Consensus selection of alternatives (CSA): a score-level consensus framework for integrating multiple MCDM rankings

Multi-criteria decision-making (MCDM) applications often produce different rankings when multiple evaluation procedures are applied to the same alternatives. This creates an integration problem in settings where analysts intentionally retain methodological plurality instead of selecting a single method ex ante. Classical aggregation rules such as Borda Count provide a simple solution, but they compress score differences and may discard useful information contained in method-specific outputs. This study proposes the Consensus Selection of Alternatives (CSA), a score-level consensus framework that integrates heterogeneous MCDM outputs on a common normalized scale and assigns method weights according to inter-method agreement. Unlike rank-only fusion rules, CSA preserves score differences while producing a single interpretable final ordering. The framework is evaluated on a reproducible defense-inspired benchmark with 151 alternatives, 14 main criteria, and 55 sub-criteria. Entropy, WASPAS, and PROMETHEE are used as constituent ranking procedures, while Borda Count and mean normalized score aggregation serve as benchmark comparators. The results show that CSA yields a coherent consensus ranking, remains broadly stable under weighting, normalization, omission, and removal scenarios, and displays bounded sensitivity to normalization choices. An external validation on an open PV technology selection dataset further indicates that the framework can be transferred to a distinct application domain without producing anomalous ranking behavior. Overall, the study contributes a transparent and reproducible consensus layer for multi-method MCDM settings in which preserving score information and reconciling heterogeneous rankings are both decision-relevant.