When participants self-select, or volunteer for a study, it is not known how representative such participants are of the population of interest. If such individuals entering or remaining in a study display different associations from those who do not, a biased estimate of the association between exposure and outcome is produced.

Sometimes people we want to study may not be available for study. This can happen in an occupational setting where the exposed workers suffering symptoms are the ones we would like to study, but may have left their jobs. Those who remain are the ones less affected.

Healthy worker effect – those severely ill or disabled are ordinarily excluded from employment.