How Charts Mislead

A chart carries authority. Because it looks like a neutral picture of numbers, readers tend to trust it more than the same figures in a sentence. That trust is exactly why a chart can mislead so effectively: a few small choices about scale, range, or dimension can make a trivial difference look enormous, or hide a real one. Most misleading charts are not outright lies — the numbers are usually correct. The distortion is in how those numbers are drawn. Here are the classic techniques, framed so you can spot them in the wild and avoid them in your own work.

The truncated axis — the classic

This is the most common distortion of all. On a bar chart, value is encoded by the length of each bar, which only works if the axis starts at zero. Move the start of the y-axis up to just below the data, and the bars get chopped off near the bottom — so a 2% difference can be drawn to look like a threefold one. The numbers on the axis are honest; the lengths are not.

How to avoid it: Start every bar chart at a zero baseline, and when you read someone else's chart, glance at where the y-axis begins before you trust the shape. A line chart encodes value by position rather than length, so it can legitimately use a non-zero scale — but only when the axis is clearly labelled and the choice is honest, not chosen to exaggerate.

Inconsistent and dual axes

Two distortions live here. The first is an inconsistent scale — uneven gaps between tick marks, or an axis that switches units partway — which warps the spacing readers rely on. The second is the dual axis: plotting two different series against two separate y-axes on the same chart. Because the designer chooses each scale independently, they can slide the axes until the two lines appear to track each other or cross at a dramatic moment. The apparent relationship is manufactured by the scaling, not found in the data.

How to avoid it: Keep one consistent, evenly spaced scale per axis. Instead of a dual axis, show two small charts stacked with aligned x-axes, or index both series to a common starting value (say, 100) so they share one honest scale. If you see two y-axes, treat any "correlation" with suspicion — and remember that even a real correlation is not proof of cause.

Cherry-picked time ranges

A chart can be perfectly drawn and still mislead by what it leaves out. Show only the months where a number rose and you imply relentless growth; start the window right after an unusual spike and you imply a crash. The trick is selecting a range that supports a story the full data would contradict.

How to avoid it: Show the full relevant period, or enough of it that the recent movement sits in context. If you must crop, say why, and note what falls outside the window. Let the reader see the whole trend before you point to a part of it.

3-D and area distortion

Adding a third dimension to a flat chart almost always distorts it. A 3-D pie tilts the circle, so slices near the front look larger than their true share; a 3-D bar chart adds perspective that makes values genuinely hard to read off the axis. A related trap is area distortion: when an icon or bubble is scaled by both width and height to represent a single number, doubling the value quadruples the area, so the eye massively overestimates the difference. The same applies to scaling a picture to "twice as big."

How to avoid it: Keep charts flat — two dimensions, no perspective, no shadows. When sizing circles or icons by a value, scale the area to the value, not the radius or the height. Decoration that distorts is worse than no decoration at all; see chart mistakes to avoid for more on stripping out chartjunk.

Pie-chart misuse

Pie charts mislead in two main ways. First, they are used for things that are not parts of a single whole — comparing unrelated quantities, where the slices add up to nothing meaningful. Second, even when composition is the right goal, people read angles far less accurately than lengths, so two slices of 24% and 27% look identical. Add too many slices, an exploded wedge, or a 3-D tilt, and the chart becomes guesswork.

How to avoid it: Use a pie chart only for a few genuine parts of one whole, and only when "which part dominates" is the message. When you need precise comparison, use a bar chart — the pie vs bar comparison spells out exactly when each wins.

A 30-second spot-check

Before you trust any chart — including your own — run through four questions:

• Where does the axis start? Bars should begin at zero; a truncated axis is the number-one red flag.
• Is the whole range shown? Watch for cropped or oddly specific time windows.
• Is there a second axis? Dual scales can manufacture correlations that are not there.
• Does any 3-D or area effect distort sizes? If so, the picture is exaggerating the numbers.

If the headline claim weakens once the chart is redrawn honestly, the chart was doing the persuading — not the data.

Make honest charts by default

The simplest defense is to build with tools that start at zero and stay flat. Every chart maker on Chart.biz does, and our guide to choosing the right chart helps you pick a type that tells the truth in the first place.