Evaluating the design of studiesSpec 1.16
In short
A well-designed health study uses a representative sample, selected without bias, that is large enough for chance variation to have little effect. Data are valid if they measure what they claim to and control other variables, and reliable if the results can be repeated. Self-reported data, small samples and uncontrolled variables reduce the trustworthiness of a study.
- Valid
- The data measure what the study claims to measure, with other variables controlled.
- Reliable
- The results are consistent, and similar results would be obtained if the study were repeated.
| Feature | What makes it better | Why |
|---|---|---|
| Sample selection | People who represent the whole population (age, gender, lifestyle), chosen at random | Avoids bias, so the findings can be applied to the population; volunteers may be unusually health-conscious |
| Sample size | Large | Reduces the effect of chance and of anomalous individuals; the results are more reliable |
| Control of variables | Groups matched for age, gender and other factors, with a control group | Reduces the effect of confounding variables, so any difference is more likely to be due to the factor |
| Data collection | Objective, measured values (blood pressure, cholesterol) | Self-reported diet and exercise rely on memory and honesty |
| Duration | Long enough for the disease to develop; follow people over time | CVD takes years to develop |
In a cohort study a large group is followed over time and the rate of disease is compared between people with and without a factor. In a case-control study people with the disease are compared with matched people without it. Both can show a link but cannot prove a cause.
Do not just write 'bigger sample'. Say why: a larger sample reduces the effect of chance and anomalies, so the data are more reliable. Link bias in selection to validity.
Written and checked against the Edexcel International AS Level Biology (XBI11) specification · Updated October 2026