Dear MorphMetters,
As I did last year, I'd like to grab your attention for a moment
to advertise a long essay of mine,
"The inappropriate symmetries of multivariate statistical analysis
in geometric morphometrics," just posted by the Springer
journal Evolutionary Biology. The article is free for download via
http://link.springer.com/article/10.1007/s11692-016-9382-7/fulltext.html
or from the journal's own website.
The paper runs to 37 double-column pages, but a summary
can be brief. I am claiming
that the ways we use GMM in most of today's organismal
biological applications, whether of growth, evolution,
or disease, don't actually make much biological sense.
I go on to offer a small assortment of better alternatives,
including but not limited to the idea of deflated Procrustes analysis
that I published in the same journal last year. The article incorporates
two appendices that might merit your closer attention. Appendix 1
is a principled criticism of the RV coefficient some people are recommending
nowadays for use in connection with PLS analyses; my conclusion is
that it serves no valid scientific purpose and should never
be used. Appendix 2 shows
what covariances of Procrustes shape coordinates actually look like
and why it follows that principal components of these
covariance matrices are usually less useful than we hoped they were.
Feel free to go get your own copy of this publication and
either learn from it or send me your objections. Yours from Vienna
with best wishes, Fred Bookstein
--
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
---
You received this message because you are subscribed to the Google Groups
"MORPHMET" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
