Selection bias

Selection bias is systematic error caused when inclusion in a study depends on exposure, outcome, or their causes, making the studied participants unrepresentative of the target population.

Collider bias: Bias produced by conditioning on a variable caused by two other variables. Selecting participants through a common effect of exposure and outcome can create a spurious association.
Causal inference: Methods for estimating cause-and-effect relationships from data. Selection bias threatens whether an observed association supports a causal conclusion.
Case-control study: An observational study that samples people by outcome status and compares their prior exposures. Choosing controls from the wrong source population can bias exposure comparisons.
Information bias: Systematic error caused by inaccurate measurement or classification of study variables. It concerns how variables are measured, whereas selection bias concerns who is included.
Berkson's paradox: A spurious association that appears when analysis is restricted to a selected group. Hospital-based sampling can link otherwise independent conditions because either condition increases admission.
Target population: The full group about which a study aims to draw conclusions. Bias depends on how the analyzed sample differs from the population the study targets.
Cohort study: An observational study that follows a defined group to compare outcome occurrence across exposures. Enrollment and continued participation can select which exposure groups contribute outcome data.
Publication bias: Distortion in the available research literature when publication depends on study results. It selects studies into the evidence base rather than participants into an individual study.
Survivorship bias: Error from analyzing only entities that persist or succeed while excluding those that disappear or fail. Survival to recruitment or measurement can select participants based on factors related to outcomes.
Sampling frame: The list or procedure that identifies units eligible for selection into a sample. Its coverage and recruitment rules determine who can enter the study.

Linked from 28 pages