09-18-2014 03:30 PM
I ran a simple proc reg step, which provided low VIF values (less than 4) for all of my variables. The intercept was not significant (with a p-value of 0.84), so I was advised to drop the intercept from my model. Running the same proc reg, but this time including the 'noint' option, two of my variables have very high VIF values (5800 and 6050).
Any advice on what may be causing this?