The pitfalls of annual testing
NEW YORK — I am pleased that the new education bill provides more resources for poor children, but am concerned that it mandates annual math and reading tests in grades three through eight. As states gear up to meet this demand, the problems encountered during the past decade in developing standards-based testing are likely to be exacerbated.
In taking on this increased burden, states may be forced to forgo the most promising approaches to testing. Consider the testing of writing, which requires students to write real words rather than fill in bubbles. Real human beings are required to evaluate the writing, and more than one person, to ensure fairness, must score each piece, so the process becomes costly. Since annual testing will drive up costs, the more expensive approaches to testing are likely to be sacrificed - but it is they that have the most beneficial influence in the classroom: When students must write on a test, they do more writing throughout the school year.
As Sen. Jim Jeffords has pointed out in opposing the bill, the appropriated funding is far from adequate. Only $400 million, for example, is provided for annual testing in the nation as a whole. This shortfall will force states to use more multiple-choice testing, although research has shown its detrimental effects on education. Even a cursory examination of the history of testing teaches a clear lesson: If enough money is not available, testing is done on the cheap and children are the losers.
What we are likely to see nationwide is already clear in the 15 states with annual math and reading tests in grades three through eight. All except Tennessee use multiple-choice tests such as the Stanford 9, even though these commercially produced tests are not in line with these states' standards. Off-the-shelf tests are certain to be used more widely in other states.
Even states that have tests aligned with their own standards use multiple-choice questions to test reading, even though they are ill-suited for tapping into the more complex thinking involved in reading. Indeed, research shows the "wrong answers" are often designed to encourage such thinking, in contrast to the "right answers," which tend to focus on surface details.
To make matters worse, when faced with important tests, educators often turn to coaching material of dubious quality. In a recent study of such material in urban schools that serve poor children, a team of doctoral students and I discovered a good deal of faulty logic and factual error in the explanations for how to answer multiple-choice questions. We also observed teachers using this material in a mechanical way that deadened children's engagement with what they were reading.
I fear that the greater resources provided urban schools by the new bill could lead to even greater reliance on coaching material. I am especially fearful that the distortion of the curriculum I have observed - with much less time given to subjects such as art and science - will only increase as children prepare for annual tests in math and reading.
The most unfortunate consequence of annual testing is that it undermines a comprehensive approach to assessing children's work throughout the year. Those of us who have worked with teachers to develop these approaches often balance extended projects with testing based on responses that the children construct themselves. Unlike multiple-choice, this testing requires children not only to focus on factual content but also to make appropriate inferences based on this content.
Unfortunately these approaches are breaking down as the school day is increasingly taken up with test prep. Children now spend their time passively responding to coaching material instead of doing work that engages their intelligence.
Clifford Hill is the Arthur I. Gates Professor of Language and Education at Teachers College, Columbia University. His most recent book is 'Children and Reading Tests' (Elsevier, 2000), coauthored with Eric Larsen.