Raising the education bar
What percentage of Georgia’s fourth-graders are good readers? It seems to depend on whom you ask. The state will tell you that 85% met or exceeded the proficiency benchmark on its 2007 test. On the other hand, that year only 28% scored high enough to be considered proficient on the National Assessment of Educational Progress, or NAEP, an exam administered by the U.S. Department of Education that is usually regarded as the gold standard. The big difference results from where the two tests set their proficiency bars. Georgia sets its bar pretty low -- so low that barely literate students can score high enough to be deemed proficient. On the NAEP, a student labeled “proficient” by Georgia could fail to score above “basic.”
Unfortunately, five states have even lower standards than Georgia’s, and eight others are at about the same level. For its part, in 2007, 52% of California’s students performed at or above the proficiency benchmark on the state’s reading exam, while only 23% met or exceeded the NAEP’s proficiency standard.
Worse, standards are declining. A recent federal study noted that 15 states lowered at least one of their proficiency standards in math and reading between 2005 and 2007.
States’ low standards have spurred a bipartisan campaign to create worthwhile national ones. But the road to national standards would be extremely tough to navigate politically. A more feasible approach would be to give all states an incentive to set objectively high standards themselves. The looming reauthorization of the federal No Child Left Behind Act gives us an opportunity to do it.
The perverse incentives of the act clearly have something to do with the downward movement of state standards. The law punishes a school when too few of its students meet math and reading proficiency targets each year. But the law has a gaping loophole: States get to define proficiency. A state can thus meet the law’s targets by defining proficiency down; toughening its standards, by contrast, handicaps its ability to meet the federal requirements.
Of course, low standards have their own appeal. The lower the standard, the more students surpass it. State governments love to tell constituents that students are doing great on standardized exams; the public usually just assumes that the criteria used on those exams are meaningful.
States can get away with setting low standards in part because proficiency remains an inexact notion. In determining the difficulty of exam questions, states typically rely on expert panels. What constitutes an expert varies from state to state -- and the experts only make suggestions, which the state may ignore. The percentage of students who answered the question correctly also gets taken into account. But it’s clear that the process has an unavoidably subjective dimension.
Given such confusion, the case for high, uniform, enforceable national standards seems strong.
But national standards run up against serious objections. Historically, schooling has been a state matter. In 2002, some conservatives argued against enacting No Child Left Behind because they saw it as an unwarranted federal intrusion into state business. A congressionally set national standard would only expand Washington’s role.
Even more troubling is that we could wind up with a single low standard. Proponents of national benchmarks seem to think that they’ll be the ones writing them. Bureaucrats in the U.S. Department of Education will have other ideas. So will members of Congress from lower-achieving states, which won’t want to be embarrassed by a national standard that their students can’t reach. Because any system for setting a common standard, either by federal mandate or voluntary state agreement, depends on the cooperation of lousy performers -- like Georgia -- it’s hard to see how a demanding national standard would survive the political process.
We could make better progress toward an effective testing regime if we changed our goal from uniform national standards to high state standards, which two simple amendments to No Child Left Behind could help bring about.
First, remove the law’s disincentives for states to adopt higher standards. Recall that the law holds schools accountable based solely on the percentage of students who score above a certain proficiency benchmark. Instead of focusing just on who counts as “proficient,” the law’s measure of school quality should also include the gains that students make from year to year on their state tests. This would stop rewarding states that push down their proficiency standards. (To work, a revised system would have to ensure that tests were scaled properly.)
Such a “value added” measure would also make more sense, because a truly effective school is one that makes a meaningful difference in how much its students learn.
An entirely value-added system isn’t desirable, however, because it’s important that we continue to provide schools with goals for learning. The best system would combine value-added measures with a minimum proficiency standard.
But even with the law’s perverse incentives removed, some states will still prefer lower standards because they inflate records of achievement. We need to give states an incentive to set higher standards.
A second tweak to No Child Left Behind is how to do it. States could still develop their own tests in math and reading and set their own benchmarks. Every few years, though, the federal government would administer these tests to a small but nationally representative sample of students. These scores would provide a uniform and objective measure that could be used to compare the relative difficulty of standards across states. For example, a test that identified 80% of the nation’s students as proficient would have a lower proficiency standard than one that labeled 40% as proficient.
With an objective measure of each state’s standards, a revised law could then link some portion of a state’s federal per-pupil funding to the standards’ difficulty, relative to those of other states. Under such a system, states would have an incentive to raise standards, but not to set them unreasonably high, because some of their funding would still depend on the percentage of their students meeting the proficiency benchmark.
This system would still allow the definition of proficiency to vary. But it would at least encourage the proficiency benchmark to increase over time. In this system, the answer to the question, “What should students know?” would always be “More!”
Marcus Winters is a senior fellow of the Manhattan Institute. This is adapted from the forthcoming issue of City Journal.