0
$\begingroup$

In my regression model I have exponentiated my coefficients, since I understand that coefficients are not directly interpretable.

Problem is, all examples of odds ratios given in books are of binary categorical independent variables.

I don't understand how it works with an independent variable with 5 levels. Do you simply multiply the odds ratio by the number of levels you have moved? enter image description here

Does this mean that as you move from 1 level to the next the odds of the outcome changing are the values in each box?

Thanks guys.

Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ This question of computing odds ratios in logistic regression has been addressed in many threads.. Please look them over; I believe you'll find a good answer. $\endgroup$
    – whuber ♦
    Commented Mar 21, 2015 at 21:35
  • 2
    $\begingroup$ I understand that for experienced statisticians it seems as though this question has been solved elsewhere but for me, none of those threads have explained, in language which I can understand, the difference between odds ratios between a binary categorical, and continuous independent variable. Which makes it immensely frustrating to have the thread marked as duplicate by someone $\endgroup$
    – HenryBukowski
    Commented Mar 21, 2015 at 22:29
  • $\begingroup$ We are sympathetic. However, we cannot read your mind, so we would appreciate more than an expression of frustration. Please edit your question to stipulate what kind of language you do understand and to distinguish your question from those that have already been asked and answered. (It can help to reference similar questions and to point out specifically how they differ from yours.) Without such assistance, we can only guess what you're looking for: and such guesses, unless they accidentally are accurate, could mislead you and are likely to mislead future readers. $\endgroup$
    – whuber ♦
    Commented Mar 22, 2015 at 17:46

0

Reset to default

Browse other questions tagged or ask your own question.