The ranking is most sensitive to the weight assigned to criterion C2. An important finding of this sensitivity analysis is that the top two SAW ranked alternatives (a3, a1) do not vary regardless of the weights assigned to the criteria in the SAW method (assuming that they remain within the DM provided ranges shown in Fig. 5). The required weights were elicited from the DM by MCPUIS, being automatically re-scaled so that they sum to 100 to facilitate the analysis. As already mentioned for ELECTRE I, “alternative a outranks alternative b” is verified using the concordance (a majority of criteria supports it) and non-discordance (no criterion is strongly opposed to it) principles. The user-defined thresholds of concordance and discordance of ELECTRE I play an important role and may influence the final outranking relations. The variation of those parameters was performed and the results obtained are presented in Fig. 6, where the set of not outranked alternatives is represented for several pairs of threshold values located in the 4 regions of the figure. It can be observed that alternative a3 is never outranked by any other alternative for any pair of values of those thresholds. However, a1 is always outranked by, at least, one other alternative. This conclusion also remains valid for weight changes (not presented here due to space limitations). As there is no “best” MCDA method [7], MCPUIS includes three methods with distinct underlying assumptions. MCPUIS can then be used to determine how sensitive the results are with respect to the MCDA method instantiated with different sets of preference information and technical parameters.