Uncertain interval TOPSIS and potentially regrettable decisions within ICT evaluation environments

Abstract

It is generally assumed that the rankings provided by Multi-Attribute Decision-Making (MADM) techniques are definitive. Once the ranking is delivered, decision makers (DMs) are expected to choose the first alternative and dismiss the remaining ones, concluding the application of the corresponding model. The MADM literature has incorporated fuzziness and imprecision to its models to deal with evaluation uncertainties but has not accounted for its consequences defined in terms of regrettable choices. That is, MADM models do not consider the possible consequences of having chosen an alternative whose actual characteristics do not correspond to those expected by the DM. This paper aims at designing an integrated MADM framework with interval variables where the DM is allowed to modify the initial alternative chosen after observing the realizations of its characteristics. In order to do so, sequences of alternatives including the initial choice as well as subsequent alternate choices should be ranked in place of single alternatives. We analyze the combinatorial decision environment that arises from defining and evaluating sequences of choices by accounting for the whole set of potential realizations and any subsequent change in the alternatives selected. The TOPSIS method is used to design the integrate evaluation framework producing the final ranking. A case study analyzing the entry decision of a firm within a group of European countries based on their levels of ICT development is presented. We illustrate how the countries selected and their order may differ substantially when accounting for the complementarities existing among them. Moreover, the selection process and any subsequent decision vary with the number of modifications considered relative to the initial country selected. The results obtained are of interest not only to firms facing a similar problem, but also to DMs or managers dealing with strategic selection processes where the wrong choice of alternatives may lead to increasingly complex sequential disruptions.