Jacques Barzun on multiple-choice tests:
Multiple-choice questions test nothing but passive-recognition knowledge, not active usable knowledge. Knowing something means the power to summon up facts and their significance in the right relations. Mechanical testing does not foster this power. It is one thing to pick out Valley Forge, not Dobbs Ferry or Little Rock, as the place where George Washington made his winter quarters; it is another, first, to think of Valley Forge and then to say why he chose it rather than Philadelphia, where it was warmer.Why Valley Forge? The National Parks Service explains.
Multiple-choice tests, whether of fact or skill, break up the unity of knowledge and isolate the pieces; nothing follows on anything else, and a student’s mind must keep jumping. True testing elicits the pattern originally learned; an essay examination reinforces pattern-making. Ability shows itself not in the number of accurate “hits” but in the extent, coherence, and verbal accuracy of each whole answer. Science and math consist of similar clusters of thought, and, in all subjects, to compose organized statements requires full-blown thinking. Objective tests ask only for sorting.
“The Tyranny of Testing” (1962). In A Jacques Barzun Reader, edited by Michael Murray (New York: HarperCollins, 2002).
A related post
Whitehead on primrose paths and external examinations
[A relevant anecdote: A student once asked, only semi-seriously, if our final exam would be multiple-choice. In life, said I, there are no multiple-choice tests. People expect you to develop answers, not choose them. A second student suggested that there was indeed one multiple-choice test in life: marriage. No, said a third student, marriage is a true-false test. No, said I, marriage is a matching test. It was a lively class, with an essay exam.]