A method used to control the probability of making a Type I error in experiments with multiple tests.