In PROC HPREG diagnostics, a predictor has VIF = 18. What does this indicate and what is the typical remediation?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because VIF measures how much coefficient variance is inflated by multicollinearity. VIF = 18 far exceeds the common threshold of 5-10, indicating severe multicollinearity.
Full explanation below image
Full Explanation
B is correct because VIF measures how much coefficient variance is inflated by multicollinearity. VIF = 18 far exceeds the common threshold of 5-10, indicating severe multicollinearity. Standard remediation includes removing one of the correlated predictors, creating composites, or applying ridge regression. A is wrong; VIF is not a percentage. C is wrong; VIF > 10 is widely considered severe. D is wrong; VIF diagnoses multicollinearity, not heteroscedasticity.