----- Forwarded message from adrien.an...@doct.ulg.ac.be -----
Date: Mon, 13 May 2013 13:01:02 -0400
From: adrien.an...@doct.ulg.ac.be
Reply-To: adrien.an...@doct.ulg.ac.be
Subject: manova tests
To: morphmet@morphometrics.org
Dear Morphometricians,
My name is Adrien, I'm new to morphometry and I'd like to understand correctly
the results I obtained for the procrustes Anova in MorphoJ.
Here is my main question:
What is the meaning of having a significant or a non-significant p-value in the
symmetric and asymmetric component manova tests?
here are the results I obtained with my data.
Classifiers used for the Procrustes ANOVA:
Extra main effect(s):
-- Extra 1: site
Individuals: name
Centroid size:
Effect SS MS df F P (param.)
Extra 1 0,941346 0,058834 16 2,23 0,0056
Individual 5,232293 0,026426 198
Shape, Procrustes ANOVA:
Effect SS MS df F P (param.)
Extra 1 0,01783523 0,0000412853 432 2,06 <.0001
Individual 0,10700362 0,0000200156 5346 11,46 <.0001
Side 0,00074918 0,0000277473 27 15,88 <.0001
Ind * Side 0,01009397 0,0000017470 5778
Shape, MANOVA tests of effects:
Symmetric component of shape variation:
Effect Pillai tr. P (param.)
Extra 1 3,14 <.0001
Note: the test for 'Individual' used the symmetric component of the residual as
the 'error' effect.
Asymmetry component of shape variation:
Effect Pillai tr. P (param.)
Extra 1 2,03 0,4523
Side 0,72 <.0001
For this analysis, i used 2 classifiers. the first one (name) is the specimen
ID, the second one, "site", is the site where the specimen come from. The last
one has been used as the extra main effect.
there is no error factor in this analysis because I didn't digitized all my
individuals twice. I did it before on a subsample, and mean squares of FA, DA,
and individual variation were found
to exceed the error component, indicating that the contribution of measurement
error to
overall shape variation was small.
thank you very much for your help
Adrien
adrien.an...@ulg.ac.be
----- End forwarded message -----
