Comment on “A Note on Collinearity Diagnostics and Centering” by Velilla (2018)

Abstract

This contribution clarifies the debate between Christensen (2018) and Velilla (2018b) concerning the appropriate calculation for variance inflation factors (VIF). Thus, we conclude that the doubts raised by Christensen (2018) in relation to expressions (16) and (17) of Velilla (2018a) are due to the measure used by Velilla (2018a) as VIF is not really VIF because the transformation indicated in (9) by Velilla (2018a) does not imply that the exogenous variables present a zero mean. In addition, we found the proposal of Velilla (2018a) appropriate to detect collinearity between the intercept and the independent variables by replacing the identification of the VIF with the notation of the index of collinearity given by Stewart.