Collider bias is a statistical phenomenon that occurs when a researcher adjusts for a common outcome variable that is shared between a predictor and its criterion. This biased adjustment can happen in one of two ways: (1) treating the collider variable as a confounder and adjusting for it during statistical analysis, or (2) incidental sample selection on the collider variable. Improperly addressing collider bias can result in invalid estimates of the population relationships psychologists wish to measure, often attenuating the sample estimates toward 0 and reducing power to detect a significant effect.