[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: Beating a dead horse[ was Re: A NEW REF, SOME TETRAPODS, AND CLADISTICS & PT, Re: Crown clades a-frikkin'-gain]



Just a few thoughts, then I'll shut up.




>
>>Did you hear about the three or four different analyses of elephant shrews
>-- >all of equal parsimony --
>        How do you compare relative parsimony between seperate analyses? I
>know of no method for doing so.

An interesting question.  The stats that actually measure "parsimony" - and
by "parsimony," I assume we're talking about the fit of the data set to the
tree - are not exportable between data sets.  Data set 1 may have a better
CI than data set 2, but without knowing data set size, this is a
meaningless comparison.

However, there are methods for (a) assessing internal support and (b)
comparing how two different data sets react to the same tree.  These can be
useful in making the kinds of comparisons we're talking about.  Data set 1
may suggest a different tree from data set 2, but the bootstrap values,
Bremer support, and (if it's molecular data) likelihood ratios might be
very low in one and not the other.  Or the differences might turn out to be
so slight that, statistically speaking, they're effectively the same tree.
This is actually what happens most of the time.  Of course, an even more
powerful approach is to (c) combine the data sets and get an answer from
one common voice.



chris



-=--=--=--=--=--=--=--=--=--=
Christopher Brochu

Postdoctoral Research Scientist
Department of Geology
Field Museum of Natural History
Lake Shore Drive at Roosevelt Road
Chicago, IL  60605  USA

phone:  312-922-9410, ext. 469
fax:  312-922-9566

cbrochu@fmppr.fmnh.org