A practical data table and chart guide for online newspapers

Tables and charts can make a number-heavy article easier to understand. A visual, however, is not automatically accurate. Connecting figures built on different definitions, narrowing an axis too aggressively, or omitting the source date can create a stronger misconception than prose alone.

Small newsrooms often work without a dedicated data desk. A repeatable sequence matters more than elaborate software: secure the source, test comparability, choose the form, check mobile presentation, and record revisions. This guide covers that practical path from public data to a published table or chart.

Write the question before opening a chart tool

State what readers should compare or discover. Useful questions include how a measure changed over 12 months, how regions differ, or what share each category represents. A precise question determines the period, fields, and visual form.

If there are several questions, use separate simple visuals rather than forcing everything into one chart. A single important number may be clearer in a sentence or small table. Do not begin with a decorative graph and invent its meaning afterward.

In the planning note, record the question, intended audience, period, and unit. This one line narrows the data and reduces unnecessary processing.

Freeze the source and its publication state

Whenever possible, inspect the original dataset and documentation rather than copying a summary from a press release. Record the file name, public URL, retrieval time, and reference date. Web figures can be revised, so the newsroom must be able to identify the version used.

Measures with identical names may have different definitions. Check the population, geographic scope, collection method, provisional status, and exclusions. Do not mix monthly and cumulative values, counts and rates, or cases received and cases completed.

Keep source columns separate from processed columns. Document unit conversions, totals, and treatment of missing values in enough detail for another editor to reproduce the result.

Test whether the figures are comparable

Comparison conditions matter more than a polished shape. Check whether periods are equal, denominators match, and classifications or boundaries changed. If a definition changes, separate the series or mark the break instead of drawing one uninterrupted line.

A blank, zero, and not collected are different states. Converting missing information to zero can manufacture a decline. Use an explicit symbol in a table and explain what it means.

For calculated rates, verify that numerator and denominator cover the same population and period. Show absolute counts beside dramatic percentages from a small base. Note when rounded categories do not add to the displayed total.

Match the visual form to the question

Lines generally suit change over time, bars support comparisons among categories, and tables help readers retrieve exact values. Pie or stacked forms work only when there are few categories and the whole is clearly defined.

When categories multiply, emphasize the relevant series and simplify the rest instead of adding more colors. Use a map only when geography is part of the question. Raw totals can make populous areas dominate, so identify whether the visual shows totals or a population-adjusted rate.

Three-dimensional effects, shadows, and pictorial symbols can distort length and area. Remove effects that do not help readers evaluate the values.

Keep axes and units from exaggerating the story

Bar charts communicate through length and are usually easiest to interpret from zero. If a truncated axis is genuinely necessary, mark it prominently and explain it. A very narrow line-chart range can also make a modest movement look extreme.

Place axis titles, units, and periods near the data. Do not make readers guess whether values represent currency, people, cases, or percent. Distinguish percentage change from percentage-point change in labels and prose.

Dual vertical axes can make unrelated movements appear connected. Split the charts unless the combination is essential. If two axes remain, make every series-to-axis relationship unmistakable.

Do not communicate with color alone

Readers who cannot distinguish particular colors may miss categories, direction, or risk levels. Combine color with direct labels, line styles, markers, or patterns. Place names near lines and bars where possible instead of requiring repeated trips to a legend.

Check contrast and avoid tiny gray text. Colors can also carry political or cultural associations not present in the data, so do not let color imply an evaluation. Reserve warning colors for information that actually requires attention.

Alternative text should summarize the main pattern and important values, not merely say “bar chart.” If every number matters, provide an accompanying data table rather than placing a long inventory in alt text.

Review tables and labels on mobile

A chart that works on a desktop can produce overlapping labels and a distant legend on a phone. Inspect the actual article width, including title, axes, notes, and source. Reduce categories or switch to horizontal bars before shrinking text beyond comfortable reading.

For wide tables, place essential columns first and shorten headings without making them ambiguous. If horizontal scrolling is unavoidable, make that behavior visible and preserve the meaning of the first column. A concise summary table can link to complete data.

For image-based charts, confirm that text remains clear when enlarged. State the central finding in ordinary article text so the story still makes sense if the visual does not load.

Separate observation from interpretation

A chart title should first identify what was measured. Unless causation has been established, describe an observed increase during a period rather than claiming one event caused another.

Notes should identify the source, period, unit, provisional status, essential definitions, and important processing. The full methodology need not fit inside the image, but readers should be able to locate the source and understand the main limitation.

Annotations on unusual points must also distinguish confirmed events from hypotheses. Two events occurring together do not establish cause. Reporting and prose should handle interpretation that the data cannot prove alone.

Trace selected values back before publication

Choose several values from the finished visual and trace them to the source. Prioritize the maximum and minimum, newest value, numbers quoted in the article, and values used in rate calculations. Recalculate totals and subtotals.

When possible, ask an editor who did not process the data to review the headline, axes, units, source, and article figures together. The review question is not only whether the chart looks good, but how a reasonable reader could misunderstand it.

For scheduled publication, check whether the source may update before release. Decide whether to refresh the figures or preserve the stated reference point, and record that decision. Never replace data automatically without another review.

Record revisions and safe reuse

An agency may finalize provisional figures or correct an error after publication. Assess whether the change affects the article's conclusion, then update chart and prose together. Changing a value without revisiting the title and explanation can create contradictions.

Before reusing a chart in another article, verify the source and reference date again. Do not copy an old image in a way that makes it look current. A design template may be reusable, but its data and explanation require a fresh review.

Log the changed value, reason, time, and source. If a revision is material, disclose it in the article's update note so readers can understand what changed.

Data visualization checklist for a small newsroom

• Is there one sentence stating the question the visual answers?
• Are the source URL, reference date, retrieval time, and version recorded?
• Are definitions, periods, denominators, and units comparable?
• Are zero, blank, and not collected kept distinct?
• Does the selected table or chart match the question?
• Do axis range and color avoid exaggerating differences?
• Are labels and alternative text available beyond color alone?
• Can readers understand titles, values, notes, and sources on mobile?
• Have central values and article figures been traced to the source?
• Is someone responsible for monitoring and recording updates?

Conclusion

A strong data visual is not a decorative picture. It is an editorial result in which the question, source, comparison conditions, and presentation agree. Readers should understand the main finding in prose and use the visual to inspect both evidence and limitations.

Before building the next chart, write down the question it answers and the reference date of its numbers. Those two lines are the simplest protection against visual exaggeration and confused updates.