generalized liner mixed model with categorical fixed effects?

#1
Hi Folks,

I am running a generalized linear mixed model with a poisson distribution and a log link function in SPSS 19.

My target/dependent variable is a continuous count and I am looking at how the count changes over time. Most of the variables I am using as fixed effects are continuous. However when I run it with categorical variables as fixed effects I get unexpected output.

With the binary variables I get coefficients for just one side (i.e. when it = 1) and for the 0 it says that it has been "set to 0 because it is redundant". I only get one F though.
My other categorical variable is my time variable (years numbered 0-43), and when I include it I get a coefficient for every year, but again only one F.
The binary variables are both related to time and one of them is the same for each case (represents when new regulations went into effect).

My question in short is why does it do this and how can I make it stop? Is there a way to just get one coefficient per variable? Or is this just what happens when categorical variables are used as fixed effects in this type of model? The categorical variables are some of my most important ones theoretically and I would really like to include them.

Thank you in advance for the help/advice/suggestions!:wave:
 
#2
Hi, I figured it out. It was happening because this model treats categorical variables as factors. Designating them as scale variables (which are treated as covariates) fixed the problem. :yup: