Misleading Charts

Truncated Y-Axes

One of the most common tricks is starting the y-axis at a value other than zero. A bar chart showing approval ratings from 48% to 52% will make a 4-point difference look enormous, while the same data plotted from 0% to 100% shows a barely visible change.

This technique is not always dishonest — sometimes small differences matter and a truncated axis helps readers see them. But it is frequently used to exaggerate trends. Always check the axis labels and ask whether the visual impression matches the actual magnitude of the difference.

Cherry-Picked Time Frames and Dual Axes

Choosing a strategic start date can make any trend look good or bad. Stock market performance looks terrible if you start from a peak and looks excellent if you start from a trough. Climate change skeptics often cherry-pick the unusually hot 1998 El Niño year as a starting point to argue temperatures have not risen.

Dual y-axis charts — with different scales on left and right — can make any two variables appear correlated. By adjusting the scales, you can make almost any two trend lines appear to move together. This is why spurious correlation websites exist, showing things like 'per capita cheese consumption correlates with deaths by bedsheet tangling.'

Worked Example: The Truncated Y-Axis

Truncating the vertical axis is the single most common chart trick. Suppose support for a proposal rises from 48% to 50% between two polls — a change of two percentage points. Here is what the same data looks like under two different axis choices:

On the honest scale the second bar is barely taller — a 4% increase in height for a 2-point change. On the truncated scale, where the axis starts at 47% and ends at 51%, the second bar is three times the height of the first. Nothing about the data changed; only the axis did. A reader glancing at the truncated chart 'sees' a dramatic surge that does not exist.

The defence is simple: always read the axis before you read the bars. For bar charts the baseline should almost always be zero, because a bar's length is what encodes the value. (Line charts tracking change over time can sometimes justify a non-zero baseline, but they should say so.) Other tricks to watch for: inconsistent axis intervals, a second hidden y-axis chosen to make two lines cross, and 3-D effects that exaggerate the front-most slice. When a chart provokes a strong emotional reaction, that is precisely the moment to check the numbers behind it.

A Quick Audit Routine

Train yourself to run a five-second audit on every chart. Read the axis labels and units first. Check where each axis starts and whether the intervals are evenly spaced. Ask what time range was chosen and whether a different start or end date would tell the opposite story (cherry-picking). Look for a second y-axis, which lets a designer make any two lines cross wherever they like. Finally, ignore all colour, shading, and 3-D effects and re-read the raw numbers. If the story survives that audit, trust it; if the drama evaporates once the axis starts at zero, the chart was doing the arguing, not the data.