Two variables at once
Everything in chapters 4 and 12 tracked one variable at a time. A scatter diagram shows two variables together, one point per individual, so you can see whether the variables move with each other. Ten students recorded their revision hours and their test scores:
| Hours revised | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| Test score | 35 | 42 | 47 | 55 | 58 | 66 | 70 | 76 | 82 | 88 |
The scatter diagram, with a line of best fit and one outlier
Example 1Describe the relationship shown by the scatter diagram above.
Describing correlation
Positive correlation
As one variable increases, the other tends to increase. Points climb from bottom-left to top-right. Height and arm span; hours revised and test score.
Negative correlation
As one variable increases, the other tends to decrease. Points fall from top-left to bottom-right. Age of a car and its value; temperature and coat sales.
No correlation
The points form a shapeless cloud — knowing one variable tells you nothing about the other. Shoe size and test score.
Add a strength word: strong when the points lie close to a line, weak when the trend is there but the cloud is loose. “Weak negative correlation” is a complete, mark-worthy description; “quite downhill” is not.
The line of best fit
When the correlation is reasonably strong, you summarise the trend with a single straight line drawn by eye with a ruler. There is no calculation at IGCSE — just three rules:
Rules for a good line of best fit
- It follows the direction of the trend of the points.
- It balances the points — roughly as many above the line as below, all along its length.
- It does not have to pass through the origin, and it does not have to hit any actual point (though it may).
Correlation explorer
Switch between four real data sets, toggle the line of best fit, and try a prediction — watch what happens when you predict outside the data.
Using the line to predict — and when not to
Inside the data: interpolation
Reading the line within the range of the plotted points gives a sensible estimate — the line is supported by data on both sides. This is what “use your line of best fit to estimate…” wants: dashed lines to the axis, read off, done.
Outside the data: extrapolation
Extending the line beyond the plotted points assumes the trend continues — and nothing in the data supports that. The mark-scheme word is unreliable. (A car’s value line, extended far enough, predicts a negative price.)
Example 2Use the line of best fit on the first diagram to estimate the score of a student who revised for 6 hours.
Example 3Explain why the line should not be used to estimate the score of a student who revised for 20 hours.
More worked examples
Example 4On the first diagram, one student revised for 9 hours but scored only 40. What is this point called, what effect would it have if included when drawing the line, and suggest a possible reason for it.
Example 5Four students each draw a “line of best fit” on the same diagram: A joins the first point to the last point; B follows the trend with about half the points on each side; C passes through the origin because “no revision means no marks”; D joins every point dot-to-dot. Whose line is correct, and what is wrong with each of the others?
Your turn
Practice set
- State the type of correlation you would expect between: (a) a person’s height and arm span; (b) outdoor temperature and jumper sales; (c) shoe size and maths score.
- Write the one-sentence description, in context, for the relationship in 1(b).
- A best-fit line for car age (years) against value ($1000s) passes through (0, 20) and (8, 4). How much value does the line suggest a car loses per year, and what value does it predict at 5 years old?
- Explain why the line in question 3 should not be used for a 15-year-old car.
- True or false: a line of best fit must pass through at least one plotted point. Explain.
- (a) positive (b) negative (c) no correlation
- The higher the temperature, the fewer jumpers are sold. (“colder → more jumpers” is equally good.)
- Drop of 20 − 4 = 16 over 8 years → $2000 per year. At 5 years: 20 − 5 × 2 = $10 000.
- 15 years is beyond the data the line was drawn from — extrapolation is unreliable (indeed the line predicts 20 − 30 = −$10 000, an impossible negative value).
- False — the line summarises the trend and balances the cloud of points; it may pass through none of them.
Key points to remember
- Bivariate data plots as one point per pair — never join the points.
- Describe correlation with strength + type + a context sentence: “strong negative — the older the car, the lower its value.”
- Correlation is not causation — a third factor may drive both variables.
- A line of best fit is one ruled line that follows the trend and balances the points; it need not pass through the origin or any point.
- Ignore an outlier when drawing the line, and be ready to suggest a context reason for it.
- Estimating inside the data range is fine; outside it is extrapolation — the mark-scheme word is unreliable.
- Show the dashed reading lines whenever you use the line to estimate.
Where next?
🎥 Take the interactive module
Representing Data — the full teaching module for this topic, with real past-paper questions and mark schemes.
✏️ Practise exam questions
Statistics — 25 exam-style questions worth 99 marks, with full answers and the mark scheme.
Progress is saved on this device only and appears on the notes home page.