Dear all,

I have two categorical variables that are socialclass (1=low, 2=middle, 3=high) and jobtype (1=manufacturing, 2=services, 3=construction, 4=other). I am running a logit regression and want to interact them with the level of unemployment in the region. Normally I would just do:

gen unemploymentsocialclass=socialclass*unemployment
gen unemploymentjobtype=jobtype*unemployment

However, I think that for jobtype the interpretation would not make sense as an increase in jobtype is not really meaningful given the variable options. In this case, should I create 4 new independent binary variables for each of the options of jobtype and then interact those with unemployment?

Thank you very much!

Joan