What Is Selection Bias?

Selection bias occurs when the people included in a study (or analysed) differ systematically from the population of interest or from each other in ways that affect the results. It arises from how participants are chosen, recruited, retained or analysed.

Common forms

• Sampling or ascertainment bias: the sample does not represent the population (for example a hospital sample for a community question).
• Non-response bias: responders differ from non-responders.
• Attrition (loss to follow-up) bias: those who drop out differ from those who stay.
• Healthy volunteer bias: participants are healthier than average.
• Berkson's bias: hospital-based samples show spurious associations because admission depends on both conditions.
• Control selection bias (case-control studies): controls do not come from the same source population as cases.
• Allocation bias in trials: assignment is predictable or manipulated, so groups differ.

Reducing it

1. Define the target population and sampling frame clearly.
2. Use random or consecutive sampling where possible.
3. Maximise response and follow-up, and document reasons for loss.
4. Select controls from the same population as cases.
5. Conceal allocation in trials.
6. Compare responders with non-responders on known characteristics.

Illustrative example

Invented for teaching. A survey about sleep is advertised on social media. People with sleep problems are more likely to respond, so the estimated prevalence of insomnia is too high. The authors should describe the recruitment method and treat the prevalence as non-generalisable.

Spotting it when you read

Look at the participant flow diagram and the table of baseline characteristics. Large numbers excluded or lost, response rates reported without comparison with non-responders, or recruitment through channels that favour particular groups are warning signs. Ask: if I could see everyone who was eligible, would the findings be the same?

Common pitfalls

• Assuming a large sample prevents selection bias.
• Not reporting response rates.

Key takeaways

• Selection bias depends on how people get into and stay in the study.
• Prevent it by design; assess it by reading methods and flow diagrams.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.