The fourth honesty check is whether uncertainty, missing data, and sample size are labeled at all

What it is

The fourth honesty check is the visibility of quality information: can readers learn from the chart how uncertain the estimate is (interval), how much data is covered (missing proportion), and how many observations the conclusion rests on (sample size)? These three are the direct measures of "how thick the foundation beneath the number is"; missing any one leaves trust calibration without a basis—absence is not nonexistence, only an improper transfer of calibration duty to the reader.

Why it happens

This check consolidates the earlier quality-annotation knowledge (error expression, missing-data expression, sample-size annotation) into one audit action: answer three questions per chart—does the estimate have an interval (error bars / band / ± annotation)? Does the data have gaps (missing markers / coverage rate)? How large is the sample (n / coverage)? All three "yes and visible" means pass. The pass standard must match the metric type: counts from large samples may have intervals too narrow to display, but the sample-size label remains; survey and experimental results cannot omit uncertainty; real-time pipelines cannot omit coverage. The check's second value is exposing "silent certainty": a chart with neither interval nor sample size asserts certainty to its readers—the audit forces the maker to confront whether that assertion holds.

Where it stops holding

Annotation form may be minimal but never absent: when space is tight, one line ("n=48,210, coverage 97%") carries sufficient information; what is checked is information availability, not visual size. The necessity of the three labels scales with the metric's evidentiary stakes—decision-facing metrics require all three; internal exploratory intermediates may relax, but the relaxation must be explicit (the check record notes "internal use, interval annotation waived") rather than a silent skip. For multi-source merged charts, audit the three per source, since uncertainty and coverage can differ drastically across sources.

Applying it

• Check action: per chart, answer "interval labeled? missing/coverage labeled? sample size labeled?"—all yes means pass.
• Mark any non-applicable item explicitly with its reason ("full-population data, no sampling; sample size equals total"); blank skips are not allowed.
• Add the three questions as required fields in the publishing template.
• Verification: sample 10 outward-facing charts and compute the pass rate across the three questions; any unavailable item indicates a systematic gap in quality annotation.

Related

• Same group: U10.04.1 Check whether axes start at zero or are truncated or broken · U10.04.2 Check whether data range, filter conditions, and exclusions are stated · U10.04.3 Check whether title and annotation claims are supported by the graphic itself · U10.04.5 Check whether the conclusion changes under another reasonable rendering · U10.04.6 The checklist runs before publication, not after being challenged
• Nearby: U8.01.3 Omitting uncertainty claims certainty · U8.04.3 Missing proportion affects conclusion credibility
• Search terms: quality annotation · uncertainty labeling · sample size disclosure