1
$\begingroup$

I have Likert-type data (ranked as 1 least important to 7 most important) for both dependent and independent variables. Can I use multinomial logistic regression? My second question is, if the dependent variable is binary and all independent variables are ordinal (Likert-type), would that make a difference in using a model for both type of these variables?

$\endgroup$
1
$\begingroup$

Ordinal logistic regression is what you need in this case. Multinomial is used when the DV is nominal. See this page for more info.

To answer your second question, logistic regression requires the dependent variable to be categorical while the independent variable(s) may be continuous or categorical, so the ordinal logistic regression will be fine.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Ordinal logistic regression assumes parallel slopes, which multinomial does not. But multinomial costs you power - with a large sample, you don't mind. $\endgroup$ – Jeremy Miles May 17 '16 at 22:59
  • $\begingroup$ For a 1-7 scale, I suspect most people would use OLS. $\endgroup$ – Jeremy Miles May 17 '16 at 22:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.