When one wishes to formulate, evaluate and implement public policies, the existence of a plurality of social actors, with interest in the policy being assessed, generates a conflictual situation. How such a conflict should be dealt with? This paper defends the thesis articulated in the following points: (1) Different metrics are linked to different objectives and values. To use only one measurement unit (on the grounds of the so-called commensurability principle) for incorporating a plurality of dimensions, objectives and values, implies reductionism necessarily. (2) Point (1) can be proven as a matter of formal logic by drawing on the work of Geach about moral philosophy. This theoretical demonstration is an original contribution of this article. Moreover, here the distinction between predicative and attributive adjectives is formalised and definitions are provided. Predicative adjectives are further distinguished into the new categories of absolute and relative ones. The new concepts of set commensurability and rod commensurability are introduced too. (3) Weak comparability of values, which is grounded on incommensurability, is proposed as the main methodological foundation of well-being evaluation. Incommensurability does not imply incomparability; on the contrary incommensurability is the only possible way to compare societal options under a plurality of policy objectives. Weak comparability can be implemented by using multicriteria evaluation, which is a formal framework for applied consequentialism under incommensurability.