

Maximum parsimony methods may use bootstrapping, Bremer support and the
Templeton test (and doubtless other tests buried in PAUP) to judge
statistical reliability. Although the Bayesian methods report a full
posterior distribution, it isn't, since there are so many assumptions that
are not investigated as to their affect on a branch arrangement of interest
if wrong.

The Bayesian gimmick is to equate the likelihood slogan that the data
contains all the information needed for statistical analysis because the
likelihood of the data equals the likelihood of the hypothesis (then they
add, also you gotta have a model). Bayesian analysis is SUPPOSED to
incorporate all factors that reasonable affect the outcome, so you can make
a bet, given the odds and what you are risking, on a one-time event.
So-called Bayesian analyses may often evaluate some few assumptions to see
whether the outcome is robust to changes in assumptions, but even this is
subject to problems, e.g. "the sequence is more than 95% correct" or "the
two genes produce trees that are more than 95% congruent, therefore we can
combine the data." For these, the posterior probability of a given
arrangement being right depends on the chance of it changing if the
assumption (sequence, congruence) is wrong. 95% time 95% is 90%, for a
single assumption.

Both maximum parsimony and Bayesian analysis can be used for morphological
data sets. Bootstrapping can be weak unless you restrict the number of taxa
to what you feel is an "uncontested" group.
______________________
Richard H. Zander
Bryology Group, Missouri Botanical Garden
PO Box 299, St. Louis, MO 63166-0299 USA
richard.zander at mobot.org <mailto:richard.zander at mobot.org>
Voice: 314-577-5180;  Fax: 314-577-9595
Websites
Bryophyte Volumes of Flora of North America:
http://www.mobot.org/plantscience/bfna/bfnamenu.htm
Res Botanica:
http://www.mobot.org/plantscience/resbot/index.htm
Shipping address for UPS, etc.:
Missouri Botanical Garden
4344 Shaw Blvd.
St. Louis, MO 63110 USA


-----Original Message-----
From: Robert Mesibov [mailto:mesibov at SOUTHCOM.COM.AU]
Sent: Friday, July 22, 2005 2:06 AM
To: TAXACOM at LISTSERV.NHM.KU.EDU
Subject: [TAXACOM] Faith in parsimony
The real difficulty is that maximum parsimony methods provide no  warning
when the single reconstruction selected has low probability of being
correct...By contrast, Bayesian methods report a full posterior distribution
on the space of possible trees and arrangements.