How to fit a general-linear mixed-effects model with categorical variables?

Hi,

I am using the function fitglme from the statistics toolbox to fit a mixed-effects model with repeated measurements and categorical predictor variables as follows:

data_nr_acquisitions=table(nr_acquisitions,problem_type,block,subject);
%mixed-effects GLM that allows the effects of the problem type, the offset, and the
%block to vary randomly between subjects.
glme = fitglme(data_nr_acquisitions,'nr_acquisitions ~ problem_type + block + (problem_type| subject) + (block| subject) + (1|subject)');
anova(glme)

The variable problem_type is categorical, but when I run anova it says that problem_type has only one degree of freedom even though it has four possible values. This suggests that Matlab is treating it as a continuous regressors rather than as a categorical variable. Hence, something went wrong.

I tried to instruct fitglme to treat problem_type as a categorical variable with the argument "CategoricalVars" but unlike fitglm the function fitglme does not accept this argument. Can fitglme handle categorical variables and how can I get it to treat a variables as categorical?

[Walter Roberson]

Replaced by http://uk.mathworks.com/matlabcentral/answers/246951-how-to-fit-a-general-linear-mixed-effects-model-with-categorical-variables

[Falk Lieder]

I have added an explanation of the error. Please reopen my question or tell me what else I need to specify in order for my question to be answerable.