WASHINGTON >> Disadvantaged students are much more likely than others to be engaged in remote schooling during the coronavirus pandemic, increasing the risk that less effective instruction will widen the achievement gap, according to the first comprehensive analysis of attendance patterns.

Using cellphone data to track movement to more than 100,000 schools, researchers at Columbia University found that closed classrooms were disproportionately composed of nonwhite students, as well as students with low math scores or limited English proficiency or who are poor enough to qualify for free meals.

About 58% of nonwhite students attend schools that rely heavily on remote learning, compared to 36% of white students. Remote learning is widely considered less successful than traditional classrooms, especially for younger children.

“Given the sheer magnitude of the students affected, this does not bode well,” said Zachary Parolin, the study’s lead author. “Inequality in learning outcomes is only more likely to grow.”

Other experts have warned that disadvantaged students often lack the support that remote learning requires, such as computer access, quiet study space and help from parents or tutors. The Columbia study shows how the students least equipped for virtual instruction are those most likely to have encountered it this year.

Consider the experience of Shereese Rhodes, a single mother in Kent, Washington, whose fifth grader, Mya Janae, has not returned to the classroom since the coronavirus closed her school in March. While the school has worked hard to improve its online learning, Rhodes said, problems have abounded.

A school-issued computer would not recharge, and the replacement’s internet connection was weak, causing images to freeze and Mya Janae to sound like a robot. Unlike the last year, when closures followed months in class, Mya Janae has never met the teachers on her screen. They get angry if she turns off the camera but her older brother objects to appearing in the background.

The biggest challenges involve Mya Janae’s learning disabilities. She had a long delay in learning to speak and has impaired hearing. “She’s not designed for school like this,” Rhodes said. “There’s not time for her to ask questions. She has breakdowns and just cries about little things.”

Mya Janae agreed that “online is bad, because for me it’s hard to pay attention,” especially with her cat, Queen, nearby. “I’m not learning anything. I feel like if I keep learning like this, I’m gonna flunk.”

Worried about lingering harm, Rhodes squeezed her budget to hire a tutor and reluctantly told her daughter, “Santa’s not visiting you this year because I want to make sure you’re set up academically.”

Parolin and his co-author, Emma Lee, analyzed anonymous data from 40 million cellphones. Since some schools let parents choose between live and remote classes, the researchers labeled a school “mostly closed” if daily arrivals in October fell by half or more from the same time the previous year. Schools where traffic declined by 25% or less were deemed “mostly open.”

The starkest differences involved race. While just 36% of white students were enrolled at closed schools, the share was 51% among Black students, 60% among Latino students and 64% among Asian students.

Schools with large shares of non-English speaking students and those with the lowest math scores were also more likely to be closed.

The pattern is largely driven by geography: minority students disproportionately live in big cities and in Democratic-leaning states, which have emphasized social distancing to fight coronavirus. Two majority-minority jurisdictions — Washington, D.C., and California — had the country’s highest rates of school closures, while two heavily white states — Wyoming and Indiana — had the lowest rates.

The pattern may also be driven by family choice: surveys have shown families of color are less likely to trust the precautions taken by school systems and neighbors. “Families of color have good reason to feel they’re at risk” given their higher rate of infection, said Robin Lake, director of the Center on Reinventing Public Education, a research group.

Among them is Rhodes. Despite her dislike of online learning, she said she would keep Mya Janae home even if her school reopened, “given where the COVID numbers are today.”

A September study by the Brookings Institution warned that school closures would disproportionately affect students of color. Richard V. Reeves, a co-author of the study with Ember Smith, said the new data, which is more comprehensive, deepened his fear of rising inequality. He urged officials to consider reopening schools to protect poor students from enduring harm.

“It’s easy to forget how perilous the process of staying on track is for many of these kids,” he said. “The consequences of short-term derailments can be very large indeed, and the students who are most affected were already most behind.”

For many families, the switch to online learning has coincided with deepening hardships at home. Job losses have disproportionately hit low-paid and minority workers, poverty rates appear to be rising, and child hunger is abnormally high.

While it remains unclear how much school closures will harm disadvantaged students, most experts are pessimistic.

NWEA, a nonprofit research group, warned in May that the spring school closures could cost students one-third of their expected annual progress in reading and half of their expected progress in math. A subsequent analysis of fall test scores showed better results — no falloff in reading and more modest declines in math — but many disadvantaged students did not take the test, likely skewing the results.

Data from Zearn, an online math program used by some schools, shows widening performance gaps, with progress among low-income students falling by 14% since January, even as it rose by 13% among high-income students. A recent study of Dutch exams found the average student made “little to no progress” during an eight-week shutdown last spring, with disadvantaged students suffering the greatest learning loss.

“It’s nearly certain that remote learning will widen the achievement gap,” Lake said. “It’s been a complete disaster for low-income students.”

Among those affected are Dehlia Winbush of Kent, Washington, and her 10-year-old daughter, Nadira, who suffers from a behavioral disorder that causes swings between depression and aggression.

The move to remote learning last spring “was extremely horrible,” Winbush said. “It was constantly a fight to get her to log on, even though it was only for an hour.” The school-issued computer malfunctioned, and Winbush, who is visually impaired, was not able to read it well enough to help Nadira with lessons.

“I personally don’t think she learned anything,” she said.

The new school year, she said, brought a longer school day and “a really great teacher.” But the isolation deepened Nadira’s depression and led to a recent hospitalization. Winbush took time off from her job in a warehouse to be at her daughter’s side, but her absences caused her to lose the job, adding financial concerns to medical woes.

While Nadira’s screen flashes with interesting lessons — the rise of cities, defense mechanisms in animals — she is missing the social and emotional development that comes from being in a classroom.

“I’m trying to keep her interest in school intact, but sometimes the positivity is no longer there,” Winbush said. “We’re still trying to get the fight back.”

© 2020 The New York Times Company