Grading Machines

The most prestigious test factory in the U.S. is in Princeton, N. J. Last week the factory worked overtime cooking up new devilments for defenseless scholars. An old client, the Navy, wanted a new test to pick 5,000 candidates for Annapolis.

The Navy job was the latest model of objective ("pick-the-right-answer") aptitude test, the current fashion in exam-making. The no-man test factory, a nonprofit outfit called the College Entrance Examination Board, tailors special exams to order for the State Department, the Pepsi-Cola Co., many another customer. But mostly it works out mechanical ways of measuring who should and who shouldn't be admitted to 55 member colleges, mainly in the East (among them Harvard, Yale, Princeton, Vassar, Smith).

In 1901, after Nicholas Murray Butler got the board started, the first 973 test-takers were private-school graduates who had taken pretty much the same studies. It was easy for the board to hit on a half-dozen broad essay questions, fairly testing their factual preparation, their grasp of ideas, their literary style. The 46,087 boys & girls who took the College Boards last year were just too many and various to grade in the old way; more than half of them were public-school students--products of a dozen different curriculums. Four years ago, the board tossed out the essay-type exam entirely (except in English composition).

Old-fashioned educators might argue that there are certain qualities in an educated man too elusive for an I.B.M. machine to measure, but the board insists that its present multiple-choice exams cover a much wider range of the student's preparation than the essay type ever did. Explains the board's new director, Harvardman Henry Chauncey: "We get a large number of candid-camera shots of the individual, 150 or more, instead of six or eight posed photos." The new exams are also quicker to take, quicker to mark (by I.B.M. machine)--and also eliminate the effect of the marker's prejudices and the state of his digestion.

Worthy Distractions. New questions for C.E.E.B. exams are "pretested" on students who are classified from ace to dunce. The test factory writes a biography of each "item" on a card and studies it. If the best students have picked the right answer and the worst ones muffed it, the "item" is ready for use. But if too many good students tripped, the test constructor knows that the wrong choices he offered were not "fair distractors," and he words the problem over again.*

The trend in C.E.E.B. exam-making is steadily away from achievement tests (what do you know?) toward aptitude tests like the Navy exam (how well can you learn?). The aptitude tests, says Chauncey, are "about as good a single index of future success in college as there is."*

But the board's 20 testmakers, not satisfied yet, are working out the super test of tomorrow: something that will measure new aspects of the student's personality. By making him perform the same simple operation over & over, for example, they hope to figure out his "persistence quotient" (one reason why steady, mediocre tortoises sometimes nose out brilliant but unstable hares).

* In academic lingo, the discarded question has a low "biserial coefficient of correlation."

*There are three types: verbal (measuring a student's vocabulary and his ability to combine ideas), mathematical (measuring his knack with problems, not his memory of formulas) and spatial relations (measuring his grasp of the three-dimensional world by means of regular and odd-shaped solids which he must count, combine into new shapes or interchange--a process in which many an unscientific intellect feels like a lost and slightly mad sheep).

This file is automatically generated by a robot program, so reader's discretion is required.