Using indifference information in Robust Ordinal Regression

In this paper, we propose an extension to Robust Ordinal Regression allowing it to take into account also preference information from questions about indifference between real and fictitious alternatives. In particular, we allow the decision maker to suggest a new alternative that is different from the existing alternatives, but equally preferable. As shown by several experiments in psychology of the decisions, choosing between alternatives is different from matching two alternatives since the two aspects involve two different reasoning strategies. Consequently,by including this type of preference information one can represent more faithfully the DM’s preferences. Such information about indifference should narrow down the set of compatible value functions much more quickly than standard pairwise comparisons, and a first simple example at least indicates that this intuition seems to be correct.