Definition
A statistic that measures the degree of multicollinearity present among a set of Independent Variables. VIF is closely related to R2 by the following equation:
VIF = 1/(1-R2)
Large values of VIF indicate a high degree of multicollinearity, but how large is a matter of debate among practitioners. Some consider values of 4 and above as too large, others are more liberal and use 10 or higher.
See Also
Multicollinearity