Logistic Regression


New Member
I am a distance learning student with a project on hand. I am new to stats and would be seeking help from colleagues. Thanks
I am confused about two issues on Logistic Regression and would be grateful to receive some clarification.
1. On running the LR, I am having both the Chi-square and the Hosmer & L test significant, i.e p=.000 Does that invalidate the LR exercise? Can I still use the LR equation for predictive purposes?
2 On running the LR with five independent variables, one of them did not satisfy the Wald significance test and was dropped out in the forward selection to give a final partial model. The question is: In the designation of the logistic regression equation to quantitatively predict the outcome, should this “insignificant” variable form part of the equation? This would then mean that the coefficient of the full model be used rather than the coefficient of the partial model.