I need some help with interactions of dummy variables.

Can I model interactions between two or more qualitative variables like we learned to do interactions between 1 quantitative and qualitative (as in module 14)?

[NEAS: Yes]

If not all the combinations in my data set are present, do they have to be included when I determine the degrees of freedom? For example, I have 5 occupation categories and 3 industry sectors, but not all occupations are present in all sectors. How do I determine the degrees of freedom for occupation * sector? How do I determine degrees of freedom for qualVar1 * qualVar2 * qualVar3? For example, occupation * sector * geographical region.

[NEAS: An interaction is another parameter. Whether there are data points in a class is not relevant; take the number of data points. You may have 100 data points in a class or 0 data points. If no data points exist in a class, there can be no interaction term for that class.]