----- Forwarded message from Rodrigo Lima -----
Date: Thu, 3 May 2012 18:23:37 -0400
From: Rodrigo Lima
Reply-To: Rodrigo Lima
Subject: tooth wear and shape: regression?
To: "[email protected]"
Dear morphometricians,
I'm analyzing the shape of voles' teeth which is know to be (very) influenced
by wear, so I classified the teeth in 5 different wear stages (from 0 to 4). I
wanted to do a regression of shape on wear stage and work with the residuals,
similar to what is done for allometry. The problem is that this classification
is a categorical variable, and a multiple/multivariate regression of shape
variables on dummy wear variables won't give me residuals that I can work with.
My questions are:
1 - Suposing that my classification is based on meaningful biological
indicators of wear, would it be conceptually correct to perform a regression of
shape on wear and work with the residuals as "wear free" shape variables?
2 - Given the impossibility of getting residuals from a multivariate/multiple
regression, would it be acceptable to use the values of the wear stage
classification (0 to 4) in a multivariate regression and use the residuals for
further analyses?
Thank you,
Rodrigo
----- End forwarded message -----
