Chart Truthfulness & Ethics for UK Journalists

What you need to know

IPSO Clause 1 (Accuracy) obliges publishers to avoid inaccurate, misleading, or distorted information — and this obligation extends to charts and infographics, not only written text. A chart that visually exaggerates a trend through a truncated y-axis or an undisclosed log scale can mislead readers just as effectively as a false statistic. The Office for Statistics Regulation (OSR) monitors the use of official statistics in public communication and has published specific concerns about charting choices that distort the perception of data.

UK data journalists are increasingly expected to apply the same editorial rigour to visual outputs as to written copy. Fact-checking organisations including Full Fact routinely scrutinise published charts for misleading design choices, and corrections have been required following coverage that used dual-axis layouts, undisclosed denominators, and 3D effects that visually inflate smaller values.

Best-practice resources from Datawrapper Academy and the ONS methodology guidanceoffer concrete standards for scale choice, labelling, and contextualisation. Embedding these standards into your editorial workflow reduces the risk of publishing a chart that is technically accurate in its numbers but misleading in its visual impression — a distinction IPSO's Clause 1 does not excuse.

When chart truthfulness issues arise

1A y-axis begins above zero to compress the visual baseline and make a small numerical change appear dramatic.
2Two unrelated datasets are placed on a dual-axis chart, creating a spurious visual correlation through independent axis rescaling.
3A 3D pie or bar chart is used, which distorts the visual proportion of segments through perspective rendering.
4A logarithmic scale is used on a public-facing chart without labelling the axis as logarithmic or explaining the compression in accompanying text.
5Percentage changes are reported without stating the underlying base figure, making small absolute changes appear significant.
6A cumulative chart is presented without disclosure that values accumulate rather than show point-in-time data.
7Colour scales on choropleth maps use non-linear intervals, causing readers to overestimate regional differences.
8Interactive charts allow readers to select their own date range in ways that enable cherry-picking without editorial guidance on context.

Red flags

The y-axis of a bar chart does not start at zero and no editorial justification or clear label explains the truncation.
A chart supplied by an external PR, political party, or government department uses axis manipulation — third-party charts inherit the publisher's IPSO responsibilities.
A 3D effect has been applied to a pie chart: the front slices appear larger than back slices due to perspective, misrepresenting proportions.
A dual-axis chart has been inherited from a wire service without sub-editing the axis scales for independent verification.
The data source cited is a press release rather than the underlying ONS, HMRC, or government dataset — primary sources should be linked directly.
The chart headline states a trend (e.g. “Cases soaring”) that is only visible because the axis has been zoomed in on a small range.

Chart accuracy checklist

The y-axis starts at zero, or a non-zero baseline is explicitly labelled and editorially justified in the caption.
If a log scale is used, the axis is labelled as logarithmic and the body text explains what this means for interpretation.
No 3D effects have been applied to bars, pies, or area charts.
Dual-axis charts have been avoided; if unavoidable, both axes are independently labelled and a caveat explains the independent scaling.
All percentages and ratios are accompanied by the underlying absolute count or base population.
The primary data source (ONS, HMRC, official dataset URL) is cited beneath the chart, not just a press release or secondary report.
The chart headline accurately reflects the data range shown, not a cherry-picked portion of a longer trend.
A colleague or data editor has reviewed the chart independently before publication.

Data journalism tools & ethics hubs

Explore our Data Visualisation Tools guide for software that builds honest charts by default, or visit the AI-Generated Content hub for ethics guidance when AI is used to produce or annotate visual data.

Data Visualisation Tools AI-Generated Content Ethics

Common mistakes

Passing a chart through from a press release or government slide deck without independently verifying the axis scales and source data.
Using percentage change as the sole metric when the base number is too small to make the percentage meaningful or comparable.
Assuming chart software defaults are editorially neutral — tools like Excel default to non-zero y-axes; always override to start at zero for bar charts.
Treating a correlation visible in a dual-axis chart as evidence of causation without statistical verification.
Publishing a chart without a data source citation, making it impossible for readers or fact-checkers to verify the underlying figures.
Applying colour gradients that are not perceptually uniform (e.g. rainbow colour maps) which cause readers to perceive false boundaries in continuous data.

Primary sources

IPSO Editors' Code of Practice— IPSO
Office for Statistics Regulation— UK Statistics Authority
Full Fact — UK Fact-Checking Organisation— Full Fact
ONS Methodology Guidance— Office for National Statistics
Datawrapper Academy — Chart Best Practices— Datawrapper
Flourish — Data Visualisation Platform— Flourish

Frequently asked questions

Does IPSO Clause 1 apply to charts and infographics as well as text?

Yes. IPSO Clause 1 (Accuracy) requires that publications do not publish inaccurate, misleading, or distorted information. The Editors' Code does not limit this obligation to written text: a chart that visually misrepresents data — for example by truncating a y-axis to exaggerate a trend — can constitute a breach of Clause 1 in the same way as a factually incorrect statement. Full Fact has documented multiple IPSO complaints involving charts.

When must a journalist disclose that a chart uses a log scale?

Whenever a logarithmic scale is used in place of a linear scale, this must be clearly labelled on the axis and explained in the accompanying text or footnote. A log scale compresses large values and can make exponential growth appear to slow dramatically. Without disclosure, readers unfamiliar with log scales will misread the rate of change. The Office for Statistics Regulation has flagged undisclosed log scales as a transparency concern in pandemic-era coverage.

What is a truncated y-axis and why is it problematic?

A truncated y-axis begins at a value above zero, which compresses the baseline and makes differences between data points appear proportionally larger than they are. For example, a chart showing a 2% rise in a metric can be made to look like a 200% rise if the y-axis starts at 98 rather than 0. While truncation is sometimes legitimate — for instance when showing very small changes in a large-scale measure — it must always be clearly labelled and editorially justified. Starting axes at non-zero values without explanation is a common source of misleading charts in UK news.

What is the dual-axis trap in data journalism?

Dual-axis charts plot two separate datasets on a single chart using two different y-axes — one on the left and one on the right. The visual proximity of the two lines implies correlation, but the analyst can manipulate the apparent relationship simply by rescaling either axis independently. The dual-axis trap is therefore a form of chart deception even when no individual number is wrong. Best practice, endorsed by Datawrapper Academy, is to use separate panels (small multiples) or clearly state that axes are independently scaled whenever a dual-axis chart is unavoidable.

What does denominator transparency mean in chart ethics?

Denominator transparency means making clear what base population or total a percentage or ratio is drawn from. A headline such as '40% rise in cases' is misleading if the base figure is very small (going from 5 to 7 cases). ONS methodology guidance requires that rates and proportions are accompanied by the underlying counts so readers can assess statistical significance. In chart terms, this means always labelling sample size, population base, or total alongside any percentage-based chart.