Glossary
Selection Bias
What is selection bias?
Selection bias happens when the way observations enter or stay in an analysis distorts the result. Who gets studied matters, but so does who drops out or gets left out later.
Hernán, Hernández-Díaz, and Robins explain how choosing an inappropriate comparison group, losing certain participants during follow-up, and restricting an analysis in certain ways can distort causal comparisons.
Examples of selection bias
If a program evaluation reports only people who completed it, and people doing poorly were more likely to leave, the reported result can make the program look better than it was for everyone who started. Self-selection raises a related question: do volunteers differ in relevant ways from people who did not volunteer?
Selection bias vs. sampling bias
Sampling bias concerns how a sample differs from the population it is meant to describe. Selection bias can also arise after recruitment, through who remains in the analysis or which comparisons are made.
Why it matters
Check who declined, dropped out, or was excluded. Random sampling helps a study represent a population. Random assignment helps compare treatments. They solve different problems, and neither prevents every later source of selection bias.