Honest infographics: axes, scales and proportions

An infographic makes numbers easier to feel. That’s its power, and it’s also the risk. When a bar is twice as tall, readers feel “twice as much”, whether or not the numbers agree. When a chart exaggerates, readers walk away with the wrong impression, and if they notice, they stop trusting the rest of the page.

Most misleading charts aren’t made on purpose. They come from default settings, a wish to make a small change look dramatic, or a decorative idea that quietly distorts the data. Here are the traps to watch for and the plain fixes for each.

Bars start at zero

A bar’s length is its value. If the axis starts at anything other than zero, the bars no longer show the values; they show the values minus an arbitrary amount.

For example, say two branches served 940 and 980 customers in a week. On an axis from 0 to 1,000, the bars look nearly equal, which they are. On an axis from 900 to 1,000, the second bar looks twice as tall as the first. Same data, very different story.

The rule is simple: bar and column charts start at zero. If the difference you care about is too small to see from zero, that’s information too. Show the figures as numbers, or show the difference itself as the chart (“+40 customers”).

Line charts can zoom, carefully

Lines are different. A line shows change and shape, not size, so a line chart’s axis doesn’t have to start at zero. Zooming in on the range where the data lives can be perfectly honest.

Still, be careful:

• Label the axis clearly, including where it starts, so readers can see it’s zoomed.
• Don’t zoom so far that ordinary variation looks like a crisis.
• Keep the same range when you show the same measure in several charts side by side.

Size things by area, not by height alone

Icons, circles and pictures are where proportion goes wrong most often. If you double the height of an icon to show “twice as much”, you also double its width, so its area becomes four times larger. Readers judge size by area. The result overstates the change.

Honest alternatives:

• Repeat icons instead of enlarging them. Two equal icons for twice as much, ten for ten times.
• Scale circles by area, not by diameter, if you must use sized circles. Most charting tools have this option.
• Use bars when precise comparison matters. Bars change in only one dimension, so length maps cleanly to value.

Parts must add up to the whole

Pies, donuts, stacked bars and waffle charts all promise that the pieces make up 100%. Break that promise and the chart lies.

• Check that percentages add to 100. If rounding makes them 99 or 101, add a note.
• Don’t show a pie when respondents could pick more than one answer. Those totals exceed 100%; use separate bars instead.
• Don’t hide an “Other” slice to make the rest look bigger. Keep it, and keep it honest.

Keep scales consistent across a page

An infographic often has several small charts. If they measure the same thing, they should share the same scale. Two bar charts of monthly sales for two cities, one running to ₹10 lakh and the other to ₹50 lakh, invite the reader to compare bar heights that aren’t comparable.

Where scales must differ, make that obvious with clear axis labels, or add a line of text saying so.

Watch the visual extras

Decoration can distort as much as a bad axis.

Feature	Why it misleads	Better choice
3D effects	Perspective makes nearer parts look bigger	Flat bars and slices
Pictorial bars (bottles, buildings)	Shape changes width as well as height	Plain bars with an icon beside the label
Cherry-picked date range	Starting at a low point exaggerates growth	Show a fair, stated period
Missing base	“75% agree” from 8 people sounds like a lot	State the base: “6 of 8 people”
Mixed units	Per month beside per year	Convert everything to one unit

Say what the numbers can’t

Honesty isn’t only about drawing. It’s about the words beside the chart.

• Name the source and date on the page itself.
• Separate correlation from cause. “Districts with more libraries had higher reading scores” is a finding. “Libraries raised reading scores” is a claim the chart may not support.
• Keep the hedges. If the source says “estimated” or “provisional”, so should you.

A title that matches the chart exactly (“Footfall rose slightly in October”) is better than one that oversells it (“Footfall soars”). If the chart doesn’t back the headline, change the headline.

A truthfulness check

Before sharing any infographic, run through these:

1. Do all bars and columns start at zero?
2. Do icons and shapes scale by count or area, not by height alone?
3. Do parts of a whole add up, with a note if rounding shifts them?
4. Are charts of the same measure on the same scale?
5. Is every figure’s source, period and base stated?
6. Does the title claim no more than the chart shows?

Friendly numbers only stay friendly if they’re also exact. For help picking the form that suits each figure in the first place, see how to choose the right chart for an infographic.