How Can Dashboards Present Uncertainty Accessibly?
From the earlier reference collection. Read the current software blog.
Show uncertainty as part of the result, with visible bounds, plain-language explanations, and labels that remain meaningful without color.
Show the range, not just the estimate
A dashboard number can look exact even when it is an estimate. If a model predicts tomorrow’s river level as 2.1 meters, readers need to know whether that figure is a fixed measurement or a central estimate with a plausible range. A shaded band around a line can communicate that range visually, but the band should have a legend and a nearby text explanation.
For example, a forecast might show a central estimate of 2.1 meters and an interval from 1.8 to 2.5 meters. Explain in plain language what the interval represents and, if known, how it was produced. Do not imply that every possible outcome is contained in the band unless that is what the method supports. The description should distinguish uncertainty about the value from ordinary variation in the observations.
Make the meaning available without color
A pale blue band on a chart may be difficult to distinguish from its background or invisible in a monochrome printout. Add a direct label such as “estimated range” and use line styles, boundaries, or patterns that reinforce the meaning. Keep text contrast high enough to read, and ensure the center line remains visually distinct from the interval edges.
The same information should be available in text or a data table. A table could list the time, central estimate, lower bound, and upper bound as separate columns. A concise chart description might say, “The estimate rises through the morning; the range widens after noon.” This helps readers who do not see the visual and helps everyone understand the important pattern.
Explain what can change the range
Uncertainty may grow because observations are sparse, the process is variable, or the forecast looks farther into the future. If the interval widens over time, explain the reason when the method supports one. If the dashboard combines different sources or assumptions, identify the relevant inputs in a method note rather than burying caveats in a tooltip.
Be careful with labels such as “confidence” or “likely.” They can mean different things to different readers. State whether the band is a model interval, a range across scenarios, or another kind of summary. If the source does not define a probability interpretation, do not invent one.
Keep the explanation close to the result
Place the key explanation beside the chart, with a link to fuller methodology for readers who want more detail. A useful presentation lets someone answer three questions quickly: what is estimated, how wide is the range, and what should the range mean for a decision? When those answers are visible in text as well as graphics, uncertainty becomes information readers can use rather than decoration they must guess at.