I'm new in econometrics and I'm working in a probabilistic model and using Stata for this, but when I was going to compare the Logit and Probit I did not know which one win in this case, because there is a tie between them. If somebody could help me, I would be so grateful.

  • 1
    $\begingroup$ Your substantive conclusions will be the same for both, so it's not worth stressing over which one to choose. It's a toss-up. Either way you should use the margins command to arrive at interpretable results. It's totally fine in a paper to report from one of these models and say that results from the other model were essentially the same. $\endgroup$
    – Noah
    May 16, 2019 at 22:32
  • $\begingroup$ Thanks for your answer, excuse me for bother you again, I've choosen the logit model but I have some troubles to interpret the marginal effects when one of the independent variable is in ln and the dependent is a dummy. In this case dy/dx=0.0649153 and the x is 13.8467. The variable is lwage, I'm not how to analize this. Again, excuse me for ask you again. $\endgroup$ May 16, 2019 at 22:57
  • $\begingroup$ That means that at an lwage of 13.8467, the instantaneous rate of change in the probability of the event is .06491. For a small change in lwage, you expect the probability of the event occurring to increase by .06491 times that change. This effect will be different at different values of lwage, so it's a good idea to choose several values of lwage to see what the effect is at different levels, possibly plotting them using marginsplot. $\endgroup$
    – Noah
    May 17, 2019 at 15:34


Browse other questions tagged or ask your own question.