For instance, if both our exposure and outcome predict participation (ie, collider bias), this can bias the relationship between these variables. (Image from doi.org/10.1038/s414...)

(Still a bit confused? Don’t worry you’re in good company - Selection bias is a notoriously counter-intuitive and slippery concept. For some nice intros, see doi.org/10.1093/ije/...doi.org/10.1038/s414...doi.org/10.1097/EDE....)