a scoring rule for multiple-choice tests such that the expected value of getting an item correct under the assumption of no knowledge is 0 rather than 1/n, where n is the number of alternatives. Thus, if a test taker incorrectly answered four questions, a further 4/n would be deleted from his or her score. By thereby penalizing inaccurate responses, the correction discourages a test taker from guessing and provides a better estimate of that person’s true ability.