Pivot-Linked Elicitation: A unified approach to hierarchical weight elicitation and its application to best–worst methods
针对多准则决策中层级权重启发的问题，提出枢轴关联式启发框架，通过少量枢轴准则连接各组，整合组内与组间偏好信息，并在最佳-最差方法中实现，保留层级认知优势的同时提升整体一致性。
We study weight elicitation in Multiple Criteria Decision Analysis (MCDA) when decision criteria are organized hierarchically. We show that the conventional practice of eliciting weights separately at each level can lead to practical and conceptual limitations, including reliance on abstract upper-level judgments and the fragmentation of consistency across independent sub-models. To address these issues, we propose the PIvot-Linked Elicitation (PILE) framework, which links groups through a small set of pivot criteria and integrates intra- and inter-group preference information into a single model. We instantiate PILE for best–worst methods, deriving nonlinear and linear PILE–Best-Worst Method (BWM) formulations, and illustrate the approach on two published use cases. The results demonstrate how PILE preserves the cognitive benefits of hierarchical structuring while enabling a more holistic and consistency-aware elicitation.