| Description: |
International audience ; Consumers are more and more concerned about the living conditions of farm animals. As a consequence, numerous trade groups (producers, distributors, retailers, chain restaurants) have created certification schemes which include animal welfare. These initiatives aim at allowing consumers to choose ‘high welfare standard’ products, but the assessment of welfare varies from one scheme to another. There is no European standard to assess animal welfare and to bring that information to consumers. The EU project Welfare Quality R seeks to fill this gap for cattle, pigs and poultry, by proposing an overall assessment system based on multicriteria analysis and widely accepted by stakeholders. In this paper we present a possible contribution of MCDA to this problem. This application has been run for cattle and will be extended later to other animal species. Animal welfare is a multidimensional concept, including several aspects like good health, comfort, expression of behaviour, and so on. Thus, the overall assessment of animal welfare on farms is a multicriteria evaluation problem that needs a constructive strategy to compound information produced by multiple animal-based measures. We propose a methodology based on the construction of 12 subcriteria, grouped to form four criteria (good feeding, good housing, good health, and appropriate behaviour ) essentially to facilitate communication toward consumers, as defined in Welfare Quality R [1]. Each subcriterion is expressed on the same utility scale [0, 100] where 50 represents a neutral element. The aggregation process includes several steps. First, using several data-sets of farms graded by experts on each criterion, we derive the mathematical model linking criterion values to subcriteria values. This step is performed using the Choquet integral, the capacity being obtained by minimization of the overall quadratic error over examples provided by different experts. The second step of the aggregation process consists in producing an overall . |