THE DIMENSION
Expressing confidence conditionally and updating it when relevant information changes.
The dimension in context
Expressing confidence conditionally and updating it when relevant information changes.
You may make the limits of a judgment visible.
You may use definite language or lack feedback about uncertainty.
An illustrative work situation
A participant must recommend a course of action while two important facts remain unknown. They can hide the uncertainty to sound decisive or state it clearly and explain why one option is still reasonable. The task is to connect uncertainty with the action and the conditions for revision.
Acknowledging uncertainty does not require indecision. It can clarify which part of a recommendation is robust and which depends on a working assumption. A self-report about uncertainty does not measure probability skill or forecasting accuracy. The reflection concerns how the participant represents what is and is not known.
A small practice to examine
List the most consequential unknowns and describe how each could affect the choice. Identify a reversible action or a review trigger where possible. Explain the recommendation together with those conditions. Later, examine whether the unknowns were the right ones and whether the agreed trigger was used when information arrived.
Excessive qualification can make an actionable conclusion unclear.
A question for closer study
How can research distinguish transparent uncertainty from vague hedging that leaves no usable decision guidance?
Read the example alongside the research
The following editorial passages come from the linked framework chapters. They provide context for the original teaching example above; the example is fictional and is not a reported study finding.
A proposed study could collect several clearly defined predictions and preserve the information available when each was made. Researchers would examine probability judgments separately from participants' reports of how they formed them. One correct prediction cannot establish calibration, and a single missed event does not show that an uncertainty estimate was unreasonable. The study would also need to avoid selecting only memorable successes for review. A useful test of the questionnaire is whether people reporting explicit confidence checks define resolvable events and update their judgments when relevant information changes. Even if that relationship appears, it would not justify calling the inventory a forecasting-accuracy test. That stronger interpretation would require evidence about actual predictions, their selection, and the conditions under which they were made.
Continue: Confidence should be connected to evidence and feedback →
A proposed exercise could hold the decision evidence constant while varying the audience's stated preference. Researchers would compare which considerations participants include, omit, or qualify. They should not assume that every adaptation to an audience is distortion; differences in technical background can legitimately change how an explanation is expressed. The important question is whether decision-relevant content changes without justification. A governance review would also distinguish responsibility for making the choice from responsibility for explaining it. The inventory cannot certify a decision as ethically correct, legally sound, or professionally competent. Its narrower contribution is to prompt an inspectable account of assumptions, alternatives, consequences, and ownership that another person can question without having to accept the respondent's self-rating as proof.
Continue: Accountability should improve reasoning, not reward a performance of certainty →
SOURCES AND FURTHER READING
Kahneman & Klein - Conditions for intuitive expertise ↗Mellers et al. - Psychological strategies for winning a geopolitical forecasting tournament ↗Mellers et al. - The psychology of intelligence analysis: Drivers of prediction accuracy in world politics ↗Bruine de Bruin, Parker & Fischhoff - Individual differences in adult decision-making competence ↗Lerner & Tetlock - Accounting for the effects of accountability ↗