Dear Splus list
I am writing a manuscript in which I present several figures showing the
results of fitting a generalized additive model (family = binomial) to a
dataset containing a number of categorical and continuous predictor
variables. Each figure contains several plots, and the plots show the
response curve for each predictor variable (e.g., Figure 9.2 in Modern
Applied Statistics with S-Plus, Third Edition). My question is how best to
describe the y-axis of each plot. As an ecologist, the nitty-gritty
statistical details of GAM are a bit over my head, and the manuscript is
going to an ecological journal, the audience of which will be even less
familiar with GAMs than I am. For a model describing the relationship
between predictor variables and the probability of presence/absence by
Species X, I conceptualize the y-axis of these plots as the "relative
effect of the predictor variable on the probability of occurrence by
species X". However, that seems too vague for my current purposes. In a
recent Conservation Biology paper, Joly et al. (2001, 15:239-248)
described the y-axes of their GAM plots as follows: "The y-axes are
log-scale (link function), are based on partial residuals, and indicate
the relative influence of each explanatory variable on the prediction."
Having failed to find any other descriptions in other sources, I would
appreciate assistance in coming up with a concise description (3-5
sentences) of what the y-axis of a GAM plot represents and briefly how it
is calculated. Particularly important is that the description be
understandable to an audience of ecologists not familiar with GAM. Thanks
for any help you can provide.
Carlos
***************************
Carlos Davidson
Department of Environmental Studies
California State University, Sacramento
6000 J Street
Sacramento, CA 95819-6001
Phone: (916) 278-6063
Fax: (916) 278-7582
Email: cdavidson@csus.edu
Web: www.csus.edu/indiv/d/davidsonc
|