Abstract (EN)

The aim of this paper is to examine on a non-theoretical ground to what extent outranking and MAUT decision-aid approaches differ.For this purpose, we chose a study using utility theory conducted by Keeney and Nair, dealing with a nuclear plant siting problem. We had to determine what the study would have been if it had been conducted with the use of the Electre III model.In this ‘experiment’, we are not interested in the practical problem for its own sake but in:•- the way to build criteria;•- the representation of a decision-maker's preferences;•- the use of the model and the nature of the derived prescription.Confronting the two models, we study:•- the differences that they induce when facing a real problem and building a set of data,•- their respective part of arbitrariness, weakness, realism, robustness,•- the convergence or divergence of their results,and insist upon the differences between ‘descriptive’ and ‘constructive’ approaches.