The saga of national education testing won't surprise those who have watched other reformist projects turn sticky and foul after brushing against the twin tar babies of the federal government and the education establishment. Once again, a worthwhile idea for upgrading American education is being muddled beyond recognition.

The tale begins in the 1980s, when governors like Lamar Alexander and Bill Clinton realized that the effort to reform U.S. schools and boost student achievement would never get far until clear academic standards were set, solid tests put in place, and real accountability mechanisms installed. It's impossible to make a successful journey if you cannot state your destination and have no markers of progress along the way. Nor should one expect much from school-choice policies -- vouchers, charter schools, and the like -- absent clear information for parents on the performance of individual schools.

The federally funded testing program called the National Assessment of Educational Progress (NAEP) had long reported only nationwide averages rather than offering a sense of the differences among the various states and school districts. By 1988, NAEP was changed to permit state scores and comparisons between states; and in order to distance it from the tar babies, an independent governing board was appointed to oversee the testing. For two years, I had the privilege of chairing that board, and we sought to ensure that the assessment content was sound, the tests challenging, and their results reported in relation to standards (how well should kids be doing in 8th-grade science or 4th-grade reading?) instead of simple averages and percentiles.

In 1989, George Bush and the nation's governors met in Charlottesville. There they set ambitious national education "goals" to be attained by century's end. Gov. Clinton, especially, toiled into the wee hours to draft them. Yet one essential ingredient was still missing: standards-based tests that would allow for comparisons between school systems, schools, and schoolchildren. A governor could find out from NAEP how his state was doing but not how Memphis was performing as opposed to Nashville, much less Nashville versus Milwaukee. Nor could the Robinson family see how little Jonah and Jonetta were doing compared with how they ought to be doing, or whether the Franklin School had higher achievement than the Jefferson.

There are hundreds, even thousands, of extant tests, but none of them was -- or is -- quite right. The "standardized" kind that most school systems use doesn't actually contain any standards -- and is prey to the notorious "Lake Wobegon" effect whereby everyone is told he is above average. State testing programs don't allow scores to be compared across jurisdictions. Even NAEP is still barred from reporting on units smaller than whole states. College entrance tests such as the SAT and ACT aren't taken by everyone -- and in any case they are no help in the elementary and middle grades.

U.S. testing in 1997 resembles a faulty model-airplane kit in which the pieces cannot be fitted together. Yet for those who believe in standards- based tests, test-based accountability, and results-based education reform, this is a plane that needs to fly.

In 1991, Bush and then-education secretary Lamar Alexander proposed a new nationwide testing scheme called "American Achievement Tests," but Congress would have none of it. NAEP's governing board suggested that NAEP tests be made available for district and school-level testing, but the education establishment -- and commercial test publishers -- rose up in outrage. The independent board was punished for its hubris in 1994, when Congress stripped away much of its independence.

Re-enter Bill Clinton, first with a wretched program called Goals 2000 that was long on federal control of state and local reform plans but short on testing and accountability, and then, just this January, with a proposal for national testing.

Initially, Clinton's January proposal seemed modest and workable. He offered the states voluntary tests of 4th-grade reading and 8th-grade math -- nothing more -- based on the tests and standards already in use by NAEP at the national level. At little cost to themselves (there would be a federal subsidy, of course), states could use these tests to find out how their school systems, schools, and children were doing. The tests would also be accessible to individual communities, private schools, someday even parents.

Gov. John Engler immediately signed Michigan up, and several other states have since agreed to join. Clinton is barnstorming the land to recruit more. Judging from the many presidential events and speeches in which it now stars, the testing plan has become one of his top education priorities.

Alas, the plan had one surpassing flaw. While it was said to be "based on" NAEP's carefully developed tests and standards, Clinton was not actually going to use those tests and standards, nor did he propose to entrust the new venture to the board I once chaired. Rather, the Department of Education -- without explicit congressional authority -- would use discretionary dollars to launch the test-development process by contracting with a private organization. The independent board would control neither test content nor standards. Instead, the executive branch would.

Immediately, the scholar Diane Ravitch, the Education Leaders Council (a group of state education officials who don't toe the establishment party line) , and I urged the administration to reconsider. We said that national testing had merit -- but that anything so sensitive as these tests must be run at arm's length from the government and education-establishment tar babies. It also seemed that Congress should have something to say about the arrangements for so momentous a shift in American educational federalism.

