In article <4va62e$1ija at uni.library.ucla.edu>,
Ron Kagan <rkagan at ewald.mbi.ucla.edu> wrote:
>For those of you who have used maximum likelihood analysis (DNAML and
>DNAMLK of Felsenstein's Phylip package): What level of significance do
>you use for the likelihood ratio between the log likelihoods w and w/o
>the molecular clock, to reject the null hypothesis of the molecular
>clock? Is p<0.05 still the standard, or should I use a stricter criteria
>(p<0.01)?
I don't think that a survey is quite the right thing to do here. It is
a matter of how cautious you want to be. If the statistical test is done
properly (which is an issue, of course), then P = 0.05 means that if there
is actually a molecular clock, you will falsely conclude that there isn't
5% of the time. If you report your P value the reader can judge for
themselves, though given biologist's statistical backgrounds it might be wise
to mention that this is what the P value means.
So it is a bit a matter of taste.
--
Joe Felsenstein joe at genetics.washington.edu (IP No. 128.95.12.41)
Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360 USA
