Binomial regression does not give coefficients for all IVs

Question

I have a dataset with the dependent variable presence / absence (0 and 1) for a certain species. I have three categorised IV's (2 IV's with 3 categories and 1 with 2 categories). To test the response of the presence / absence to the 3 IV's, I have to perform a binomial regression, right?

I can run a model in R which looks like this:

glm(species.presenceabsence~TREATMENT+WEEK+TYPE, data=speciesdata, family=binomial))

The outcome looks like this:

Coefficients:
               Estimate Std. Error z value Pr(>|z|)
(Intercept)    -19.4779  2955.2811  -0.007    0.995
TREATMENTHigh  -19.1984  2955.2812  -0.006    0.995
TREATMENTLow    -0.9217     0.8101  -1.138    0.255
WEEK10          18.2767  2955.2812   0.006    0.995
WEEK15          19.1984  2955.2812   0.006    0.995
TYPEF           -0.3083     0.7881  -0.391    0.696

What I don't understand is why R doesn't give the coefficients of all the IV's. How do I interpret these data? I'm a bit surprised that none of the coefficients has a significant effect.

This is a FAQ. There are lots of threads that address this issue. The linked thread is just the first one I thought of. Try reading through some of the threads categorized under the [tag:categorical-data] tag. — gung - Reinstate Monica, Apr 23 '15 at 16:46
The default coding scheme for categorical predictors is reference-level coding. See e.g. here. (See also the [tag:hauck-donner-effect] tag.) — Scortchi - Reinstate Monica, Apr 23 '15 at 16:48

Binomial regression does not give coefficients for all IVs

0 Answers0