Hi Everyone,
I have the following data: Species is a nominal main effect, Sample
is a nominal random effect nested within Species, Time is a
continuous predictor variable, and Absorbance is a continuous Y
variable. In other words, within each Sample there are data for a
single linear regression.
I am interested in the slopes of the regressions, and whether they
vary between Samples, and between Species. I'm not actually
interested in the intercepts per se, although these are bound to vary
as well
So this is an analysis of covariance with one level of nesting. The
basic ANCOVA is covered in the Help Guide, but not a version with
nesting. My question is how to code the model. I need to allow the
slopes to vary across samples (and obviously across species). As I
understand it, the test for whether slopes differ across a nominal
effect is the test for the crossed effect of the X predictor of the
regression and the nominal effect. But I don't now how to do this in
the nested example. Do I cross Time with Sample and Time with
Species, or Time with Sample[Species] and Time with Species? Right
now I have
Species
Sample[Species]
Time
Time*Sample[Species]
Time*Species
Can I add a triple cross? How does that work with the nesting? Or am
I on the wrong track altogether?
Thanks,
Gareth
=====================================================================
Gareth J. Russell
Department of Mathematical Sciences (Division of Biological Sciences)
New Jersey Institute of Technology
Department of Biological Sciences
Rutgers University
Phones: (973) 642-4299 (NJIT)
(973) 353-1429 (Rutgers)
Fax: (973) 596-5591
E-mails: russell@njit.edu
grussell@andromeda.rutgers.edu
WWW: http://web.njit.edu/~russell
=====================================================================
|