Congress is on the verge of enacting what may turn out to be the largest and most far-reaching mandate on elementary and secondary schools in 25 years: President Bush's plan for annual testing of students in grades 3-8. Remarkably, however, there has been almost no discussion of the magnitude of this proposal. As a result, the president has thus far failed to articulate a policy to make his plan effective, and Congress has not forced him to do so.

Instead of engaging in discussion of annual testing, the president has cast the issue as a choice between his plan and perpetuation of an unacceptable status quo. Secretary of Education Rod Paige wrote on this page [May 13] that "anyone who opposes annual testing of children is an apologist for a broken system of education."

That sort of rhetoric is irresponsible. By framing the issue in such stark terms, it stifles debate on this far-reaching proposal and obscures the complexity of the issue. The question confronting Congress is not whether to require tests; the policy arguments in favor of doing so are clear, and rising test scores for minorities in Texas and elsewhere demonstrate the potential of effective standards-based reform. Instead, the issue is how to design a testing policy that is more than a cosmetic victory for the president and will improve the quality of education for students.

In 1994 the last reauthorization of the Elementary and Secondary Education Act (ESEA) required states to test students at least three times in math and reading during their schooling to help increase accountability for schools and school districts. Since then every state has worked to develop standards and assessments to accomplish this.

The Bush plan is a quantum leap beyond this requirement; it would mandate the development of at least 260 additional tests nationwide and require each state to integrate the additional tests into its evolving standards and assessment system. Congress and the president must ensure that the final legislation is as robust about the quality of these tests as it is about the quantity.

First, all testing is not equal, and the tests in use around the country are of varying quality. In the near term, Congress can't solve this problem, but it can require states to develop high-quality tests that are not just multiple choice or "fill in the bubble" tests but require students to construct answers and write essays. In addition, Congress can require that these tests measure student performance against established benchmarks, rather than just the performance of other students. These "criterion-referenced" tests are more expensive to develop but more useful for educators.

Second, Congress should require states to use the same tests statewide and to hold schools accountable not only for overall student scores but for the scores of minority and disadvantaged students too. Allowing different tests within different parts of a state, as the president appears willing to accept, reopens the door for different expectations for different groups of students.

Third, a realistic discussion of the costs of this proposal is essential. Maryland, which is considered to have a high-quality test, spends about $30 per student for each testing. Development of tests costs far more. While some estimates of the cost of the president's proposal are clearly exorbitant and politically motivated, the $320 million he has proposed to divide annually among the 50 states for test development is far too little to be meaningful. Leaving this testing mandate unfunded will drive states to use lower-quality, preexisting commercial tests that frequently are not in alignment with the curriculum being taught in schools.

Finally, states will need time to develop these additional tests and should be encouraged to do the job right rather than fast. States such as Virginia that have rushed this process have encountered predictable resistance and implementation problems. And rather than each of the states reinventing the wheel on math and reading, collaboration among them should be encouraged. While to conservatives this reeks of "national testing," it is more a matter of simple economics and common sense. Reading and math really aren't all that different among the various states.

Properly implemented and supported, the president's call for annual testing can result in better schooling for millions of children, particularly many who are not receiving a high-quality education now. But if the president simply accepts a lowest-common-denominator policy on testing to avoid a thorough debate or political complications, he not only will accomplish little but could actually hamstring the move for better standards and tests.

Andrew Rotherham is director of education policy at the Progressive Policy Institute and a former education adviser to President Clinton.