Statistics can mislead even when every number quoted is technically correct. A graph with a truncated axis, a sample chosen from a biased group or an average selected to favour a particular conclusion can all create a false impression. Spotting these tricks is a key GCSE skill.
What is a truncated axis and why does it mislead?
A truncated axis starts above zero, so the bars or line graph appear to show a much larger difference than actually exists.
Example: A company's sales last year were £980 000 and this year £1 020 000 — a real increase of about 4%. If the y-axis runs from £950 000 to £1 050 000 rather than from £0, a bar for this year looks about five times as tall as last year's bar, suggesting a far more dramatic rise.
How to spot it: Look at where the y-axis starts. If it does not start at zero (or if a "zigzag" break symbol is shown), any visual comparison between bars or points is distorted. The actual values printed on the bars or points are still correct — it is only the visual impression that misleads.
In an exam question: If asked "give one reason why this graph might be misleading", state that the y-axis does not start at zero and explain that this makes differences appear larger than they are.
How does sample bias create misleading statistics?
A biased sample is one that does not fairly represent the whole population. Results from a biased sample cannot be generalised reliably.
| Type of bias | Example | Why it misleads |
|---|---|---|
| Voluntary response | Online poll: "Do you think our product is excellent?" | Only enthusiastic customers respond |
| Convenience sample | Asking only your friends | Not representative of the full population |
| Leading question | "Don't you agree our school lunches are poor?" | Question nudges respondents towards a specific answer |
| Small sample | Concluding from 5 people | High random variation; results unstable |
| Timing bias | Surveying shoppers on a Tuesday morning | Misses people who work weekdays |
When critiquing a data-collection method in an exam, always state who is included and who is excluded, and explain why this makes the sample unrepresentative.
Why does the choice of average matter?
There are three different averages — mean, median and mode — and each can be chosen to tell a particular story.
Example: An estate agent wants to attract buyers. Average house price in an area:
- Seven houses priced at: £150k, £155k, £160k, £165k, £170k, £180k, £900k
- Mean: (£150k + £155k + £160k + £165k + £170k + £180k + £900k) ÷ 7 = £268 571 — inflated by the mansion
- Median: the middle value = £165 000 — more representative
- Mode: none (all different)
The estate agent might quote the mean to imply the area is upmarket. A buyer should ask which average was used.
Key rule: whenever you see a claim about an "average", ask: mean, median or mode?
How can graphs misuse percentages?
Example 1 — missing sample size: "Sales up 50%!" sounds impressive. But if you sold 2 items last year and 3 this year, that is a 50% increase from a tiny base — misleading because the absolute numbers are tiny.
Example 2 — percentages that don't add up: A pictogram or pie chart where the sections sum to more than 100% has been drawn incorrectly — or respondents were allowed to choose more than one option and the results presented as exclusive categories.
Example 3 — changing the base: "Crime fell 10% this year" could be misleading if the government changed how crime is recorded — the fall might reflect counting methodology, not actual crime.
How do pictograms mislead?
Pictograms use picture symbols, each representing a fixed quantity. A 2D pictogram misleads when the maker doubles the height of a symbol to show doubled values, but doubles the width too — so the area quadruples, greatly exaggerating the difference.
How to spot it: Read the key, count the symbols, and calculate the actual values. Trust the numbers, not the picture size.
What clues should you look for in an exam question?
- Check the axes — where do they start? Is there a gap or break symbol?
- Identify the sample — who was asked? How many? When?
- Identify the average used — mean, median or mode? Are there outliers that would skew the mean?
- Read the percentages critically — is the base given? Do they add up?
- Look at the scale — is it consistent? Does a bar chart have bars of equal width?
Frequently asked questions
Will GCSE exam questions always tell me explicitly that a graph is misleading?
Not always. Some questions ask "comment on the reliability" or "state one criticism" of a statistical claim. Treat these as an invitation to identify any of the problems above: biased sample, truncated axis, inappropriate average, small sample size, missing context, or inconsistent scale. One clear, specific criticism earns the mark more reliably than a vague statement that "the data may not be representative".
Can a graph be technically accurate but still misleading?
Yes — and this is the key insight. A graph with a truncated axis shows the correct values at each data point; it is the visual scale that misleads, not the numbers. Similarly, a mean calculated correctly from a biased sample is arithmetically correct but statistically unreliable. Statistical literacy means evaluating whether a claim is appropriate, not just whether the arithmetic is right.
What is the difference between a biased sample and a random sample?
A random sample gives every member of the population an equal chance of being selected, which minimises systematic bias. A biased sample systematically over- or under-represents certain groups. In practice, perfect random sampling is difficult — telephone surveys miss people without phones, online polls miss people without internet access. Real-world samples almost always have some bias; the question is how serious it is.
How should I phrase my answer to a "why is this misleading?" exam question?
Be specific. Do not write "the data could be wrong". Instead, write: "The y-axis starts at [value], not zero, so the bars appear to show a much larger difference than the actual [x%] change." Or: "The sample was collected from [specific group], which is not representative of [the full population], so the conclusion may not apply generally." One specific reason with a clear explanation earns full marks.
Want to sharpen your GCSE statistics skills with Professor Pi? Visit AI Tutors for guided practice.