Multicriteria evaluation applied to farm animal welfare: difficulties and solutions from the Welfare Quality® project

More and more often European schemes for farm certification imply the fulfillment of animal welfare. The Welfare Quality (R) project aimed at proposing an overall assessment model of animal welfare at the unit level (farm or slaughterhouse). This model relies on a multicriterion evaluation. First, 12 welfare criteria, grouped into 4 principles, were defined. After identifying the measures needed to check that an animal unit complies with the various criteria, the model was progressively constructed so as to finally classify animal units into four categories reflecting the level of animal welfare (from excellent to very poor). We used multicriteria evaluation methods, trying to avoid the pitfalls of a naive aggregation of the information (e.g. weighted sums authorizing full compensations between criteria, not compatible with the concept of welfare). Different persons (animal and social scientists, stakeholders) were consulted. They were presented virtual datasets to which they had to assign scores. The model was developed and parameterized so as to reflect their opinions. A post-hoc analysis of the results showed that the model gives the priority to animals in bad condition while also taking into account the average state of the group of animals, and that compensation between criteria are strongly limited. Following this rigorous methodology we were able to propose an evaluation of a multidimensional concept. Such an exercise requires value judgments that must be clarified and modelled. The methodology followed to construct this animal welfare assessment model could be transposed to the assessment of other multidimensional concepts like the sustainability of production systems.