1
$\begingroup$

I would appreciate it if someone could help me answer this.

I recently carried a logistic regression model using the following basic formula: y ~ x + factor1 + factor2

My supervisor suggested it would be good to carry out a likelihood ratio test, comparing the above model output to the output from: y ~ factor1 + factor2

I understand that the LRT compares the goodness-of-fit of two models, but I dont understand why the second model (null model) would be y ~ factor1 + factor2 and not y ~ x?

$\endgroup$
1
$\begingroup$

Well, if the difference between y ~ x + factor1 + factor2 and y ~ factor1 + factor2 is not significant, that would mean that x is not needed to predict y - is that what you want to check?

$\endgroup$
2
  • $\begingroup$ Ah I see, thank you. Yes the difference is significant, so I guess that means that x is necessary to predict y. Thanks! $\endgroup$
    – HotDesk
    Apr 4 '17 at 11:43
  • $\begingroup$ The answer says that x is not necessary to predict y and not the other way around. $\endgroup$ Apr 4 '17 at 12:21

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.