Following the experiments by Marsh and Parducci (1978) and Molina and Fabre (2000), this research investigates the role of a dichotomous pre- established norm in judgment tasks. This norm remains anchored at the middle of the scale, while the range-frequency compromise operates separately on each side of the norm. The originality of the present study rests on a crucial condition in which a discrepancy between the dichotomizing criteria of stimuli (the zero-point with positive and negative numerals) and ratings (the pass-fail value with exam grades) is introduced. In this case, only the rating criterion operates, and asymmetric endpoints about the norm do not produce symmetrical extensions of the range. The discrepancy condition was crucial in showing that the symmetrizing effect is not sufficiently robust to overcome a conflict between the two criteria, and therefore that this effect is not independent of all rating criteria.