As we head toward the start of the academic year, schools around the country are grappling with how to educate their students while keeping everybody safe. Some districts will have kids back in class full-time, while others will only teach them via a screen. But an estimated 20 to 30 percent of the nation’s school districts—including New York City, which alone serves more than one million students—are planning to implement a “hybrid” model, where groups of kids attend in person on alternating, part-time schedules. U.S. Governors, school chancellors, and state epidemiologists have touted this approach as the Goldilocks solution: the number of children in classrooms is kept low enough for proper social distancing, while students still receive some amount of essential in-person learning. It appears to be the perfect compromise.

But this widely held assumption may be grossly incorrect.

“The hybrid model is probably among the worst that we could be putting forward, if our goal is to stop the virus getting into schools,” says William Hanage, an epidemiologist at Harvard’s T.H. Chan School of Public Health. “I don’t see how, in the end, this helps teachers,” says Jennifer Nuzzo, an epidemiologist at Johns Hopkins Bloomberg School of Public Health. “I don’t fully get the hybrid model.”

Their argument is simple: If you want to limit children and teachers’ exposure to infection, it’s better to have students spend their time within a consistent group of peers.

In a hybrid model, when students are kept out of school for multiple days each week, or every other week, a sizable percentage of them are likely to intermingle with other children and adults. This is especially so for younger kids with working parents, as the kids may need to be in daycare, exposing them to another set of social contacts, and all their possible infections. Meanwhile, older kids and adolescents will be inclined to hang out with their peers on their copious “off” days. (In many districts, “remote learning” plans include just a short amount of live-streamed teaching every day, leaving many hours to fill in other ways.) The hybrid model, Nuzzo says, “only works if students stay home, alone, during all of that time they are out of school.” This is a strangely unrealistic assumption by policy makers.

All those extra interactions are more likely to increase transmission risk than to reduce it, according to Dimitri Christakis, a pediatrician and epidemiologist at the University of Washington School of Medicine, and the editor in chief of JAMA Pediatrics. “There’s a real chance a hybrid model could advance the spread of the virus.” In his view, it would be preferable to have 30 kids in a classroom even if there weren’t sufficient space for six-foot social distancing, than to switch off groups of half that size. In the latter scenario, each of those students would likely be exposed to more people overall. Their teachers, too, would be at higher risk—since instead of teaching one cohort every day, they’d be in charge of two.

Martin Kulldorff, a biostatistician at Harvard Medical School, frames it as a simple matter of arithmetic: “With full-time schooling, children primarily are in just two places and with two groups of people, at school and at home. With a hybrid model, many young children must also be in a third place with additional people, such as a grandparent, uncle, neighbor, nanny or day-care provider.” By increasing everyone’s exposure from two places or groups to three, he says, the hybrid model is “the worst of both worlds.” He suggests a “hybrid-teacher” model, instead: The children stay in school full-time, while the most vulnerable teachers work permanently off-site, helping their colleagues to grade exams, prepare course material, or provide online tutoring for children who must be at home themselves.

While a number of countries in Europe implemented a hybrid strategy this past spring, none of the experts interviewed for this article were aware of any studies on its effects on viral transmission. Beyond the potential for greater spread of infection, the educational benefits of the hybrid model may be somewhat marginal. Meira Levinson, an educational expert and ethicist told me that some students might find real value in even intermittent opportunities to learn in-person—from dissecting frogs, for example, or the occasional team-building exercise. But many others will be adversely affected by an inconsistent schedule. She also pointed out that hybrid models do little to ameliorate the child-care crisis that results from having children on remote-learning schedules.