Forward stepwise regression: Round 3

- [Instructor] This movie continues my explanation…of the rounds in forward stepwise…logistic regression modeling.…This time, we are focused on round three.…Round three is particularly important…because it's where you achieve your final, final model.…You want it to be a robust model.…I'll explain what that means.…You do that by trying to break the working final model,…also called the candidate final model…we talked about in the last movie.…We did this in logistic regression too,…but I wanted to go over it again here…to make sure you understand…exactly what we are trying to achieve.…

So at the start of round three,…you have a candidate final model,…also called a working final model…that came out of round two.…All of the covariates in it should be significant…or close to significant.…You run the show, you make the call.…If you think a covariate is empirically important,…meaning you think it just belongs there, keep it.…This can happen when a covariate is almost significant…like with a P value of 0.08, between 0.05 and 0.10.…