Best-Worst Method (BWM) is a structured approach for eliciting criteria weights in multi-criteria decision-making. This review paper revisits the original BWM formulation and procedural logic, and discusses behavioral motivations for dual anchoring and structured elicitation. It then synthesizes major methodological developments, covering linear and multiplicative formulations, interaction modeling extensions (nonadditive models), Bayesian formulations, group-aggregation models, tradeoff-based elicitation, parsimonious and disaggregation-based variants, sorting extensions, and fuzzy and belief-based treatments of imprecision and epistemic uncertainty. Because the reliability of inferred weights depends on judgment quality, the paper consolidates and compares consistency checking approaches and discusses their implications for practice. Representative application domains are reviewed to illustrate how BWM is deployed in practice. Finally, future research directions are outlined, emphasizing behavioral validation, integration into complete decision pipelines, scalable elicitation support, and cautious human-AI co-production that preserves problem-specific preference meaning. These directions also include transferring BWM’s anchored elicitation principle to other preference elicitation approaches, including MACBETH, UTA, and conjoint analysis. • Revisits BWM’s dual-anchored elicitation and behavioral rationale. • Synthesizes BWM extensions: linear, multiplicative, Bayesian and interaction models. • Compares output- and input-based consistency ratios for BWM judgments. • Maps representative application domains and growth of BWM research over 2015–2025. • Sets agenda: behavioral tests, full pipelines, scalable elicitation and human-AI use.
Jafar Rezaei (Sat,) studied this question.