morphmet wrote:
> Hello to everyone.
>
> I need your help!
>
> Can I perform an F-test on individual coordinates from the alligned
> specimens matrix (from tpsRelw)?
You can, but why would you want to? Individual coordinates are a
function of position in the coordinate system and unlikely to be
biologically meaningful unless, say, you are using some anatomical
coordinate system.
>
> I want to compare two samples of skulls (same species, little
> difference) and I need the *p-*value of individual landmark
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ unlikely
> coordinates (after GPA). The alligned specimes matrix seems convenient
> because the number of coorinates is retained.
>
> I made multivariate regression of the samples in MORPHEUS, I have the
> plots of splines and now I want to test the differences on the level of
> individual coordinates.
>
> Is this possible, can it be done?
At best, you might consider 2D or 3D *sets* of coordinates for each
landmark (Bookstein has a chapter in the Whitebook on this), but even
that violates the total shape approach implicit in Procrustes analysis.
I would view such results as only suggestive or illustrative.
Alternative alignments could give different results at the level of
individual landmarks, but comparisons based on overall shape
similarities or differences should be consistent (sensu latu).
Whatever you decide, once you have the aligned coordinates you can treat
them as any other set of variables and do whatever tests you want up to
the limitations imposed by a singular covariance structure, e.g.,
anything based on Wilks' lambda won't work, but Pillai or
Hotelling-Lawley trace-based tests will.
-ds
>
>
> Thank you very much in advance!
>
> Tina Klenovsek
>
> ------------------------------------------------------------------------
> Sick sense of humor? Visit Yahoo! TV's Comedy with an Edge
> <http://us.rd.yahoo.com/evt=47093/*http://tv.yahoo.com/collections/222>to
> see what's on, when.
>
--
Dennis E. Slice
Department of Anthropology
University of Vienna
========================================================
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
