A JUSTIFICATION OF THE AVERAGING METHOD FOR ABSTRACT PARABOLIC EQUATIONS