I have a dependent variable (Y) say choices of different modes (car, bus, train, car share), a categorical variable and several independent variables (age: categorized into three different groups 18-25, 26-40 and 40-60 & education level, gender), all categorical variables. Is it possible to form a multinomial logistic regression model to establish relation between the dependent and the independent variables?
I also have perception of safety, security of the respondents (in 1 to 5 scale) about using different modes. What different tests can I conduct?
Hi Tarapanda,
I agree with what Paul said. You should not try to make a continuous variable into a categorical one because you are losing information and cutoffs are arbitrary. Also, values on either side of the cutoff are very close than what your categories tell you. You should treat them as continuous variables.
The answer to your main question is yes, you can use it. I suggest reading up on multinomial logistic regression before you blindly apply it using R or another statistical software.
- Badges
- Science topic
- Similar topics
- Journal Impact Factor