A statistical technique that removes the effect of one variable on another to isolate independent signals.