Chapter 06 · Field guide

How Can a Chart on X Mislead You?

Short answer

A chart misleads when its design changes the impression the data gives, most often through a bar chart axis that does not start at zero, a hand-picked time window, two mismatched axes, or totals that ignore population size.

Key takeaways

  • Bar charts should start at zero. If the axis starts at 95, a 2% difference can look like a tenfold one.
  • Ask what happened just before and after the time window shown. Short windows can make any trend appear.
  • Two lines on two different axes can be made to cross, diverge or track each other by rescaling either one.
  • Totals favor big places. Compare per person or per 1,000 before ranking cities, states or countries.

A screenshot shows two bars. One towers over the other. The caption: "Brand A batteries last WAY longer." Look at the left edge of the chart and you see the axis runs from 95 to 100 hours. Brand A lasts 99 hours, Brand B lasts 97. The bars are honest about the numbers and dishonest about the size of the difference. Charts are fast to read, which is exactly why they are easy to bend.

Why does a truncated axis change what you see?

A bar chart whose axis does not start at zero exaggerates differences, because your eye compares the lengths of the bars rather than the numbers at their tips.

In the battery example, Brand B's bar is 2 units tall on a 95-to-100 scale and Brand A's is 4. Brand A looks twice as good. In reality it lasts about 2% longer.

Line charts are a different case. A line showing a city's average July temperature over 50 years might sensibly start at 70°F, since nobody is comparing heights to zero. The test is whether the scale is labeled and whether the chart's message survives once you read the numbers. If you want to check how big a gap really is, enter both values in the percent versus points tab of the percentage check.

Starting or ending a chart at a carefully picked date can make a flat or falling series look like it is rising, or the reverse.

Say a smartwatch brand posts a chart of its monthly sales from January to March, showing a steep climb. If the previous December had a holiday spike twice as high, a chart starting in December would show a decline. Neither chart contains a false number.

Questions to ask:

  • What happened before the first date and after the last?
  • Is the period a full cycle, such as a whole year for anything seasonal?
  • Why does it start here? A start date at an unusual low point is a warning sign.

This is cherry-picking applied to time. The fallacy finder has a practice question on it.

What goes wrong with two axes on one chart?

When two lines use separate vertical axes, the chart maker can rescale either axis to make the lines appear related or unrelated, so the visual link proves nothing.

Picture ice cream sales on the left axis and sunglasses sales on the right. With the right scaling, the two lines lie on top of each other and look like twins. Shift one axis and they drift apart. Both series probably rise in summer for the same reason: hot weather. The chart hints at a connection that the data cannot establish on its own.

If a post uses a dual-axis chart to suggest that one thing causes another, look at each series alone and ask what third factor might drive both. That is the confounder problem from reading a cited study.

Why should you check for per-capita numbers?

Raw totals mostly measure how big a place or group is, so rankings should compare rates, such as per person or per 1,000 people.

A map showing "most bike thefts by state" will usually highlight the states with the most people and the most bikes. That tells you where people live. Divide by population, or by the number of bikes, and the ranking can change completely.

What the chart shows Question to ask Better comparison
Total incidents by city How many people live there? Incidents per 100,000 residents
Cumulative signups since launch What were new signups each month? Monthly new signups
Spending in dollars over 30 years Was it adjusted for inflation? Inflation-adjusted dollars
Icons or circles of different sizes Did area or height carry the value? A plain bar chart

The last row matters more than it looks. If a chart doubles the height and width of an icon to show a doubled value, the icon's area grows four times, and your eye reads the area.

What other design choices should make you pause?

Missing labels, missing units, cumulative totals and three-dimensional effects all make a chart harder to check, and each one deserves a second look.

A chart with no axis labels or no units cannot be verified at all. A cumulative chart, which adds each period to the running total, can only go up, so it can hide a sharp fall in new activity. Three-dimensional bars and tilted pie charts distort proportions because the front slices look bigger. Colors matter too: a red-to-green scale can imply "bad" and "good" when the data only shows "more" and "less."

How do you read any chart in a screenshot?

Work through the source, the axes, the window and the units before you accept the chart's headline.

  1. Find the source. Search the title or source line to reach the original.
  2. Read both axes. Where does each start, and what unit is it in?
  3. Check the time range. What is just outside it?
  4. Look for rates. Are totals being compared across groups of different size?
  5. Watch for two axes, and do not read a link between the lines as a cause.
  6. Check whether the series is cumulative.
  7. Restate the finding in one plain sentence with the real numbers.

If that sentence is less exciting than the caption, trust the sentence. For the numbers behind the chart, the chapter on misleading statistics goes further.

Frequently asked questions

Does every chart axis have to start at zero?

Bar charts should, because readers compare bar lengths. Line charts can reasonably start elsewhere when the point is change over time, as long as the axis is labeled clearly so you can see the scale.

How can I check a chart that is only a screenshot?

Look for a source line, title or watermark, then search for it to find the original publication. The original usually shows the full time range, the units and any notes the screenshot cropped out.

What is a cumulative chart and why does it mislead?

A cumulative chart adds each period's total to all the previous ones, so the line can only go up or stay flat. It can hide the fact that new cases, sales or signups are actually falling.

Sources

  1. U.S. Census Bureau, Statistical Quality Standards
  2. U.S. FDA, Communicating Risks and Benefits: An Evidence-Based User's Guide
  3. Our World in Data, About (methods for comparing data across countries)

Examples in this chapter are invented for teaching. Facts about platforms were checked in November 2026 and can change. See how we research.