PSD.9:4.5 - Elicit trade-offs over consequences, not criterion labels

Before assigning weights, ask which concrete consequence profiles the value holder prefers and why. For an additive value model, a weight concerns the value of an improvement over a stated consequence range, not the abstract importance of the criterion’s name. Changing that range can require elicitation again.

Test whether a proposed compensatory model fits the expressed preferences. If one consequence’s value depends on another, retain that interaction or use a suitable model. If a participant refuses the trade-off, do not encode refusal as an extreme but negotiable weight unless that faithfully represents the judgement.

Several forms can be useful: qualitative preference statements, a priority order, threshold rules, an additive or non-additive value model, or an outranking relation that can preserve incomparability. Choose the least burdensome form that preserves the decision-changing distinction. The domain Method supplies its elicitation and validity conditions; naming MCDA does not satisfy them.

Keep each materially different value account attributable. A common account requires an explicit and legitimately governed way to combine or reconcile the judgements. Voting, averaging, or a sponsor’s preference is not a default aggregation rule. Where several preference models remain compatible with the expressed judgements, retain that family rather than fitting one arbitrary point estimate.