Is it valid to use a likelihood ratio test to compare 2 nested linear models instead of anova? I'm trying to assess whether the quadratic model is a better fit to the data. I know anova seems to be the standard test for comparison but I was wondering whether I could use the likelihood ratio test?

i.e. comparing

model 1: Height~Age

model 2: Height~Age + I(Age^2)

  • $\begingroup$ Would you please post a link to the data, or add the data to the question? $\endgroup$ – James Phillips Aug 21 at 17:31
  • 1
    $\begingroup$ Yes you can use the LRT. Both it and an F-test are appropriate in this case. They aren't exactly equivalent but will give you similar results. $\endgroup$ – logistic Aug 21 at 17:35
  • $\begingroup$ Based on some super quick simulation studies, it looks like the LRT is a little more powerful for fixed $\alpha$ and when the normal errors specification is accurate. $\endgroup$ – logistic Aug 21 at 17:39

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.