Spotting Tricks in the Wild
Three real-world graph problems: an approval rating chart, a tech startup's growth graph, and two competing climate visualizations.
Work through each problem. Use the hint if you're stuck, and check your answers when you're done with each part.
A news network shows a bar chart of two candidates' approval ratings: Candidate A at 53%, Candidate B at 47%. The y-axis runs from 44% to 56%.
In the truncated chart, roughly how much taller does Candidate A's bar appear than Candidate B's?
Show answer
On the truncated axis (44%–56%), A's bar spans from 44% to 53% — 9 units. B's spans from 44% to 47% — 3 units. So A's bar looks about 3× taller than B's, even though A is only 6 percentage points ahead.
What would the chart look like if the y-axis ran from 0% to 100%?
Show answer
Both bars would be tall and nearly the same height. A's bar would reach 53% of the chart's height; B's would reach 47%. The visual difference between them would be barely noticeable — about 6% of the total chart height.
Does the truncated axis make the race look closer or more lopsided than it actually is?
Show answer
More lopsided. The 6-point gap looks enormous on the truncated chart but is actually within typical polling margins of error. Many elections with this gap are considered competitive.
Is there any situation where the truncated axis would actually be the more informative choice for this data?
Show answer
Possibly — if the goal is to show the precision of the measurement, or compare changes over time in a race where both candidates are known to be above 44%. But for a simple "who's ahead" comparison, starting at zero is almost always more honest.
HintShowHide
Try imagining (or sketching) the same chart with a y-axis from 0% to 100%. How tall would each bar be relative to the chart? Now compare that to the truncated version.
A tech startup shows investors a graph of their monthly active users. The y-axis runs from 9.4 million to 10.3 million. The line runs from the bottom-left to the top-right of the chart, looking steep and exciting.
Actual numbers: users grew from 9.5M to 10.2M over 12 months.
What was the actual percentage growth in users?
Show answer
(10.2 − 9.5) ÷ 9.5 × 100 = 0.7 ÷ 9.5 × 100 ≈ 7.4% growth over 12 months.
What would this graph look like with a y-axis starting at 0?
Show answer
The line would be nearly flat — a very slight upward tick at the right edge. The bars would take up roughly 95–102% of the chart's height, with almost no visible slope.
Is the growth real? Is the graph misleading? Can both be true?
Show answer
Yes, both can be true. 7.4% annual user growth is real and meaningful. But the chart presents it as if users nearly doubled. The data is accurate; the visual impression is misleading. This is the key tension in graph literacy: technically true ≠ honest.
HintShowHide
Percentage growth = (new value − old value) ÷ old value × 100. Use the actual numbers, not the chart.
You find two graphs of average global temperatures over the last 100 years. Graph A shows a steep upward curve. Graph B shows a nearly flat line that ticks up slightly at the end. Both use the same underlying data.
List three specific things you'd want to check about each graph before deciding which to trust.
Show answer
Any three of: the y-axis range (what temperature scale is used?), the x-axis range (what years are shown?), the data source, whether a trend line is smoothed or raw, the units (°C vs. °F), and whether uncertainty ranges are shown.
If both graphs are technically accurate, what might explain the visual difference?
Show answer
Almost certainly the y-axis range. Graph A likely zooms in on the range of observed variation (say, 13.5°C to 15°C), making the upward trend fill the chart. Graph B likely starts near 0°C, making the same change look tiny. Same data, completely different visual story.
If you were a scientist presenting this data to government officials and lawmakers, what choices would you make — and why?
Show answer
Good answers typically include: show the full time period, use a y-axis range that represents meaningful context (e.g., relative to pre-industrial baseline), include uncertainty bands, label the data source, and choose a scale that honestly represents the magnitude of change compared to what's considered dangerous warming. Honesty and clarity should override dramatic presentation.
HintShowHide
Focus on what the axes tell you. Check: what range does the y-axis cover? What years does the x-axis include? These two choices alone can make the same data look dramatically different.