This article is from the archive of our partner National Journal

Fifteen years ago, presidential candidate George W. Bush stood before the NAACP's annual convention, and talked about advancing racial harmony and economic opportunity through education.

"Under my vision, all students must be measured. We must test to know," Bush said. He proposed giving low-performing schools—"those schools that won't teach and won't change"—three years to improve before helping parents move their children elsewhere. "No child should be left behind in America," he said.

The education law President Bush signed, the No Child Left Behind Act (NCLB), was based on a straight-forward theory: Give schools data and pressure them to improve, and they'll better educate children. Simple. But the law didn't always work that way in practice.

Now Congress is revising the law and grappling with its complicated legacy. A Senate bill would likely preserve annual, statewide testing while giving states more flexibility to design their own accountability systems.

NCLB essentially attached strings to federal education funding under the Elementary and Secondary Education Act. It required states to test students annually from third through eighth grades and once in high school, and to track test scores by students' race and ethnicity, low-income status, disability status, and status as English-language learners.

If students in any subgroup persistently failed to achieve certain scores on state tests, schools could be shut down. The law set a goal of ensuring all students in the country were reading and performing math calculations at grade level by 2014—a goal that proved so impossible to meet, the Obama administration has since granted most states a waiver from meeting NCLB's targets.

When Bush addressed the NAACP, he and his audience knew that African-American, Hispanic, and low-income students were more likely to struggle academically than white and affluent students. Since at least the 1960s, national standardized tests have revealed an achievement gap.

NCLB revealed the extent of those gaps in every school in the country—and pushed districts to hone in on low-scoring students. It fueled ambitious reform efforts in cities like New York and Washington, D.C. The disaggregated data "became a vehicle for us to start to explore equity issues, and look at basic principles of accountability that we wanted to build out," says Joel Klein, former chancellor of the New York City Education Department.

Reformers like Klein shut down schools, reorganized others, and expanded new models like charter schools. The NCLB era coincided with the rise of public-charter networks like the Knowledge is Power Program (KIPP), which proved that it was possible to raise the test scores of low-income and minority students.

But after more than a decade of shuttered schools and sanctions, the achievement gap remains as wide as ever, even in cities like New York. Nationally, scores on the low-stakes National Assessment of Educational Progress (NAEP) have risen, but researchers say the rise is just part of a long-term trend. A 2011 National Academy of Sciences study found that NCLB raised elementary-school math scores, but not much else.

Sandy Kress, a former Bush adviser, says he's troubled by how few schools responded to the law the way its architects intended: by reaching for proven solutions, like better teaching strategies. In successful accountability systems, he says, "people are highly motivated to go to what works."

That didn't happen in the Mississippi Delta, where Renee Moore taught high school until 2006. In her area, most districts serve almost entirely low-income, African-American students. "All of the schools around, we have nothing but poverty," Moore says. The area has a chronic teaching shortage and schools get less per pupil funding than the national average.

By requiring all schools and all students to meet the same standards, NCLB set schools like Moore's up for failure. There's no way to reallocate resources when resources are tight to begin with and almost all students earn low scores. Under pressure to raise test scores dramatically and quickly, Moore says, local elementary and middle schools started drilling students for state tests during social-studies and science classes.

The law also generated data that, while useful to policy wonks, education researchers, and civil rights activists, it wasn't particularly useful to teachers. "The test data that we were getting, it came late," says Moore. "It never came until after the school year was over." The state tests didn't generate information specific enough to help her focus her instruction.

What NCLB has demonstrated, 15 years in, is something policymakers already knew—that standardized test scores are strongly correlated with a student's family income. To education scholar Diane Ravitch, the achievement gap in test scores doesn't prove that there are bad teachers, or bad schools—it just proves that poverty matters.

It is possible to raise test scores, using a different theory of action. "If we only asked, are they [students] making annual academic growth, if that was the only thing we talked about all the time—that would end up skewing your organization in ways, and schools in ways, that aren't healthy," says Richard Barth, CEO of the KIPP Foundation.

KIPP schools assess students on what they're learning constantly. They also measure student satisfaction, parent satisfaction, and teacher satisfaction. Students occasionally take tests that show how they compare with their peers nationally.

Teachers and school leaders talk about how to learn from the data all the time. "It's not about, you know, accountability and punitive measures," Barth says. It's about continuous improvement.

Next America's Education coverage is made possible in part by a grant from the New Venture Fund.

This article is from the archive of our partner National Journal.

This article is part of our Next America: Higher Education project, which is supported by grants from the Bill & Melinda Gates Foundation and Lumina Foundation.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.