
Greetings. I have fit habitat attribute data to warbler presence/absence data. My global model has categorical variables and continuous variables. It looks like this:
Barbuda Warbler Presence ~ Landscape form + Vegetation type +...a few more variables.
Upon fitting to a logistic regression model with "glm" R automatically split the categorical variables into its components. So "Landscape Form," for instance, got split up into "Highlands," "Wetlands," "Plains," etc, and an effect of each of those on the presence of the warbler was calculated (Estimate, along with Standard Error, z value, and Pr(>z).
My question is: is the sum of the Estimates (effects of Highlands, Wetlands, Plains, etc) equal to the overall effect of Landscape Form?
After the model fitting, I am interested in presenting the effect of each habitat attribute as above (Barbuda Warbler Presence ~ ....), and I need to know if it is valid to present, say, Landscape Form and Vegetative Type, as a sum of their parts.
Thank you,
Kevin
