Charts carry an air of objectivity that plain sentences don't. A line going up feels like a fact, not an argument — which is exactly why a misleading chart is so persuasive. The distortions rarely involve fake numbers. More often the data is real and the framing does the lying: a cropped axis, a hand-picked start date, a percentage stripped of its base. The good news is that the same handful of tricks show up again and again, and once you know where to look they're hard to un-see.
This guide is about method, not verdicts. It won't tell you whether any particular claim is true — it'll give you the checks to work that out yourself. For claims that arrive as an article rather than a chart, pair this with our guide on verifying a news article.
Reading a chart critically — the workflow
Read the axes before you read the headline
A chart's story lives in its axes, not its title. Before you absorb the takeaway, find what each axis measures, what units it uses, and what range it covers. Most people do the opposite — they read the bold caption, glance at the shape, and move on. The distortions are almost always hiding in the small print along the edges, which is exactly where the eye doesn't go.
Check where the y-axis starts
For a bar chart, the vertical axis should start at zero. When it starts higher — say a jobs chart running from 95 to 100 instead of 0 to 100 — a trivial change balloons into a cliff. This isn't a fringe trick: a Cognitive Research study (Yang et al.) found that truncated bar graphs persistently mislead viewers even after they're told the axis is cut. Line charts have more leeway, but a bar chart that doesn't touch zero deserves suspicion.
Ask what timeframe was chosen — and why
Cherry-picking the window is the quietest distortion of all. Start a climate chart in an unusually hot year and warming looks flat; start an economic chart the month after a crash and any recovery looks miraculous. When a trend is dramatic, ask what the data looked like just before and just after the frame. If widening the window would kill the story, the story was the framing.
Separate raw numbers from percentages
"Cases up 200%" sounds alarming until you learn it went from 1 to 3. Percentages hide their base; raw counts hide their scale. A trustworthy stat gives you both. Be especially wary of a percentage change applied to another percentage ("support rose 5 points, a 50% jump!") — technically true, deliberately confusing. Whenever you see one form, mentally ask for the other.
Trace the number back to its source
A statistic shared as an image has usually been through several hands, and each pass is a chance for a caption to drift from what the data actually said. Click through to the original study, dataset, or agency release. Check the sample size, the date, and whether the source measured what the post claims. If there's no traceable primary source — just a screenshot of a chart with no citation — treat it as an assertion, not evidence.
The distortions to watch for
Most misleading charts use one of a small set of moves. Learn to name them and you'll catch them at a glance.
The truncated axis
A y-axis that starts above zero, magnifying a small difference into a dramatic gap. The most common chart distortion, and one that keeps working even on people who know the trick.
Dual axes
Two lines on two different vertical scales, arranged so they cross or track each other. By sliding the scales, almost any two datasets can be made to look correlated.
The cherry-picked window
A start and end date chosen to flatter one conclusion. Zoom out and the trend often reverses or flattens entirely.
Area and 3D distortion
Doubling a circle's width quadruples its area, so bubble and pictograph charts routinely overstate differences. 3D tilt exaggerates whichever bar sits in front.
Correlation styled as cause
Two lines moving together is not proof one drives the other. Spurious pairs (ice-cream sales and drownings) track perfectly and mean nothing.
Where to check the numbers
The primary source
The single most useful check: find the original dataset or study behind the chart. Government statistics agencies, peer-reviewed papers, and institutional releases beat any redrawn version.
Our World in Data
Interactive, well-sourced charts on health, economics, and environment where you can change the time range yourself — a fast way to see whether a viral chart cherry-picked its window.
Reverse image search
Drop the chart image into Google Lens or TinEye. Recycled and doctored charts often trace back to an older, differently-labeled original.
Full Fact & Sense about Science
Non-partisan organizations that regularly break down how a specific statistic was misused, and publish plain-language guides to reading numbers in the news.
"How Charts Lie" (Alberto Cairo)
A short, readable book by a data-visualization professor that catalogs these distortions with real examples. The reference many fact-checkers point newcomers to.
Misleading isn't the same as false
A truncated axis or a tight time window isn't automatically dishonest — a stock analyst tracking a 2% move genuinely needs a zoomed-in scale. Context decides. Interestingly, a PLOS One study found that people shown misleading graphs in a realistic setting were less fooled than lab experiments predicted, partly because surrounding text and prior knowledge corrected the visual. The lesson isn't to distrust every chart; it's to read the whole thing — axis, source, and caption together — rather than reacting to the shape alone.
Before you share a striking stat
The charts most likely to travel are the ones engineered to provoke — a vertiginous spike, a shocking gap. That reaction is the cue to slow down, not speed up. Ask the five questions above, spend the thirty seconds it takes to find the source, and if the number doesn't hold up, don't pass it on. A misleading chart only works if people forward it before they read it.
Check a claim in seconds
FAXTR searches 100+ fact-checking organizations in one query — free, no login required.