How to spot misleading graphs in the news: a 60-second check
Spot a misleading graph by checking the axis, scale, time window and source. A 60-second routine, six common tricks and a worked example.

Short answer: to spot a misleading graph, read the axes, the time window and the source before you read the shape. Check where the vertical axis starts, whether the scale is even, which dates were left out, and whether the numbers are raw counts or rates. Most bad charts fail one of those four checks in under a minute, and none of them needs a statistics degree.
Charts feel objective because they look like measurement. That is exactly why they work so well for spin. Newsrooms that publish interactive charts, and the data visualization consulting services they hire to build dashboards and charting tools, make the same choices you are about to check from the other side: where an axis begins, which years go in, whether dollars are adjusted for inflation. Every one of those choices can be honest or slanted, and a reader can usually tell which.
This guide gives you a 60-second routine, the six tricks that cover most misleading charts in the news, a worked example with real arithmetic, and what newsrooms can do to earn trust. If you came here from a viral post, our guides on how to tell satire from fake news and how to spot a deepfake cover the other two things that travel without context.
What makes a graph misleading?
A graph is misleading when the picture tells a different story from the numbers behind it. The data can be accurate and the chart can still deceive, because the design decides what your eye compares.
The statistician Edward Tufte gave this a number in The Visual Display of Quantitative Information (1983). His “lie factor” divides the size of an effect shown in the graphic by the size of the effect in the data. A lie factor of 1 is honest. Tufte treated anything far from 1 as distortion.
Darrell Huff’s How to Lie with Statistics (1954) described most of the same tricks seventy years ago, which tells you how durable they are.
Not every misleading chart is a lie. Many are rushed: a default setting in a spreadsheet, a designer who cropped the axis to fit a phone screen, an intern who pasted two series on one chart. The checks below work the same whether the cause is intent or haste, and that matters, because you rarely know which one you are looking at.
How do you check a graph in 60 seconds?
Run these six checks in order. Stop when one of them changes what you think the chart says.
- Name the measure. Read the title and the axis labels out loud. Is it a count, a percentage, a rate per 100,000 people, a share of a total, an index? A chart titled “crime” that plots reported incidents is not the same as one that plots victimization surveys.
- Find zero. Look at the bottom of the vertical axis. If bars start at anything other than zero, the differences between them are exaggerated.
- Check the spacing. Are the gaps between tick marks equal? Do the dates on the horizontal axis run at even intervals, or does the chart jump from 2010 to 2020 to 2022?
- Check the window. When does the series start and stop? Ask what the line looked like five years earlier or one year later.
- Find the denominator. Raw totals usually grow with population and prices. Look for “per capita”, “as a share of”, or “adjusted for inflation”.
- Follow the source. Who produced the data, and can you find the original table? A chart with no source line, or a source like “studies show”, deserves the least trust.
What each check catches
| Check | Question to ask | Red flag |
|---|---|---|
| Measure | What exactly is counted? | Vague labels such as “impact” or “activity” |
| Zero | Where does the vertical axis start? | Bars starting at a high number |
| Spacing | Are intervals equal on both axes? | Skipped years, uneven ticks, unlabeled log scale |
| Window | Why these start and end dates? | A series that begins at a peak or a trough |
| Denominator | Per what, adjusted how? | Raw dollars across decades, totals across states |
| Source | Who made the data and the chart? | No source line, a cropped screenshot |
Why does the starting point of the axis matter?
The truncated axis is the most common trick in news graphics because it is the easiest to do by accident. Spreadsheet tools often start the vertical axis just below the smallest value, which turns small differences into dramatic ones.
A worked example with real arithmetic
Imagine a bar chart comparing a tax rate of 35% with a proposed rate of 39.6%, with the vertical axis starting at 34%.
- The first bar is 1 point tall on screen (35 minus 34). The second is 5.6 points tall (39.6 minus 34).
- So the second bar looks 5.6 times as tall, or 460% bigger.
- The real change is 4.6 points on a base of 35, which is about 13%.
- Tufte’s lie factor here is roughly 460 divided by 13, or about 35.
Redraw the same two bars from zero and the difference is visible but modest, which is the honest picture.
When a line chart does not need to start at zero
Bars encode value by length, so cutting the bottom off a bar breaks the comparison. Lines encode change by slope, so a line chart can zoom in when the movement matters. A chart of body temperature that starts at zero would hide a fever, and a chart of a stock index that starts at zero would flatten every crash. The fair test for a line chart is whether the axis is clearly labeled and whether the zoom suits the size of changes that matter in real life.
Watch for an upside-down axis
Some charts flip the vertical axis so that higher values sit lower on the page. A widely shared 2014 news graphic about gun deaths in Florida did this, and many readers took a rise for a fall. The designer said the style was meant to echo dripping blood. Whatever the intent, the convention is that up means more, and breaking it misleads.
How can scale and shape distort the data?
Even with an honest zero, the geometry of a chart can mislead. Look for four patterns:
- Uneven intervals. A horizontal axis that runs 1990, 2000, 2010, 2015, 2020 at equal spacing compresses the early decades and makes recent change look steeper.
