The canonical bias-and-validity reference for all research — every named bias paired with a concrete control, internal vs external validity separated. A CANONICAL REFERENCE other skills load before designing a study or trusting a finding, the way craft-critique is loaded for evidence discipline. Use when someone asks 'is this biased?', 'what biases are in play?', 'validity threats', 'will this generalize?', 'how do I control for X', 'is my sample skewed', 'non-response / self-selection / survivorship', or before fielding any study and before believing any result. Names biases as an engineering problem with controls — never 'just be objective'.
