Express uncertainty with numerical probabilities or intervals

Replace verbal uncertainty words with explicit numerical values or intervals on the display. For example, show a percentage probability, a ±X% variability estimate, or a credible interval instead of labels such as “rare,” “good evidence,” or “stable measurement.”

Numerical uncertainty labels give readers a concrete readout instead of forcing them to guess what a field-specific word means. When the display uses only verbal qualifiers, different readers can leave with very different levels of confidence from the same message.

Mechanism: Explicit numbers reduce reliance on private language conventions and make the intended amount of uncertainty easier to extract from the display.

Evidence: The paper reports that laypeople often misinterpret verbal expressions of uncertainty, while most people can use explicit quantitative expressions well enough to extract their main message; it also argues that imprecise summaries leave observers guessing about uncertainty (Fischhoff & Davis, 2014).

Notes: The paper treats credible intervals as the preferred quantitative summary when uncertainty goes beyond observed variability alone.

• User Goal: Judge how uncertain a result or recommendation is.
• Data: Uncertainty can be summarized as a probability, variability estimate, or interval.
• Chart Setting: The display currently communicates uncertainty through labels, captions, legends, or boxed summaries.
• Audience: Readers are outside the field or do not share its verbal uncertainty conventions.
• Success Criterion: Readers can state the amount of uncertainty without translating a vague word.

Break it when: The intended audience already shares the field’s verbal conventions and user testing shows the verbal term is interpreted consistently. Why: The paper frames misunderstanding as a problem when recipients do not know what experts mean by seemingly common terms, and it treats clarity as an empirical question.

Sacrifice: You give up some brevity. Risk: A bare number can suggest more precision than intended. Mitigation: Pair the number with the supporting uncertainty rationale or interval width rather than presenting a lone precise-looking value.

Mistake: Replacing one vague uncertainty word with another vague uncertainty word. Why it fails: Readers still have to infer magnitude from unstated language conventions.

Failure Sign: The display says “rare,” “likely,” “good evidence,” or similar phrases without a number. Quick Check: Can a reviewer point to a numerical probability, interval, or ± estimate for each uncertainty claim? Stronger Test: Ask representative readers to paraphrase the uncertainty in their own words and compare whether their interpretations converge.

• Replace each verbal uncertainty label with a percentage probability or interval.
• Add a ±X% variability estimate where the display currently says a measure is stable or variable.
• Add a credible interval when the display summarizes an estimated effect or outcome.
• Add the short rationale for that number when the estimate could otherwise look more precise than intended.