Health data, study design and perceived risk

Correlation is not causation

A correlation means two variables change together. It does not prove one causes the other: a third factor may cause both, or it may be chance. To argue for causation you want:

• a plausible biological mechanism
• a consistent link in many independent studies
• a dose-response pattern (more exposure, more disease)
• evidence from controlled trials where changing the factor changes the outcome.

When studies conflict, compare their size, design, how well they controlled other variables and who funded them.

Judging a study

• Sample size: larger samples reduce the effect of chance, so results are more reliable (repeatable).
• Sample selection: a representative sample (age, sex, ethnicity, lifestyle) lets results be generalised. Matching groups for other variables improves validity.
• Controls and placebos: a control group allows comparison.
• Validity means the study measures what it claims to; reliability means similar results would be found if repeated.
• Peer review and repetition by other scientists add confidence.

Risk is often given as a rate: for example deaths per 100 000 people per year. Relative risk compares the rate in an exposed group with the rate in an unexposed group.

Perceived and actual risk

People often overestimate risks that are rare, dramatic, unfamiliar or outside their control (such as plane crashes) and underestimate familiar, long-term risks they choose (such as smoking, poor diet or inactivity). Optimism ('it won't happen to me'), media coverage and delayed effects all distort perception.

Worked example

Smokers in a study have 60 heart attacks per 100 000 people per year; non-smokers have 20. What is the relative risk for smokers?

1. Relative risk = rate in exposed / rate in unexposed
2. = 60 / 20

Answer: 3: smokers have three times the risk.

Check yourself

1. In a study, people with high blood pressure had 45 strokes per 10 000 per year; those with normal blood pressure had 15. What is the relative risk?
   Show answer

   3. 45 / 15 = 3.

2. In a town of 250 000 people, 500 deaths from CHD occur in a year. What is the rate per 100 000 people per year?
   Show answer

   200. 500 / 250 000 x 100 000 = 200.

3. A study finds that people who drink more coffee have more heart disease. What can you conclude?
   • Coffee causes heart disease
   • Heart disease causes people to drink coffee
   • There is a correlation, but coffee is not proven to cause heart disease
   • There is no link
   Show answer

   There is a correlation, but coffee is not proven to cause heart disease. A correlation could be due to another factor, such as smoking, linked to both.

4. Why does a large sample make a study more reliable?
   • It guarantees the study is valid
   • It reduces the effect of chance and unusual individuals on the result
   • It makes the study cheaper
   • It removes the need for a control group
   Show answer

   It reduces the effect of chance and unusual individuals on the result. Larger samples average out random variation.

5. Two groups in a diet study differ in average age. Why is this a problem?
   • It makes the sample larger
   • Age has no effect on CVD
   • It improves reliability
   • Age is another risk factor, so differences may not be due to diet
   Show answer

   Age is another risk factor, so differences may not be due to diet. Uncontrolled variables reduce validity.

6. Which risk are people most likely to underestimate?
   • Risk of dying in a plane crash
   • Long-term risk from a poor diet
   • Risk of a shark attack
   • Risk from a rare new virus in the news
   Show answer

   Long-term risk from a poor diet. Familiar, voluntary, long-term risks tend to be underestimated.

Practise with instant feedback Print this lesson

Sources

• Pearson Edexcel International Advanced Level in Biology specification (XBI11/YBI11, first teaching 2018)

Last updated 7 November 2026. Spotted a mistake? Tell us.