How to spot a misleading chart: truncated axes, cherry-picked ranges, dual axes

TUREFF• •0

Charts feel objective. Bars and lines look like plain facts. But the same numbers can be drawn to tell very different stories, sometimes deliberately and often by accident. Learning a few common tricks makes you a much sharper reader of the news, adverts and social media.

Truncated axes

The most common trick is a vertical axis that does not start at zero.

On a bar chart, this misleads. A bar's length is meant to represent the value. If the axis starts at 90, a bar of 95 looks twice as tall as a bar of 92.5, although the real difference is under 3 percent.
On a line chart, it is often fine. Lines show change over time, and zooming in can be the honest way to show small but meaningful movements, such as body temperature or interest rates.
What to check: look at the lowest number on the axis. If it is a bar chart and the axis does not start at zero, mentally rescale it. Ask how large the difference really is in absolute and percentage terms.

Cherry-picked time ranges

Any trend can look rising or falling depending on where you start and stop.

A share price chart starting at the bottom of a crash looks like a spectacular recovery. The same data from the previous peak looks like a loss.
A crime or temperature series can be made to look flat or alarming by choosing an unusual starting year.
What to check: ask why the chart starts and ends where it does. Look for a longer view of the same data. Be wary of very short windows on volatile data.

Dual axes

A chart with two vertical axes, one on the left and one on the right, puts two different series on one picture.

The author chooses both scales, so they can make any two lines appear to move together, cross at a dramatic moment, or diverge.
Two unrelated trends can look as if one is driving the other.
What to check: read both axes carefully. Ask whether the lines would look related if each had its own chart. Better alternatives are two charts side by side or indexing both series to 100 at the same start date.

Other tricks to watch for

Area and 3D effects. Doubling the height of a picture icon also doubles its width, so it looks four times bigger. 3D pie charts make slices at the front look larger than those at the back.
Pie charts that do not add up to 100 percent, usually because respondents could choose more than one answer.
Inconsistent intervals. An axis that jumps from 2010 to 2020 and then goes year by year distorts the shape of the trend.
Cumulative totals presented as if they were showing growth. A running total can only go up, even when the underlying numbers are falling.
Missing context. Raw counts where rates would be fairer, such as total cases in a big country compared with a small one.
Logarithmic axes without a label. Log scales are useful for growth, but a reader who does not notice will misread the size of changes.
Smoothing and averages that hide important spikes.
Missing sources. If you cannot see where the numbers came from, be cautious.

A quick checklist

When a chart makes you feel something strongly, pause and check:

What exactly is being measured, and in what units?
Where does the vertical axis start?
Why does the time range start and end where it does?
Are there two axes, and do they share a scale?
Are these counts or rates, and is that fair for the comparison?
Who made it, and where did the data come from?

Being fair to chart makers

Not every odd choice is a deception. Space is limited, conventions differ between fields, and sometimes a zoomed-in axis is the right call. The aim is not to distrust every chart, but to read the axes and labels before trusting the shape. Have you seen a chart that fooled you at first? Share it in the comments.