Failure by Design: Why Tests Don’t Show What Students Can Do

No standardized, norm-referenced test, assert Coalition leaders Theodore R. Sizer and Deborah Meier, can measure the real payoff from serious study-“an examined and useful life,” as Sizer puts it-nor can it describe the good school that works to achieve that end.

“Any assessment that correlates poorly with a student’s intellectual future offends us, putting stress on teachers and students and taking up their time,” Sizer says. “Essential schools, instead, call for evidence that a student not only understands something now, but can use that understanding in a new situation, now and very likely in the future.”

A mountain of evidence on how people learn, from two decades of research on cognition, bears out that position. If their goal is real understanding on the part of students, Howard Gardner and others have shown, teachers must use a variety of approaches, allowing students with very different kinds of intelligence to increase their strengths and to struggle productively with areas they find difficult.

The standardized tests in wide use across the country today do not support this kind of teaching-nor, most research shows, do they predict later academic success, except at the extremes. Success later in life, in fact, correlates better with learning experiences that are rarely tested, such as extracurricular activities, data from the National Center for Education Statistics and American College Testing indicate.

“Conventional tests spotlight children who have certain abilities-especially memory and abstract-analytical ones-but leave in the dark children with other kinds of abilities, such as creative and practical ones,” says Yale professor Robert Sternberg, the author of Successful Intelligence (Simon and Schuster, 1996). “They predict only about ten percent of the variation among people in real-world measures of success-the ones that matter most in the life activities for which school is supposed to prepare our children.”

If schools instead identify and teach to students’ strengths, Sternberg’s studies show, kids with practical and creative abilities perform far better than if only their memory and abstract analytical abilities are emphasized and tested.

Early in their lives, he argues, we are derailing from the fast track students who never get the opportunity to show what they can really do-at the same time providing almost limitless opportunities for others who may not have the practical and creative abilities they will need to use their other skills effectively in the real world.

The very design of all norm-referenced tests ensures this outcome, Deborah Meier points out. Because their goal is a bell-shaped curve of scores on which certain students must by definition do poorly, test-makers actually discard items when too many, too few, or the “wrong” subsets of test takers (such as disadvantaged children) know the answer. Otherwise, she explains, the results would not correlate to all the other norm-referenced tests (from IQ tests on) with which they must agree in order to be valid. “All these tests predict,” she says, “is how students will perform on other tests just like them.”

Though some new tests attempt to measure students’ progress against standards of content and performance, they fall into the same trap the moment they start yielding percentile scores, Meier observes. And if they try to establish and test content that every student must know, they must either eliminate diversity to the point that they are trivial and inunteresting-or individualize exams until they cost far more than schools can pay.

“What if the driving tests were deliberately made up so that 50 percent of all license seekers would not be able to pass?” she asks. “We have an auto industry that would squash such nonsense in a hurry. What industry is making sure that our tests are sensible and fair to all children?”