logistic regression

JFI

New Member
#1
Hi
I am using a logistic regression to determine siginificant variables for dependent variable as well as to develop a function to predict the devendent variable.

I have not done regression modeling for a while and need help with refreshing my memory.

1. I have only one independent variable and an intersection left after working thourgh step wise method. These two are siginificant, but correlated. How should I deal with this type of case - Remove the indipendent variable and leave only the intercept?

2. If the case above, what can I conclude in terms of independent variables which have siginificant influence on the dependent variable?

3. In order to build a predictive model, should I keep all the independent vairable (i.e. full model) rather than reduce the model with a stepwise method? Well, the result of the step wise is, as I mentioned, only the intercept.

Thank you very much,