Les modèles d'analyse des comportements à risque face à l'infection à VIH : une conception trop étroite de la rationalité

Moatti (Jean-Paul), Beltzer (Nathalie), Dab (William). - Analysing Unsafe Behaviour in Face of HIV Infection. The Limits of Rationality Social science research on the prevention of HIV infection has pointed to correlations between individuals' knowledge and beliefs about AIDS on the one hand, and their reactions - individual or collective - to infected persons and the risk of transmission, on the other. However, it has also shown that the connection between these correlations and actual behaviour has remained ambiguous. As is the case in other areas of preventive medicine, our results reinforce the view that improved information is not in itself sufficient to change individual behaviour, and thus reduce risk. The international literature that deals with the determinants of exposure to the risk of sexual transmission of HIV has been dominated by existing social-psychological models (the Health Belief Model, or social learning theory), and these models have been used with some success in the' community' implementation of preventive measures in San Fransisco's homosexual and bisexual communities. This paper points out the limitations of these models in explaining or predicting behaviour related to the risk of HIV transmission. It shows that they have relied implicitely on a reductionist notion of individual rationality, in which rationality is equated with complete avoidance of risk and with concern for absolute safety. Comparing these social- psychological models with models of expected utility - standard in the micro-economic analysis of behaviour under conditions of risks and uncertainty - and applying this theory to ACSF survey data, makes it possible to identify other underlying reasons for the persistence of individual exposure to the risk of sexually transmitted HIV, and to show that 'no- change' or 'intermediate change' strategies that do not give complete protection against HIV risks may yet prove to conform better with individual effective rationalities. Finally, the paper suggests some new directions for behavioural analyses that are less machinistic, and which would account better for the specific context and temporal dy- namics in which exposure to the risk of HIV occurs. Such analyses would benefit from the latest theoretical developments in micro-economics and social psychology.