I am new to logit regressions and have some questions which might be pretty basic for most of you on this forum. My main question is regarding forcing the dependent variable to 0 or 1 in order to perform logit regression using STATA. I think eventually I need to do maximum likelihood estimation, but logit regression seems like a good first step.
I am trying to estimate the adoption of a technology. I currently have data on % people who have currently adopted the technology, call it A (dependent var). I have explanatory variables x, y (and many more)...none are binary.
In order to perform logit regression, I need to force the parameter A to 0 or 1, correct? (i.e.in order to perform logit a x y)
The way I tried it was ln(A/(1-A)). If this result is negative, A is 0, if positive, A is 1-- is my understanding right? In my case, all the results are negative and so all the As are zero. This shouldn't matter I hope.
If someone can clarify this it will be immensely helpful.
Advertise on Talk Stats