Selection Bias

Selection bias is a distortion in a study's results caused by how people were chosen for it, or by who dropped out. If the people studied differ from the population they are meant to represent in a way related to both exposure and disease, the study's estimate of risk can be too high or too low.

Origin

Selection bias is a foundational concept of epidemiologic methods, set out formally as the field developed rules for designing cohort and case-control studies in the mid-20th century.

Significance in Asbestos Context

Selection bias is one of the most common critiques experts make of asbestos studies. A cohort that includes only workers still employed on a certain date misses those who already left because they were sick. A case-control study that draws its cases from a specialist center may over-represent people with clear occupational histories. Studies that lose track of many workers, or that rely on volunteers, may not reflect the full workforce. The healthy worker effect is a well-known type of selection bias. In court, both sides may use selection bias to argue that a study favoring the other side is unreliable, and judges consider such arguments when deciding whether expert opinions based on those studies can be heard. Selection bias does not automatically make a study worthless; good studies describe how subjects were chosen and test how much the bias could change the result. For readers, the key question is who was counted and who was left out.

Example Sentence

The plaintiff's expert argued that the mine study suffered from selection bias because it excluded workers who left before 1950, when dust levels were highest.

Related Terms

Healthy Worker Effect, Recall Bias, Confounding, Cohort Study, Case-Control Study, Daubert Standard, Epidemiology

Notes

• When a study is cited for or against your case, ask your lawyer who was included and who was left out.
• Studies of company workforces can depend heavily on company records; ask whether any groups were excluded.