Patrick Arnold <[email protected]> ha scritto:
Dear all,
I am looking for patterns of integration between 3 substructures. I
subjected the procrustes variables to pair-wise PLS analyses and
received 3 RV values. Does the highest RV value correspond to the
strongest integration or are RVs from different PLSs not comparable?
Is their any possibility to perform a PLS on 3 structures
simultaneously?
Dear Patrick,
yes, there are approaches to perform PLS on multiple blocks (see
Bookstein et al. 2003 - Journal of Human Evolution; also search for
"multi-block PLS"). A multi-set RV coefficient has been also proposed
by Klingenberg (2009 - Evolution and Development) as the average of
the pairwise RV coefficients.
The value of the RV coefficient also depends on sample size (Fruciano
et al. 2013 - PLoS One) so comparisons of the value itself across
datasets with different number of observations are inappropriate,
unless this is taken into account (please, see the paper for details).
Please, also note that, although you might find them mentioned
together for convenience, partial least squares analysis and the
Escoufier RV coefficient are distinct things.
Best,
Carmelo
--
Carmelo Fruciano
Honorary Fellow - University of Catania - Catania, Italy
e-mail [email protected]
http://www.fruciano.it/research/
--
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
