Dear morphometricians,
I greatly enjoyed the discussion on MDS, which reminded me about a
similar discussion on MANOVA/MANCOVA, DA and related techniques that we
had in MORPHMET about one year and half ago. In that case also the
discussion was about how appropriate (or inappropriate) is to use
multivariate methods, which have been routinely employed in traditional
morphometrics, in the analyses of geometric morphometric data. That
issue (use of MANOVA etc. for GMM analyses) has been recently discussed
by Klingenberg & Monteiro (2005, Syst. Biol., 54: 678-688). Not
everybody will agree with them, probably, but I found their explanations
extremely clear and very helpful.
I haven't got anything particularly interesting to add to the discussion
on MDS. Personally, I used it a few times when my only purpose was to
find the ordination which best summarized similarity relationships among
specimens to visualize them in a 2D or 3D scatterplot. At least
sometimes, the matrix correlation between the matrix of Procrustes
distances and the matrix of Euclidean distances computed using the MDS
axes seems to be slightly higher than the correlation between the PRD
matrix and  the matrix of Euclidean distances computed using the first
(2 or 3) PCs. Whether this improvement is negligible and PCA should be
preferred anyway, given the issues with MDS discussed by Prof.
Bookstein, I cannot say.
I strongly hope to have soon more papers like Klingenberg & Monteiro
(2005), or Rohlf (1998, Syst. Biol., 47:147-158), that may provide us
with clear explanations on otherwise rather difficult issues and that
may act as guidelines in a form accessible not only to statisticians. It
seems to me that GMM has indeed been a revolution (Rohlf & Marcus, 1993,
TREE, 8:129-132), but after more than 10 years from that paper there are
still some (or several) issues that needs to be discussed and clarified,
hopefully in a way that everybody with a basic knowledge of statistics
can understand.
That there are still topics open to discussion was recently suggested to
me also by reading in Fink et al. (2005, Proc. R. Soc. B, 272:
1995-2001) that a regression was done using the first five pairs of PWS,
contrary to the common view, expressed also by Prof. Bookstein, that PWS
should be used as a joint set and not as separate variables. However,
there might be exceptions to this commonly accepted 'rule', which I do
not understand but I'd like to.
I wonder whether it might be time to try to organize a morphometric
workshop on these and similar issues.
Thanks in advance for your comments and advice.
Cheers

Andrea



Dr. Andrea Cardini
Hull York Medical School
The University of York, Heslington, York YO10 5DD, UK
&
The University of Hull,Cottingham Road, Hull HU6 7RX, UK
tel. 01904 321752
fax 01904 321696
E-mail: [EMAIL PROTECTED]
http://www.york.ac.uk/res/fme/people/andrea.htm
-------------------------------------------------------------------------
Visita http://domini.interfree.it, il sito di Interfree dove trovare
soluzioni semplici e complete che soddisfano le tue esigenze in Internet,
ecco due esempi di offerte:

-  Registrazione Dominio: un dominio con 1 MB di spazio disco +  2 caselle
   email a soli 18,59 euro
-  MioDominio: un dominio con 20 MB di spazio disco + 5 caselle email 
   a soli 51,13 euro

Vieni a trovarci!

Lo Staff di Interfree 
-------------------------------------------------------------------------
-- 
Replies will be sent to the list.
For more information visit http://www.morphometrics.org

Reply via email to