I am running a LOGIT where all my independent variables are dummies. I get
the expected signs for them. But its ability to separate 1 from 0 is
extremely poor. In fact my estimated probability scores turn out to be
between 0 and .33. One another piece of information that may help is that
the split between 1 and 0 (dependent variable) is 0.022/1.... I have very
few 1s.
So the questions are:
1. Anything that one should be concerned when working with only quantitative
variables?
2. Any other ways of handling data where the dependent variable split is
just 0.022/1 That is I have in 5710 observations I have 127 bad (=1) and
5583 good (=0).
