80

RÉSUMÉ Cet article s’attache à distinguer, dans une perspective synthétique, les différents modèles d’évaluation des risques élaborés dans le champ des disciplines psychologiques. À défaut d’une définition générique du risque, rendue inopérante par la variété des situations concernées, l’analyse s’appuie sur les propriétés fonctionnelles communes aux situations à risque pour les caractériser, en particulier comme multidimensionnelles et incertaines. Trois grands paradigmes permettant l’évaluation sont ensuite présentés, analysés et discutés de façon à mettre en évidence les options qu’ils retiennent. Le premier repose sur la détermination de l’utilité espérée ou de valeurs voisines, et permet, en combinant plusieurs apports, de déterminer un risque acceptable. Le second, connu sous l’appellation de “ paradigme psychométrique ”, s’appuie principalement sur des représentations et des valeurs sociales. Le troisième se réfère aux processus cognitifs utilisés pour traiter les informations et parvenir à une évaluation. L’article se termine par l’analyse de quelques questions épistémologiques relatives à la diversité des démarches d’évaluation.

ASSESSING AND MODELLING RISKS : CONTRIBUTIONS AND LIMITS OF SEVERAL PARADIGMS IN SECURITY DIAGNOSIS The importance of risk evaluation, either at an individual or a collective level, was stressed by the numerous risky situations with which we are daily confronted in personal, occupational and social life. Risk evaluation becomes very difficult because of the extreme diversity of risky situations as well as because of differences of diverse orders identifiable in the evaluation of the consequences of risk. Because it is impossible to have a satisfactory definition of risk, the present article proposes that risk evaluation strategies be studied in terms of five functional characteristics of risky situations : multidimensionality, need to integrate information, uncertainty, presence of objectives and constraints, need to evaluate long term or short term consequences (or effects). A differential analysis lead the authors to identify three paradigms which generate different models according to their structures and the indexes that they consider as bearing information in each situation. Three paradigms of risk evaluation were then described, analysed and discussed in order to emphasize the different options chosen. The first one concerns the determining of expected utility or of related values, and allows, by combining several contributions, to determine what is called “ acceptable risk ”. The second one, known as the psychometric paradigm, is principally based on social representations and values, and shows that risk is socially determined, for the most part, from two orthogonal references showed by factorial analysis. The third paradigm is based on cognitive processes used to process information and to do an evaluation and allows to show the incidence of heuristics in decision making and the incidence of biases. The authors concluded their purpose by noting that none of these evaluation strategies is able to pretend to be the most efficient in all circumstances and in all contexts. They stressed the importance of the badly known meta-cognitive contributions, and deplore the rarity of their use in risk evaluation.