A set of variables that must be controlled for in analysis to isolate the true causal effect of a treatment.