Can I use categorical independent variables in tobit regression? If yes, how to interpret the results?