If the 100 trees are trees sampled from a Bayesian posterior, or else trees
from bootstrap samples of your data, then you might just take the estimates
from each tree (say estimates of a regression coefficient). Consider their
distribution and ask whether the null hypothesis value (such as having
slope 0) is in the tail of that histogram.
The overall P value will be the proportion of estimated slopes that are
below zero (unless you want to do a two-tailed test).
Under the null hypothesis this will have the proper rejection probability.
As you take more and more trees the power increases some, but the type I
error rate stays the same.
If the 100 trees are something else, such as the personal opinions of 100
of your friends, then there is no statistical justification for this.
J.F.
----
Joe Felsenstein j...@gs.washington.edu
Department of Genome Sciences and Department of Biology,
University of Washington, Box 355065, Seattle, WA 98195-5065 USA
[[alternative HTML version deleted]]
_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
