The national movement to establish “high stakes” tests for students, and the widely publicized problems in several states with the implementation of such policies, provided a dominant theme last week as education researchers gathered here for their annual meeting.
About half the states now use tests to make such important decisions as whether students will move on to the next grade or graduate from high school, and which schools qualify for rewards or penalties. But in many of the conference sessions devoted to state assessments and accountability programs, researchers expressed concerns that such tests are being widely misused.
Rather than improving student learning, some of the scholars suggested, high-stakes tests may be narrowing the curriculum, stultifying teaching practices, and driving some teachers and school administrators to cheat.
“We put teachers in a situation where they cannot teach properly to the needed knowledge and skills, and we just sit there and let it happen,” said W. James Popham, a professor emeritus of education and information studies at the University of California, Los Angeles.
He spoke at one of more than 30 sessions on accountability programs at the April 24-28 meeting here of the American Educational Research Association, which drew more than 13,000 researchers.
But the AERA, whose 23,000 members include a large number of state and district testing directors and other experts on assessment, was not yet ready last week to take that kind of stand. A proposed resolution calling for states and districts to adhere to a set of agreed-upon standards for student testing failed to make it out of the group’s governing council.
The association will take up the issue when its council meets next month. But some of its leaders warned against waiting, saying that debates on the issue are now taking place in many states.
“Having a document that is 98 percent perfect at the time we need it is better than a 100 percent perfect document two months later,” said Catherine Snow, the organization’s new president and a professor of education at Harvard University.
Concerns About Cheating
One problem with the move to high-stakes assessment, many researchers said, is that policymakers misunderstand the purposes for which they were designed. A graduation or promotion test, for example, should reflect what students have been taught—not how they stack up against students across the country on a set of general, unrelated skills.
“The Iowa Test of Basic Skills is not valid as a measure of who should be promoted or who should be retained in the city of Chicago or anywhere else,” said Jay P. Heubert, an associate education professor at Teachers College, Columbia University, and an adjunct professor at the university’s law school.
Just as troubling, researchers added, are vague state curriculum standards or guidelines from test publishers that do not make clear to educators the content the tests are designed to assess.
One heavily attended session, “School Scandals Waiting To Happen: Consequences of Current Pressures to Raise Test Scores,” explored the likelihood that, given a lack of direction and the pressure to raise their students’ scores, more educators may follow widely reported allegations about teachers in New York City, Austin, Texas, and other locations and resort to cheating.
At its most blatant, cheating might involve changing students’ answers or keeping poor-performing students from taking the tests. But, some researchers said, teachers are also cheating—sometimes unwittingly—when they memorize test questions from previous years and rehearse them with their students.
As an example of how not to use tests, many researchers cited Texas, which since 1990 has been using results from its state exams as the sole criterion for determining which students graduate and deciding which schools are at risk of a state takeover.
At another session, Linda McNeil, a researcher from Rice University in Houston, contended that test preparation had elbowed out better-quality instruction in many Texas schools.
“We hear stories about middle school children who read no prose from September to January,” she said. Instead, Ms. McNeil maintained, they are reading only the kinds of two- or three-paragraph selections commonly found on the state tests.
But the data from Texas were also in dispute at the convention. Some researchers said, for example, that the tests were driving more poor and minority students to drop out, while others contested that claim. Overall, the scores of poor and minority students on Texas Academic Assessment System tests have been rising.
“Linda McNeil assumes there’s some wonderful alternative out there,” said Martin Carnoy, a professor of education and economics at Stanford University, who described research he has conducted on Texas high schools at another session on state standards and accountability policies. “But what was happening before for those kids was probably even worse.”
Better Tests Sought
Despite the widespread concerns, several researchers said last week that they were not willing to give up on the kind of leverage that high-stakes testing programs provide for school reform.
“I’m probably the only living psychometrician who is in favor of high- stakes testing,” said Andrew Porter, a professor at the University of Wisconsin- Madison. “But I prefer it to be done well rather than poorly.”
What schools need are newer forms of assessment that are better attuned to how students learn, said Lorrie A. Shepard, a professor of research methodology at the University of Colorado at Boulder. As the AERA’s outgoing president, she made the accountability-testing movement the focus of her major address.
“Our goal should be to find ways to fend off the negative effects of externally imposed tests,” she said, “and to develop, instead, classroom assessment practices that can be trusted to help students take the next steps in learning.”
