Print or save as PDF — or build a custom worksheet from any module's questions.
Have you ever seen a graph in a news article or advertisement that made a change look much bigger or smaller than it really was? What tricks might someone use to mislead you with a graph?
Graphs are powerful communication tools — but they can also be weaponised to deceive. Learning to spot misleading graphs is a critical life skill for evaluating news, advertising, and political claims.
The most common trick is a truncated y-axis. If the axis doesn't start at 0, even a tiny change looks dramatic. Before reading any bar or column chart, always check: where does the y-axis start? If it doesn't start at 0, ask why — and mentally rescale the graph.
A truncated y-axis cuts off the bottom of the scale so bars appear much taller in proportion to each other than they really are.
Example: A company's sales were 98, 99, 100, 101 (units) over four quarters. On an axis from 0 to 110, the bars are almost the same height. On an axis from 97 to 102, the bars look like they are growing dramatically.
The actual growth is $\dfrac{101-98}{98} \times 100 \approx 3\%$. The truncated graph can make it look like the value tripled or quadrupled.
Rule: Bar and column charts must start at 0. A break symbol (zigzag on the axis) signals a truncation — treat the graph with suspicion.
The numbers on an axis should be equally spaced. When intervals are unequal, the visual distance between values is misleading.
Example of a dishonest axis: 1, 2, 5, 10, 20, 50, 100. The jump from 1 to 2 looks the same as the jump from 50 to 100, even though the second jump is 50 times larger. A trend that is actually slowing down can look like constant growth.
Note: logarithmic scales are sometimes legitimate in science (e.g. for data spanning many orders of magnitude), but in everyday graphs they are almost always misleading. Always check whether tick labels are evenly spaced.
3D pie charts: Tilting a pie chart in 3D makes the slices at the front appear larger than slices at the back, even if the percentages are equal. 3D effects should never be used for accurate data communication.
Pictograph area trick: If sales double, an honest pictograph doubles the number of icons. A dishonest one doubles the width AND height of a single icon — but this makes the area 4 times larger ($2 \times 2 = 4$), not 2 times. Our eyes read area, so doubling the icon looks like a quadrupling of the value.
$$\text{Area scales with both dimensions: doubling width AND height} = 4 \times \text{original area}$$
Cherry-picking means choosing only the time period or data points that support your argument.
Example: A politician shows a graph of unemployment from 2010 to 2014 only — the years when it happened to be falling. They ignore 2008–2010 (when it rose sharply) and 2014–2018 (when it rose again). The selected window makes it look like unemployment is consistently improving.
Similarly, a company might compare sales only to their worst year to make current performance look impressive. Always ask: What data is NOT shown?
Misleading Graph Checklist:
A bar chart shows a company's profit over 4 years. The y-axis starts at $950,000 instead of $0. What misleading technique is being used?
When a y-axis is truncated (does not start at zero), differences between bars appear:
In a pictograph, a company doubles the HEIGHT and WIDTH of a coin icon to show sales doubled. What does the icon's area actually represent?
Which statement describes an honest, accurate graph?
Two graphs show the same data: sales of $100 vs $104. Graph A: y-axis 0–120. Graph B: y-axis 99–105. Which graph represents the data most accurately?
Q6. A bar chart shows "Number of students who passed" for four classes. The y-axis starts at 80 (not 0). The bars show heights suggesting Class A ≈ 3× taller than Class D.
(a) Explain why starting the y-axis at 80 is misleading.
(b) If the actual values are Class A = 95 and Class D = 85, what is the true percentage difference between Class A and Class D?
(c) Sketch (or describe) how the bar chart would look with a fair y-axis starting at 0.
Q7. A company reports: "Our sales DOUBLED last year!" Their graph shows sales rising from $98,000 to $102,000 on a y-axis starting at $96,000.
(a) Calculate the actual percentage increase in sales.
(b) Explain how the truncated graph could make this look like sales doubled.
(c) What would the graph need to show for the claim "sales doubled" to be visually accurate?
Q8. A graph of annual temperature rise shows only the years 2010–2020. Another climate scientist points out that 2000–2010 showed a cooling trend.
Identify TWO flaws in the original graph and explain how each distorts the overall picture of the data.
(a) Starting at 80 compresses the axis so that the visual difference between Class A (95) and Class D (85) appears to be far larger than 10 students. The eye compares bar heights, not axis values.
(b) Actual % difference = (95 − 85) ÷ 85 × 100 ≈ 11.8%. This is a modest difference, not a tripling.
(c) A fair graph would have the y-axis from 0 to about 110. Class A (95) and Class D (85) bars would be very similar in height, accurately reflecting the small real difference.
(a) Actual % increase = (102,000 − 98,000) ÷ 98,000 × 100 ≈ 4.1%. Sales did NOT double.
(b) With the axis starting at $96,000, the Class A bar goes from 0 (visual) to 2 units and Class D bar goes from 0 to 6 units — a visual ratio of 1:3. The compressed scale makes the 4% increase look enormous.
(c) Sales doubling would require the second bar to literally be twice as tall as the first on an axis starting at 0, meaning sales would need to rise from $98,000 to $196,000.
Flaw 1: Cherry-picking the time window (2010–2020 only). By omitting 2000–2010, the graph hides the cooling trend that occurred before 2010. The viewer forms a false impression of a consistently rising trend.
Flaw 2: No context for what came before 2010. Even if 2010–2020 is accurate, presenting it without the preceding decade creates a misleading overall narrative. A fair graph would include the full available time range.
"A politician shows a graph of unemployment dropping from 5.2% to 4.8% with the y-axis running from 4.5% to 5.5%. The graph makes it look like unemployment was cut in half."
(a) What was the actual percentage-point decrease in unemployment?
(b) On the truncated graph (axis 4.5% to 5.5%), what fraction of the way up is 5.2%? What fraction is 4.8%? Calculate the visual ratio between the two bars — does the graph make it look like it roughly halved?
(c) Redesign the graph with a fair axis. Describe how the bars compare on the fair version.
(d) Is the politician's claim "unemployment has been dramatically cut" justified by the data? Explain.