Dear Pablo,
I'm not sure I got all the nuances of your problem.
Just some random thoughts:

- if you cannot measure how similar/consistent are different methods on the same skulls (or at least a subset) it becomes very hard to say with certainty if combining different sources of data does or does not affect the analysis

- perhaps one way to have an idea is to get a homogeneous subset (relative to what you are studying) with both methods and see if the different observations cluster by method

- often times, centroid size tends to be more consistent across repetitions than shape information. This, of course, doesn't say if you will have a bias between dataset but personally I would be more worried about shape

- if you have a scale bar in your pictures, I would imagine that the computation of size will not be affected much while shape might be much more affected by stuff like different levels/types of distortion introduced by using different equipment (cameras). If you don't have a scale bar, well, that's a can of worms unless you have some other measurement that you can use as scale (that's perhaps what you're alluding to when saying that you will "compute centroid size from traditional measurements")

It is a quite vast topic. I have touched upon this and other issues in a review on measurement error in morphometrics which should be online soon(ish). I will post it here when it will be available.

Best,
Carmelo




Pablo Fisichella <fisichellapa...@gmail.com> ha scritto:

Dear Morphometricians

I am new in geometric morphometrics. I have a question perhaps quite simple
with regards to 2D GM. I am comparing the size/shape of human skulls from
different sources using photographs taken in norma lateralis. One part was
obtained through a standardized protocol while the other part was obtained
from published pictures (with permissions) which were obtained using
different protocols. I have traditional measurements for each skull for the
published data set in order to compute the centroid size. However, taking
into account some theoretical issues I feel that such comparison may be
biased especially regarding size (i.e. CS) because both sets of images were
obtained using different distances between the camera and the target. In
order to minimize the size bias I plan to digitize landmarks and
semilandmarks only along the skull contour given that its shape can be
optimally characterized using a 2D approach. The location of landmarks in
other structures poorly characterized from the 2D approach (i.e distinct to
the contour) would increase the size bias. I will do a procrustes analysis
using both data sets and through a multivariate regression of size on the
procustes coordinates I will obtain residuals (i.e. shape variables). To my
knowledge using this approach I will minimize the size bias mentioned.
Obviously I will not use the centroid size in any subsequent analysis, only
the shape variables. Does anyone tell me if this approach is right? Some
suggestions?
thanks in advance

cheers
Pablo


Este correo electrónico se ha enviado desde un equipo libre de virus y
protegido por Avast.
www.avast.com
<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
<#1979225450_DDB4FAA8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

--
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 morphmet+unsubscr...@morphometrics.org.




--
Carmelo Fruciano
Postdoctoral Fellow - Queensland University of Technology - Brisbane, Australia
Honorary Fellow - University of Catania - Catania, Italy
e-mail c.fruci...@unict.it
http://www.fruciano.it/research/

--
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 morphmet+unsubscr...@morphometrics.org.

Reply via email to