You can count on one familiar refrain in the 2020 presidential campaign: Fix the schools. Faith in education is one of the nation’s bedrock values. Better schools would (we think) narrow economic inequalities and help people reach their personal potential. Promises to revitalize schools are inevitable.

There’s a magical quality to these proposals. The message seems to be that, if we can find the right combination of ideas, we can unleash education’s uplifting power. Be skeptical.

Already, at least two Democratic presidential candidates are pitching major educational proposals. Sen. Kamala D. Harris (Calif.) would give most teachers a huge pay raise, reportedly averaging about $13,500. Teachers, it’s argued, are underpaid. This makes good ones hard to recruit and retain. Meanwhile, ex-San Antonio mayor Julián Castro advocates universal pre-K classes to prepare children for school.

Both ideas sound sensible. But aside from the sizable costs, history suggests that creating gains in achievement and academic skills for the poor is extraordinarily difficult.

That’s the finding of a major new study. It reviewed test scores for Americans born between 1954 and 2001 to see how much the achievement gap had closed between students with low and high socioeconomic status.

The startling result: hardly at all.

“The achievement gap fails to close,” headlined an article in Education Next. “Half century of testing shows [a] persistent divide between haves and have-nots.”

The explanation is not that public policy wasn’t trying. The discouraging conclusion occurred despite the federal government’s decision to provide extra funding for poor schools under Title I of the Education and Secondary Education Act of 1965. Previously, public schools were funded mainly by localities and states. Corrected for inflation, overall spending per student nearly quadrupled from 1960 to 2015.

Still, there was little effect on the achievement gap. The study was conducted by Eric A. Hanushek and Laura M. Talpey of Stanford University, Paul E. Peterson of Harvard University and Ludger Woessmann of the University of Munich. Tests were given at two ages, 14 and 17. Here are highlights:

●The central problem seems to occur in high schools. Tests administered at age 14 actually showed improving student performance. But most of the gains reversed by age 17, just when students were preparing for college or work.

●During this roughly half-century, there was no general rise in achievement, which would have been a partial victory — almost everyone’s achievement level would have increased even if the gap between top and bottom hadn’t closed.

●The population’s changing ethnic and racial composition doesn’t explain the stubborn achievement gap. (In 1980, the population of children ages 5 to 17 was 74.6 percent white, 14.5 percent black, 8.5 percent Hispanic and 2.5 percent other. By 2011, the corresponding figures were 54.2 percent white, 14 percent black, 22.8 percent Hispanic and 8.9 percent other.) Separately, the study found similar trends among whites, suggesting that race or ethnicity aren’t major causes.

●The study did not find, contrary to at least one other major study, that the achievement gap has actually worsened over the past half-century. If this conclusion holds up, it would qualify as the study’s main bit of good news.

Repeatedly, the study’s authors express frustration that they can’t explain what happens in high school to undo previous gains in achievement. They dismiss “senioritis,” the reputed tendency of students to slacken in their studies, as a major cause. They speculate, though they admit that they don’t know, that teaching in high school is harder than at lower levels.

“The high school is a broken institution,” Peterson said in an interview. “We need to create more learning opportunities for kids in high school.”

Broadly speaking, the study vindicates the results of earlier research conducted by sociologist James Coleman (usually called the “Coleman report”) in 1966. As part of President Lyndon B. Johnson’s War on Poverty, Coleman examined what factors promoted educational success. He found parental education, income and race to be strongly connected to student achievement, while per-pupil expenditures and class size were much less so.

The upshot is that schools are being asked to do for their students what families usually do. This is a tall order that is probably beyond the capability of most schools.

As a society, we should keep trying. But we should not ignore history. The national strategy of controlling the country’s schools — through subsidies and regulatory requirements — has prevailed for half a century. It has failed. The federal government should exit the business of overseeing K-12 education. Federal aid would halt, and the financial loss would be offset by having the national government assume all the states’ Medicaid costs.

We should let states and localities see whether they can make schools work better. The grandiose fix-it national plans are mostly exercises in political marketing. We need solutions, not slogans.

Read more from Robert Samuelson’s archive.