0
$\begingroup$

So, in case the predictor variable is continuous, then it's easy to interpret the coefficient. Let's say the gpa score would work as a predictor and admitted to the university ( Yes or No) would be a response variable. So, if the gpa coefficient is 0.43, then we simply take exp(0.43) = 1.54 and conclude that by every unit increase in gpa score the student is 1.54 more likely to be admitted to the university.

But in case the predictor variable also is logistic. Let's say the predictor variable is the type of educational system the student has been exposed to - Private / public school. Then if the reference level would be public school, how should I interpret the same coefficient 0.43? Any advice is welcome!

$\endgroup$
0
$\begingroup$

Exactly the same as for a numeric variable: Changing from public to private school, while keeping all other predictors constant, increases the odds (not probability!) of admission by factor 1.54.

Internally, in regression (including logistic regression), nominal variables are treated as numbers, with possible values being 0 and 1. So, in your case, public school would be encoded by a dummy variable having value of 0, while a private school would be the same variable with the value of 1.

$\endgroup$

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.