Chart verification checklist: the reviewer's sign-off

What to take away

A chart reaches readers faster than the table behind it, so it needs its own sign-off once the numbers are settled. These checks cover tables, maps, dashboards and polls and charts. Record the reviewer, date, result and evidence path, and mark a source insufficient when the question exceeds the data.

What the reviewer receives

Four files go to the reviewer: the source file, the formulas, the chart data and the final version. Without all four, the reviewer can only judge how the chart looks, which is the weakest check there is.

A chart can look tidy and still rest on the wrong file, the wrong table or a stale release. Keep all four with the published version. When a later correction is needed, they show which number changed and which distributed copies carry it.

A figure copied by hand from a PDF can carry a slip into the chart without anyone noticing. Saving the file as retrieved lets the reviewer compare like with like, and it shows when the producer later revised a value. The reviewer also checks whether each figure is a count, a rate or proportion.

Is the data fit for this claim?

Fitness comes before design. A dataset can be reliable enough for one narrow claim and inadequate for another, so the reviewer states the claim first. The sign-off records that judgment instead of applying a universal quality label.

The U.S. Government Accountability Office's data reliability guide frames reliability around accuracy, completeness, and applicability for the intended purpose, using a risk-based assessment. Say in one line what the chart is meant to show, and for which place and period, because every later check is measured against that line.

The checks below apply to that line and to nothing wider.

What the reviewer reproduces

The reviewer works from the source file, not from the finished graphic. That is the only way to catch a transcription slip or a formula that quietly drops a category.

Where the reviewer and the author disagree, the source file settles it. Record the disagreement and its resolution beside the evidence path, so the next person who opens the file sees why the chart reads as it does and what was changed.

Gaps and uncertainty on the graphic

Missing data is part of the picture, so draw it as missing. Suppressed and unavailable values are not zeros, and breaks in series are displayed rather than connected silently. Small subgroup estimates are flagged or omitted under the source's rules.

Uncertainty belongs on the graphic, not only in a footnote. Ranges, model and forecast bands and sample caveats should sit where the reader looks first, because a point estimate drawn alone invites a stronger reading than the data allow.

How to read the same gaps in a source table is covered in the statistics verification guide.

One check on the axis

Bar and area charts start at zero. The UK Office for National Statistics axes and gridlines guidance calls for zero baselines on bar and area charts, consistent scales across comparable charts, and clear treatment of nonzero or nonlinear axes.

A line chart can sometimes start above zero, if the axis break is clear and the range does not exaggerate the claim. The reviewer marks any exception and says why it stands. The reason belongs in the file and, where space allows, on the chart.

Headline, caption and alt text last

Headline, deck, chart, caption, alt text and social copy should agree. Test any highest, lowest, majority, average, typical, risk or record claim against the data, and use estimate language when point values are uncertain. Examples of these failures are collected in common chart reporting problems.

Read the finished page as a reader would, with the chart, the headline and the caption together. If the caption says more than the chart supports, change the caption. Do not widen the claim to fit the graphic.

Common questions