SISTI: a multicriteria approach to structure complex decision problems

A decision aid process should be the result of interactions between analysts, decision makers, knowledge sources and stakeholders. However, decision aiding is sometimes required when the problem situation is so complex and new that a decision process is not yet activated and a formal decision system does not exist. Any type of interaction with the participants in the decision process becomes very difficult or even impossible. SISTI is a methodological approach that can be used to deal with these situations; it involves cyclic applications of a multicriteria method and the analyses of the results of these applications. Draft models and unprocessed data are used in the first applications, and sensitivity analysis is used to orient and control the model evolution. Several possible paths and modelling hypotheses can be iteratively proposed and tested in this knowledge structuring context, until the conceptual structure of the problem and its model are considered consistent with the situation. SISTI is a SImulated decision aid approach, because the decision process and system are not yet activated and a decision aid process is not oriented towards an immediate decision. However, it may also be described as a STImulating approach, because the study is developed together with a few actors that perceived the need to understand and propose structured elements for later phases of a still not activated decision process. SISTI is not a new multicriteria method, but rather a methodological modelling approach that uses a multicriteria method to structure a new and complex problem and to elaborate and validate a new model, when decision makers do not exist or cannot participate, or do not want to be involved in the decision aid process. This approach has been applied several times to clarify and structure complex situations and has been made more and more general so that it can be proposed to young practitioners to help them understand what a “good“ model is and how the robustness of their conclusions can be improved.