A DEA-inspired procedure for the aggregation of preferences

摘要

Among the most commonly applied methods for the aggregation of individual preferences, weighted scoring rules associate each alternative with a weighted sum of votes received and then rank them in terms of this aggregate value. Some authors have argued that it is not possible to fairly evaluate a set of alternatives by only considering an externally imposed weight scheme. For this reason, researchers have developed certain procedures in which the weights associated with the votes become variables in the model. Data Envelopment Analysis (DEA) represents one class of such models.In this paper, I propose a new preference-aggregation procedure. The procedure maintains the philosophy inherent in DEA, allowing each alternative to have its own vector of weights, but also introduces a new objective in the evaluation, the optimisation of the rank position in which the alternative is placed, and to avoid the problem of diverse weights by determining a social ranking that uses a common vector of weights.