Considering data uncertainty in the water and sanitation sector: application to large number of alternatives and criteria

Abstract

Priority-driven and target-focused management of water and sanitation services is essential to achieve long-term sustainability. However, in the context of water and sanitation, statistical uncertainty is an inevitable feature, as surveys are the primary sources of the information used in the decision making process. Therefore, it needs to be considered to have a greater confidence in the decisions made based on the collected data. Outranking approaches provide an opportunity to handle imprecision and uncertainty characteristic of water management problems. In particular, ELECTRE-III uses pseudo-criteria to consider uncertainty in the decision-maker’s preferences, but does not take into account statistical uncertainty. A new interpretation of ELECTRE-III, in which pseudo-criteria are linked to the data uncertainty, is presented in the paper. The modified outranking approach (E3M) is used to evaluate water and sanitation services in Kenya and rank its 338 subdistricts according to their performances on 9 different criteria. E3M proved to be an effective ranking tool, providing a detailed hierarchy between the subdistricts, even when dealing with hundreds of alternatives. However, a quadratic dependence between the number of subdistricts ranked and the computational cost of the method was found, indicating that for larger number of alternatives (order of thousands), the time required would rise to the order of hours. Future work should focus on optimising the method’s algorithm and allowing its scalability to thousands of alternatives.