What tolerance value signals serious multicollinearity problems?

• A. Values below 0.2 indicate serious problems, with 0.1 as a strict threshold.
• B. Values above 0.9 indicate strong multicollinearity.
• C. Tolerance is unrelated to multicollinearity.
• D. Values below 0.1 indicate serious problems, though values below 0.2 are worthy of concern.

Answer: (D)

Tolerance shows how much of a predictor’s variance remains after accounting for the other predictors. When this value is very small, the predictor is almost a duplicate of the others, which makes the regression coefficients unstable and inflates standard errors due to multicollinearity. Because of that, very small tolerances are taken as signaling serious multicollinearity problems. A tolerance that is still small but not as extreme can still raise concern, while a high tolerance indicates that little multicollinearity is present. Tolerance is directly related to multicollinearity, not unrelated. So a statement describing that very small tolerances signal serious problems, with a slightly larger threshold still warranting caution, best captures how tolerance is used to assess multicollinearity.