Summit Learning, the Zuckerberg-backed platform, says 10% of schools quit using it each year. The real figure is higher.

When nearly 100 students walked out of their Brooklyn high school in protest last year, saying they were spending too much of their days in front of a computer, the story took off.

The students were complaining about their school’s use of Summit Learning, a curriculum and online learning system backed by Facebook founder Mark Zuckerberg. But the organizations behind Summit pushed back, saying the issues raised by the Brooklyn students weren’t representative of what was happening at the nearly 400 schools using the program.

One piece of evidence they offered: just 10% of schools quit using the platform each year, a number that ended up in multiple news stories.

New data obtained by Chalkbeat — from Summit itself, in response to a public records request — shows that figure is misleading. Since the platform was made available, 18% of schools using it in a given year had quit using it a year later.

Asked about the discrepancy, a Summit spokesperson explained that its 10% figure comes from averaging the dropoff rates for each of the first three years. The number of schools adopting the platform was 19 in the first year and 338 in the third year, so Summit’s approach is skewed heavily in favor of the first year’s low attrition number.

Looking just at schools that signed on to the platform last school year, a quarter of them are no longer using Summit this year.

“We are extremely proud that the vast majority of schools that launch Summit Learning continue to stay in the program,” spokesperson Catherine Madden said in a statement. A spokesperson for the Chan Zuckerberg Initiative, which has also cited the 10% number, declined to comment.

The numbers offer a starting point for understanding the scope of pushback to Summit Learning — which has made headlines in Brooklyn, Connecticut, Kansas, and elsewhere.

Still, the data provide something of a Rorschach test for one’s views on Summit and the broader movement to “personalize” learning to better serve students. A solid majority of schools opting to use the program are re-upping the next year, the data show. On the other hand, if the current attrition rate holds, half of the schools now using it may stop within a few years.

John Pane, who has researched personalized learning at RAND, said an 18% annual dropoff rate would seem high if all schools planned to use Summit on a permanent basis, but would be more reasonable if many were just piloting the program. At least some schools in the program have described their adoption of Summit as a pilot, and Summit itself notes that many schools try it out in a single grade.

“This kind of thing, the ‘personalized learning’ model that Summit has, is a really big change to how schools operate,” he said. “When you’re making a big change like that, there’s a lot of work to do to get everyone on the same page, including parents.”

The Summit platform emerged from a network of charter schools in California and Washington state. The model includes a technology platform that guides students through lessons but also a structure for the school day and year that includes weeks set aside for investigations into specific topics and time to work with a dedicated mentor.

It’s grown with help from the Chan Zuckerberg Initiative. The New York Times recently reported that CZI has given $99 million to Summit since 2016 — a number that doesn’t include the team of engineers CZI has paid to work on the platform. (CZI is also a funder of Chalkbeat.)

For the last several years, Summit has made its technology available to schools across the country for free. As more schools adopted it, scrutiny has ramped up, most prominently in the Times story highlighting complaints from parents and students in two Kansas school districts. Again, Summit and its supporters said that their experience was not representative.

The list of schools using the program each year sheds new light on this issue.

Of the first 19 schools to adopt the platform in 2015, all but two — nearly 90% — are using it this year. But as Summit has rapidly spread, attrition has also jumped.

Another way of looking at the numbers: about 25% of schools that have adopted Summit at any point before this school year are no longer using it.

Madden, the Summit spokesperson, told Chalkbeat last month that the the precise share of schools that left the program each year was 11%, or nearly a 90% re-up rate. Madden declined to provide more details, but Chalkbeat obtained additional information through a public records request to Summit’s charter network.

From there, Chalkbeat calculated that 18% annual dropoff rate. Summit’s number diverges from Chalkbeat’s because it weights each year equally, regardless of how many schools participated. Summit also didn’t count four schools that dropped the program shortly after adopting it.

Summit has recently spun off efforts to offer its platform to other schools into a separate nonprofit, T.L.P. Education, which would not be subject to public records laws.

So, is an 18% annual dropoff rate high, or in line with what one would expect for an effort to overhaul how classrooms work? That’s hard to say.

As one point of comparison, Education Week reported in 2017 that a quarter of schools that had ever used the math personalized learning program Teach to One eventually dropped it. Unlike Teach to One, though, Summit is offered free to schools, which might encourage some to try it on a trial basis.

“Maybe the district feels like it’s less of a risk, or may have less commitment, if they’re not making a financial purchase,” said Pane.

The Summit spokesperson said that some schools drop the platform because of changes in school leadership or the teaching staff or because their school district adopts a common curriculum and requires schools to follow it.

Some schools have done so in response to criticism from parents and students. A few have tried making compromises: After controversy, one Pennsylvania district allowed sixth-graders to opt into the program before announcing last month it was canceling Summit altogether.

“At this point there is declining interest, we couldn’t sustain it with the staff, and our other class sizes are rising,” said Superintendent Michael Vuckovich. “And we weren’t offering a program with the fidelity it should have had.” (This district is counted in Chalkbeat’s data as continuing to use Summit because it only canceled the program last month.)

But most schools are sticking with Summit.

“In our own schools, we’ve seen real progress since implementing this approach,” wrote Gordon Mohn, superintendent of one of the Kansas’ districts featured in the New York Times article. “As students become more comfortable with the changes, we see greater student engagement and increased interaction between the teacher and student regarding the learning process.”

Meanwhile, there remains virtually no external research on the effectiveness of Summit’s model. Chalkbeat reported earlier this year that Summit had paid Harvard researchers to design a study of its model, but ultimately decided not to move forward with being evaluated.

Asked recently by Chalkbeat about research on the model, Sandra Liu Huang, head of education at the Chan Zuckerberg Initiative, didn’t offer specifics.

“We want to be able to figure out how to do a broader scale research study of that, but also understanding … it is a living, evolving, growing program that looks a little bit different in every context,” Huang said during a question and answer session at the Education Writers Association conference earlier this month. “We’re trying to figure out the best ways to do that.”