Multiple Logistic Regression question

Greetings, I have three categorical IVs and one categorical DV. However, the DV is a self-reported ranking from 1 to 4 (1= not important, 4 = very important). I want to know if any of the IVs (separately or in combination) predict perceived importance of a training module. My understanding of logistic regression is that the DV should be dichotomous. Currently my DV is not. In this instance, is it acceptable to dichotomize the responses? E.g., scores of 1, 2 = not important, 3,4 = important. Should I be considering another analysis? Thanks for any thoughts?