Pulse Brain · Growing Health Evidence Index
Tier 3 — Observational / field trialPeer-reviewed

Collider scope: when selection bias can substantially influence observed associations

Marcus R. Munafò, Kate Tilling, Amy E. Taylor, David M. Evans, George Davey Smith

International Journal of Epidemiology · 2017

Read source ↗ All evidence

Summary

This methodological paper addresses a critical but underappreciated source of bias in epidemiological research: the role of selection bias in inducing collider bias. The authors argue that whilst selection bias is known to affect representativeness and prevalence estimates, its substantial impact on association estimates is often overlooked. Through simulation evidence, they demonstrate that selection mechanisms linked to phenotypes can substantially distort genetic and phenotypic associations, and propose that understanding and adjusting for factors influencing study selection and attrition—such as through polygenic score sensitivity analyses—is essential for valid causal inference.

UK applicability

These findings are directly applicable to UK birth cohorts and epidemiological studies such as the Avon Longitudinal Study of Parents and Children (ALSPAC) and UK Biobank, where selection and attrition are known challenges. UK researchers conducting association studies should consider implementing the suggested sensitivity analyses and adjustment approaches to quantify potential collider bias.

Key measures

Bias in phenotypic and genotypic association estimates under varying selection and attrition scenarios; effects of conditioning on colliders; polygenic score predictive validity for study participation

Outcomes reported

The study examined how selection bias and attrition in large-scale cohort and cross-sectional studies can induce collider bias, leading to substantially biased estimates of both phenotypic and genotypic associations. Through simulation, the authors demonstrated that even modest influences on study selection and attrition can generate misleading association estimates.

Theme
Measurement & metrics
Subject
Measurement methods & nutrient profiling
Study type
Research
Study design
Simulation study with methodological analysis
Source type
Peer-reviewed study
Status
Published
Geography
International
System type
Human clinical
DOI
10.1093/ije/dyx206
Catalogue ID
BFmor3gaas-2mw6x9

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.