Misdirected post 4 of 7. -mod
-------- Original Message --------
Subject: Combining response variables (e.g. PW + linear measures) and
views (e.g. side + ventral view)
Date: Tue, 05 Feb 2008 15:24:35 -0500
From: morphmet <[EMAIL PROTECTED]>
To: morphmet <[EMAIL PROTECTED]>
-------- Original Message --------
Subject: Combining response variables (e.g. PW + linear measures) and
views (e.g. side + ventral view)
Date: Tue, 5 Feb 2008 12:09:21 -0800 (PST)
From: Cristian Correa <[EMAIL PROTECTED]>
To: <[email protected]>
Hi all,
Can someone comment on combining different types of response variables
in single multivariate analyses (MANOVA, MANCOVA, (NP)MANOVA, CVA, etc.)?
1) I know one should always keep partial warps as a pack not to break
the variance-covariance structure of morphological data, but can
morphological data from one set of landmarks (e.g. PW from lateral
view) be combined with another set (e.g. PW from ventral view) and/or
alternative morphological variables such as (standardized) linear
measurements?
2) Further, since the number of variables would increase considerably
combining these data sets, would it be appropriate to reduce
dimensionality through PC before conducting the e.g. MANOVA? I think
this may obscure the interpretation of the results, specially if
disparate variables load equally heavy in early PCs. Are there other issues?
3) But an alternative would be to first reduce dimensionality separately
for each data type and then combine the first few principal components
of each set into the single analysis. This method would probably
facilitate interpretation of the results but may loss meaningful
information and variance. This method raises the issue of how much
importance one gives to PCs of different dataset and similar eigenvalues.
I'd appreciate your comments, thoughts, and suggested refs.
Cheers,
Cristian
----------------------------------------------------------------------------------
Cristián Correa Guzmán, PhD Candidate
Redpath Museum, Dept of Biology, McGill University
859 Sherbrooke Street West
Montreal, Quebec H3A 2K6
[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
http://www.biology.mcgill.ca/faculty/hendry/
----------------------------------------------------------------------------
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
