Overcoming multicollinearity by deducting errors from the dependent variable Academic Article uri icon

abstract

  • Abstract A method to overcome multicollinearity, based on deduction of the stochastic disturbance from the dependent variable, was developed. This method resulted in a marked increase in the coefficient of determination—R 2 associated with the “corrected dependent variable”. Simulation results revealed that the proposed method yielded good estimation of those stochastic disturbance. The resulting “corrected” dependent variable may be incorporated into some of the existing methods for overcoming multicollinearity. Application of least-squares regression to the “corrected” dependent variable yields estimators of normal size, and the predictive ability of the derived equation is excellent.

publication date

  • January 1, 2001