This video covers how you can run a regression model when you have a binary (a.k.a. dummy a.k.a. indicator) dependent variable. I go through the pros and cons of linear probability models, probit, and logit.
Thank you, no one explained it so clearly, step by step, the previous step making it easier to understand the next step. Really got the Understanding of what is going on and not simply learned the formulas
I don't know pure-stats books well enough to compare them, but for introductory econometricians I'd say Wooldridge for the technical side, or Bailey for the intuitive side.
Thank you for the explanation. So for the Logit, the 0.53 is in the index. As its a curve between 0-1. The change of one unit of x will vary with 4 units of x. How would one calculate that?
@@NickHuntingtonKlein Hi thanks for the reply. Actually no specific portion of the video, just a general question. As logit is nonlinear, I was curious how one measures changes in x's. Like let say if you ran a logit regression and you got 0.53 for your caffeine coefficient [one could have other independent variables, but those are held constant] Your dependent variable is binary, sleep/awake. How would one measure the effect of a change of caffeine[1 cup, 3, cups, x cups]? Thanks again
@@emptyxnes ah I see. That's what marginal effects calculations are for. I can't remember if I cover them in this video or the next one but I have a video on it.
