Page 1 of 1

Complete Separation on Probit

Posted: Mon Mar 16, 2015 4:25 pm
by id90525
Hi guys, I am facing a Complete Separation issue on my probit model and was wondering if a Firth Logistic Regression is possible on eViews. If so, how is this performed?

Also, when I regressed the dependent variable on all my independent variables the p-value became much larger than when I regressed them individually. Does this suggest multicollinearity? And if so, how could I perform a VIF test to pick out the offending variable?

Thank you.