There has been no effort to address our concerns. The Education Department has been rushing to put the original plan into operation. The independent board has been marginalized. Congress has been stiffed. Critics have been ignored. And to the extent that one can forecast real harm from procedural missteps, damage lies ahead. As often in education-reform efforts, the procedure has been hijacked by the tar babies. The hijacking takes the form of contracts that are already being signed with neither congressional approval nor independent oversight.

The main contract so far is with the Council of Chief State School Officers to develop test specifications. "The chiefs," as they're known in educator- land, are the Washington-based association of state superintendents, and they form one of the establishment's most change-averse crews. The chief of the chiefs, Gordon Ambach, is a former New York state commissioner of education, staunch advocate of a larger federal role in education -- a key backer of Goals 2000, for example -- and a veteran federal grant-getter. He and his group have an ancient and cozy relationship with the Education Department and can be counted on to do its bidding, down to such particulars as Spanish- language math tests and other worrisome wrinkles in the Clinton plan.

A Berkeley, Calif., consulting firm has been engaged to help the chiefs with this project. They've named three new panels: two "content committees" to shape the reading and math exams and a "National Test Panel" to advise on the whole process. (A fourth committee, to furnish "technical" advice, will appear any day now.) Several meetings have been held, and draft specifications are due this month. By September, the Education Department will let contracts for actual test preparation. A huge consortium of private testmakers and publishers is positioned to get the job -- at a pricetag rumored to be near $ 50 million. The plan is to have these tests ready to roll in 1999.

For a new federal project, it's moving extraordinarily fast, apparently so that binding commitments can be made during the current fiscal year and before Congress cranks up to do anything. (House education-committee chairman Bill Goodling, himself a former school superintendent, dislikes tests that allow children and schools to be compared and has tried -- so far without success -- to attach a killer amendment to various bills.)

In truth, it's moving far too quickly for something as momentous as national tests in a country that never had anything of the sort before. But haste isn't the main problem. What's most alarming is that these tests will be creatures of the education establishment and prone to its postmodern curricular faddishness.

Consider these clues:

First, the contractor is in charge of its own advisory committees. Rather than an independent board that sets policies and tells the contractor (and the executive branch) what to do, the chiefs -- in consultation with hierarchs at the Education Department -- have their own handpicked panels, yet remain free to ignore them.

Second, the committees are narrowly based. Whereas NAEP's independent board contains governors, business leaders, local school-board members, and parents, the chiefs' new National Test Panel consists almost entirely of experts, interest-group representatives, and -- surprise -- several chiefs. Nobody knows whether it is bipartisan.

Third, and most worrisome, the two subject-matter committees, for reading and math, include only educators and experts; there's not a "consumer" or employer in the lot. And while each group contains some wellregarded scholars and teachers, each is also tilted toward today's dominant educationist bias: the theory of teaching and learning known as "constructivism." That's the approach in which children are expected to create their own meaning and teachers are supposed to "facilitate" learning instead of directly instructing their pupils in worthwhile knowledge. Constructivism is heavy on " critical thinking" and "problem solving" but light on specific knowledge and basic skills. It doesn't so much care whether you know something as whether you can look it up. It doesn't judge certain information to be important and certain books to be best, but, rather, partakes of fashionable academic relativism; It's friendly to "whole language" reading, wary of systematic phonics; it asks children to guesstimate the solutions to math problems and discounts the importance of correct answers. Tests grounded in this philosophy will be applauded by the ed-school professoriate and the deconstructionists, but they won't test the sorts of things that a governor like Engler -- or, once, Clinton -- really wants the children of his state to know.

Am I reading too much into lists of panelists? Probably not. People I respect in math and reading who have eyeballed these lists use words like " disaster" to describe the tests and standards that they expect to emerge.

If national testing is headed that way, the country would be better off without it. Congress should apply the brakes before a wreck occurs. Then maybe -- just maybe -- let a different driver take a turn at the wheel. If a fully independent version of the board I chaired in the 1980s were put in full charge, the risk of crashing would be reduced. Alternatively, the whole idea might be privatized, turned into a commercial (or philanthropic) testing program that picks up Clinton's basic concept but with no government entanglement or federal funds. We still need a means to compare achievement across state borders. But it's worth doing only if the twin tar babies can be avoided. The one thing indisputably worse than no national tests would be bad national tests.

Next Page