HARVARD PROFESSOR Roland Fryer has made a discovery with the potential to transform public education. To understand it, though, it helps to first hear a story about the conundrum of the coconut cake.

Fryer’s grandmother makes an astounding coconut cake, a magical confection of sweetness and air he’s loved since he was a kid growing up in Florida. Fryer wanted to learn to make the cake himself, but every time he pressed for a recipe, she gave him directions like “use a good amount of sugar, a little flour but not too much, and just a bit of baking powder.”

She wasn’t hiding anything. He’d seen her make it, and the truth is that she works her kitchen by intuition, grabbing what she needs and pouring in what feels right. The secret of grandma’s coconut cake, it seemed, would follow her to the grave.

But Fryer had a thought. A few Thanksgivings ago he watched her while she made the cake, writing down everything she did. Every time she was about to drop an ingredient into the bowl, he stopped her and measured. It drove her a bit crazy, but he ended up with the recipe, and a piece of family history he can share with his own grandkids some day.

By day, the 34-year-old Fryer is a brilliant economist – among the youngest scholars ever to earn tenure at Harvard – who studies schools. And education, he realized, has its own coconut-cake problem. There are public schools that are performing near miracles in deeply troubled urban districts, but nobody, not even those who run these schools, can say for sure what makes them work. Everyone has their theory – a longer school day, a little more discipline – but nobody knows the actual recipe.

So Fryer went to New York City and measured the ingredients. Working with Harvard’s Will Dobbie and a team at the Harvard EdLabs, Fryer collected an unprecedented amount of information from a diverse group of 35 charter schools – everything from test scores and spending per pupil to educational philosophies and videotapes of classroom instruction. Then, using rigorous statistical techniques, he compared differences in student achievement with all the other variables, extracting five principles that the star charters all share.

These principles are themselves a major stop forward: what is on the list, and particularly what is not on the list, is quite surprising. But here the story takes an amazing turn. Fryer worked with the superintendent of Houston’s public schools, Terry Grier, to apply the five principles to a group of failing schools.

The results have been positive. In just one year, kids in one of their schools went from 40 percent proficient in math on a standardized test to 85 percent proficient; high school seniors were 50 percent more likely to enroll in a four year college. Overall, reading scores have moved up only modestly, but math scores have climbed dramatically and the experiment has only just entered its second year. And all this has been accomplished in ordinary public schools, without converting a single one into a charter school; no students were kicked out.

So what is this magic recipe? It turns out to be remarkably straightforward: Give frequent feedback to teachers, use loads of data on individual students to guide their instruction, employ heavy tutoring, increase instructional time, and maintain very high expectations. All of the very best performing charter schools in his New York sample did these things aggressively, according to a paper Fryer released last month, and this is what turned around the Houston schools.

The other things you hear about improving schools – such as smaller classes and spending per pupil – do not appear to be important. This seems to defy logic, but not for Seth Andrew, founder of the Democracy Prep charter schools in Harlem. He uses larger classes because this means fewer teachers lecturing at any given time, and the free ones can do tutoring, professional development, and other essential activities. He spends less money per pupil than at other schools because the administration is lean and careful. Last year, his class of black and Hispanic 10th graders outperformed students from wealthy Westchester County.

The next steps are clear. Massachusetts, and every state that allows charter schools, should require all their schools to submit to the kind of analysis that Fryer does. With data from thousands of schools, we will have a very precise picture of what works. Meanwhile, as Fryer has shown in Houston, there is no reason we can’t start turning around schools now.

Excellent urban public schools are an artisanal product, the handiwork of a few geniuses who, like Fryer’s grandmother, arrived at their recipes through intuition. It’s time to move to mass production. There are a lot of desperate kids out there, hungry for the real thing.