I have two IV's: cognitive ability test scores (a continuous variable), and task difficulty levels (3 levels: easy, medium, difficult) I want to predict a categorical outcome with 3 types of behavior (type 1, type 2, type 3) with the main effects and interactions of both IV's. I'm using multinomial logistic regression.

My questions: Should I, like in multiple regression, center the continuous IV and dummy code the categorical IV?

  • $\begingroup$ Dummy coding: yes. Centering continuous IV: if you want to. (However, mean-centered values may make interpreting the "intercept" of the model(s) easier.) $\endgroup$ – Gregg H Apr 21 '18 at 14:50
  • $\begingroup$ Thanks! With dummy coding, the results show a row for every dummy variable. How do I interpret this? can you refer to any references? $\endgroup$ – user3315563 Apr 21 '18 at 16:01
  • $\begingroup$ There should be one fewer rows for the dummy variable than the number of categories. The missing category is the reference category (that is the "intercept"). The dummy variables indicate how much should be added to that intercept to account for being in that category. $\endgroup$ – Gregg H Apr 21 '18 at 16:07

Your Answer

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

Browse other questions tagged or ask your own question.