I mistyped in stata "logistic a" where a is a binary variable and to my surprise it actually ran.

What is a logistic regression doing when it's running without any independent variables?

enter image description here

  • 5
    $\begingroup$ It might be just fitting the null model (intercept only, log-odds of being in the class of interest). To verify, please edit your question to show the class breakdown of your binary variable a and the output from the command you ran. $\endgroup$ – EdM Jan 3 '20 at 16:50
  • 3
    $\begingroup$ Edited my answer, does this suffice? $\endgroup$ – Paze Jan 3 '20 at 16:54
  • $\begingroup$ Incidentally, you can do this with regress, ologit, etc. In all cases you get an intercept only model: a heavy-handed way to characterize the mean/proportion, etc. of the dependent variable. Contrast mean a with regress a with Odds/(Odds+1) from your output above. $\endgroup$ – Alexis Jan 4 '20 at 18:35

It's an intercept-only model. Stata's notation _cons is the intercept term. With no other regressors, the odds is $p/(1-p)$ where $p$ is the proportion with positive response.

  • $\begingroup$ I understand what the intercept and the constant is, but I'm still not sure if this actually tells me anything about my variable? $\endgroup$ – Paze Jan 4 '20 at 21:31
  • $\begingroup$ @Paze I have already shown you how the value of "Odds" relates to the frequency of your variable ShuntEffekt. $\endgroup$ – AdamO Jan 7 '20 at 16:52

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.