1
$\begingroup$

used the glm function in R to model a logistic regression of a binary repsonse and 3 categorical predictors. My problem is that according to the summary non of the levels is significant.

$\endgroup$

1 Answer 1

2
$\begingroup$

That happens sometimes even with the ordinary regression analysis. Individual coefficients (their z-values) are not significant when whole regresssion model is significant by the F-test.

$\endgroup$
2
  • $\begingroup$ the problem ist that non of the over 600 levels is significant...so this happens quite a lot. Is there some way to change my data to get better results? $\endgroup$
    – Hein
    Commented Dec 31, 2014 at 11:47
  • $\begingroup$ What are the 600 levels? You have a binary response and three predictors? $\endgroup$ Commented Dec 31, 2014 at 18:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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