Abstract
In Group Decision Making, there are situations in which the decision makers may not be able to provide his/her opinions properly and they could contain contradictions. To avoid it, in this contribution, we present a new selection process to deal with inconsistent information. As part of it, we use a method based on granular computing to increase the consistency of the opinions given by the decision makers. To do so, each opinion is articulated as a certain information granule instead of a single numeric value, offering the necessary flexibility to increase the consistency. Finally, the importance of the decision makers' opinions in the aggregation step is modeled by means of their consistency.