(a) Explain why it might be useful to include a quadratic term in a regression. (b) Explain why it might be useful to include an interaction term between two predictors in a regression. (c) Name a drawback to including quadratic or interaction terms in a regression.

(a) State the formula for a variance inflation factor (VIF) for a predictor. (b) Why does the VIF provide a more general test for multicollinearity than a correlation matrix or a matrix plot? (c) State a rule of thumb for detecting strong variance inflation.