- Unlabeled logarithmic scales. On a log scale each step up means multiplying, not adding. Log scales are useful for growth that compounds, as many pandemic charts showed, but readers who don’t notice the scale badly misjudge how fast things change.
- Pictures and 3D. When an icon doubles in height and width to show a doubled value, its area quadruples, and your eye reads area. 3D pie charts tilt the front slice toward you and make it look larger than an equal slice at the back.
- Stretched or squashed frames. Making a chart very wide flattens a line; making it tall and narrow steepens it. Social media crops often do this on their own.
How does cherry-picking the time window change the story?
Choosing the start and end dates is the quietest way to steer a chart, because every individual data point is real. Start a series at an unusual low and any recovery looks like a boom. Start it at a peak and an ordinary year looks like a collapse.
Three questions catch most cherry-picking:
- Why does the chart begin in this particular year? Is there a natural reason, such as when the data series began?
- What happened just before the first point? Look up a longer version of the same series.
- Is the latest point final? Recent months are often preliminary and get revised.
The fix is usually a longer chart. Official sources such as the Bureau of Labor Statistics or the Census Bureau publish long series that let you see whether a “record” is really one.
Why do raw numbers mislead without a denominator?
Totals grow when populations and prices grow, so raw counts across states or decades often mislead even when they are correct.
- Population. A large state will usually lead in total cases, crimes or traffic deaths. The rate per 100,000 residents is the fair comparison.
- Inflation. Dollar amounts from different years are not comparable until adjusted. The BLS CPI inflation calculator converts them in seconds, and “record” box-office or price headlines often vanish after adjustment.
- Base rates. A rise from 2 to 4 cases is a 100% increase and still only two cases. Look for the absolute numbers behind any percentage.
- Percentage points versus percent. Unemployment moving from 4% to 5% is a rise of one percentage point, or 25%. Both are true; the headline will pick the louder one.
What are the tricks with two lines on one chart?
Dual-axis charts put two series on one picture with separate scales on the left and right. Because the designer picks both scales freely, the two lines can be made to cross, track each other or diverge at will. If you see two scales, mentally cover one side and ask what each line shows on its own.
Two lines moving together also invite a false conclusion: that one causes the other. The site Spurious Correlations collects real data series that track each other closely for no reason at all. A chart can show that two things changed at the same time. It cannot, by itself, show why.
Why do misleading charts spread so fast online?
Most people meet charts as screenshots on a phone, not on the page where they were published. That format strips the context that would help you judge them:
- The source line and the footnotes are cropped off to fit the post.
- The headline above the chart is replaced by a caption written by whoever shared it.
- Small axis labels become unreadable at phone size, so only the shape survives.
- Charts made with AI image tools can look polished while plotting numbers that exist nowhere.
Treat a chart without a source like a quote without a speaker: a reverse image search or a search for the claim plus the word “chart” usually leads back to the original. Fact-checking tools help here too; our guide to AI fact-checking tools for journalists covers which ones can trace a claim to its source.
What should you do before you share a chart?
- Find the original. Locate the publication or agency that produced the data, not just the account that posted it.
- Read the fine print. Check the axis labels, the units and any notes about methods or revisions.
- Look for a longer or adjusted version. A per-capita or inflation-adjusted chart of the same data is often one search away.
- Check whether it has been debunked. Fact-checkers regularly cover viral charts; a quick search for the chart’s claim plus “fact check” often settles it.
- Share with context. If you still want to post it, link the source and say what the chart does and does not show.
How can newsrooms build charts readers trust?
Publishers and newsletter writers can make honest charts the default rather than an extra step. A short house checklist covers most of it:
- Start bar charts at zero; label any line chart that zooms in.
- Keep the source line and the date inside the image, so it survives screenshots.
- Use rates and inflation-adjusted figures when comparing places or decades.
- Avoid dual axes, 3D effects and pictograms unless there is a clear reason.
- Publish the underlying table or a link to it next to every chart.
- Write down these rules, as we do in our editorial standards, so they do not depend on who is on deadline.
Teams that build charting into their own site or app can also bake these rules into templates: a zero baseline that a reporter has to override on purpose, a mandatory source field, and an alt-text description that states the main number in words for screen readers.
Frequently asked questions
How can you tell if a graph is misleading?
Check four things: where the vertical axis starts, whether the scale is even, which time window was chosen and whether the numbers are raw totals or rates. Then find the original source. Most misleading graphs fail one of these checks within a minute.
Should a graph always start at zero?
Bar charts should, because bars show value by length and a cut-off bottom exaggerates differences. Line charts do not have to, because they show change by slope, but the zoom should be clearly labeled and suit the size of changes that matter.
What is the most common type of misleading graph?
The truncated axis, where a bar chart starts well above zero so small differences look huge. It is common partly because spreadsheet tools often set the axis just below the smallest value by default.
Can a graph be misleading if the data is accurate?
Yes. Accurate numbers can mislead through a cropped axis, a hand-picked time window, raw totals instead of rates, unadjusted dollars or two series on separate scales. The design decides what your eye compares.
What is the lie factor in a chart?
It is Edward Tufte's measure of distortion: the size of the effect shown in the graphic divided by the size of the effect in the data. A value of 1 is honest; values far above or below 1 signal a distorted chart.



