One of the goals U.S. Education Secretary Arne Duncan set when he launched the $3.5 billion School Improvement Grant (SIG) program in 2009 was to turn 1,000 schools around annually for five years. “We could really move the needle, lift the bottom and change the lives of tens of millions of underserved children,” he said.

I like Duncan and much of what he and the Obama administration have done for schools, but that goal is a harmful fantasy.

The overlooked truths about fixing schools are vividly revealed in an Education Writers Association research brief, “What Studies Say About School Turnarounds,” by Andrew Brownstein, a freelance journalist who reports on federal education policy. (Turnaround schools are those whose low achievement rates have been significantly improved by a change in operations.)

Brownstein said “successful turnarounds are extremely rare.” Veterans of education reform efforts “might be forgiven for thinking of turnarounds as the unicorns of federal education policy,” he said.

We don’t know how many turnarounds have occurred since SIG began, but it is far, far less than a thousand a year, and what first look like turnarounds may prove to be disappointments. With the meager studies available, Brownstein said all he could do was see whether the most popular turnaround methods — such as replacing the principal — have proved effective in the past. The research indicates that new leadership practices, not new leaders, are the crucial factor, he said, and assessing their effectiveness is difficult.

Some of the turnarounds cited in the studies Brownstein examined were only a couple of years old. Such unicorns often disappear into the mist. Tom Loveless of the Brookings Institution had concluded that “examples of large-scale, system-wide turnarounds are nonexistent.” So why would otherwise intelligent education leaders such as Duncan put so much faith in them?

I think the problem is representative democracy. Your schools can’t get money for new programs unless elected officeholders agree. Politicians demand optimistic goals, or they can’t sell the program to voters and to enough legislators to gain a majority. Rep. George Miller (D-Calif.), who helped forge the bipartisan majority that passed the No Child Left Behind Act in 2001, said they had to set a goal of 100 percent proficiency in reading and math — an impossible target. If they settled for a more reasonable number, such as 70 percent, opponents would say they were leaving 30 percent of kids behind.

By predicting 1,000 turnarounds a year, Duncan secured funds he knew would help teachers and students, even if they never reached his unlikely goal. Media fact checkers such as my colleague Glenn Kessler don’t usually critique hopeful guesses because miracles, like the collapse of the Soviet Union or my birdie on a par-5 last week, sometimes happen. I have criticized D.C. Schools Chancellor Kaya Henderson’s plan to raise achievement in the 40 lowest-performing schools by 40 percentile points in six years, but she needs that loony goal to get some of that federal money for her schools. Everyone has been forced to play that game for many decades.

There are better ways to improve schools than turnarounds. Many charter and experimental public schools have shown it makes more sense to create new schools with no old bad practices to expunge. Students and teachers can start fresh. A 2006 NewSchools Venture Fund study cited by reform expert Andy Smarick found only four of 36 organizations with strong records in improving school achievement had expressed interest in restructuring existing schools.

KIPP, a charter school network, has produced scores of schools that outperform others with similarly disadvantaged students trying the turnaround approach. Some educators have criticized KIPP for never turning around an existing school, which to me is like criticizing quarterback Peyton Manning for almost never using a huddle in his successful speeded-up offense. If new ways work better, why insist on the old?