Banning Phones for 3.3 Million Students: Small Academic Gains, Larger Lingering Questions

Education PolicySchool Phone BansEmpirical ResearchNBER

Sources:CEPR + NBER working paper 35132

3.3 Million Students for 0.01 Standard Deviations

On November 17, 2026, Harvard’s Center for Education Policy Research (CEPR), in partnership with Dartmouth College and assessment provider NWEA, released the inaugural paper of its “States Leading States” initiative. The study covered a massive sample: 3.3 million students across 4,600 schools, spanning from fourth grade through tenth grade and tracking math and reading scores across the past five academic years. The core conclusion boiled down to one sentence: strictly enforced cell phone bans do lift academic test scores, and the magnitude of that increase is 0.01 standard deviations.

A standard deviation measures the spread of student achievement, with the vast majority of students scoring within two standard deviations above or below the mean. At 0.01 standard deviations, the gain represents one-hundredth of that overall distribution—translated by the researchers themselves into roughly one to two additional weeks of instruction per student per year. A sample of 3.3 million students is large enough to surface any genuine effect; because the measured outcome is this small, the underlying effect itself is just that modest.

The research team provided historical context: starting around 2013, American student achievement entered what the paper termed a “slow leak,” shedding the equivalent of two to three weeks of learning each year. By the eighth grade, the cumulative loss approached an entire school year of learning. The one to two weeks recovered by phone restrictions plugs roughly half of that annual leak. That constitutes the complete definition of “effective” here.

Strict enforcement in practice typically means a magnetic lock pouch: students place their phones into a pouch upon entering school, it snaps shut with a magnetic clasp, and it can only be unlocked when leaving at the end of the day. The vast majority of these 4,600 schools relied on Yondr, the dominant supplier in the market. A simple fabric pouch with a magnetic clasp has become one of the largest educational interventions in recent American history, with over 40 states already introducing legislation to restrict phones in schools.

A Pennsylvania student demonstrates the Yondr lockable pouch

Figure: A student in Pennsylvania’s Wyoming Area School District demonstrates how the Yondr pouch works. School phone bans are often enforced through this pouch. Source: NPR / Jason Ardan, Citizens’ Voice via Getty Images

Scores Have Suffered a “Slow Leak” of 2-3 Weeks Per Year Since 2013

Thomas Kane, the lead author of the CEPR paper, previously leaned toward a different explanation: he suspected the decline in American test scores was primarily caused by weakened test accountability—when schools stopped prioritizing standardized testing, student effort naturally waned. Kane changed his mind after looking internationally: over the exact same period, multiple Western European nations and Canada observed comparable downward slides, particularly in reading.

Examination systems and accountability pressures varied widely across countries, yet the declines occurred in lockstep. Once uniquely American factors were ruled out, the remaining explanations pointed toward technologies that expanded globally at the same time. The smartphone matches that timeline perfectly: nearly a decade ago, smartphone ownership among American teenagers approached saturation; by January 2027, the iPhone will mark its 20th anniversary.

A direct causal chain proving that smartphones single-handedly drove the decline has not been conclusively established. Nonetheless, it remains one of the explanations with the most concentrated empirical evidence, serving as the common starting point for both new studies.

The Same 4,600 Schools, Two Conflicting Answers

In NBER Working Paper 35132, released in April 2026 and revised in September 2026, authors Allcott, Baron, Dee, Duckworth, Gentzkow, and Jacob studied that exact same cohort of 4,600 schools. Their dataset was broader: Yondr pouch sales records, school geolocations, teacher surveys, and district administrative records.

Their findings presented a nuanced picture: the overall average academic effect was “consistently close to zero.” Broken down by grade band, high schools experienced modest positive effects (most visible in math), while middle schools saw small negative effects. Attendance remained unchanged. Self-reported classroom attention and perceived online bullying showed no measurable improvement. The chronological trajectory added further complexity: during the initial year of rollout, school disciplinary infractions actually ticked upward, and student subjective well-being dipped. Beginning in the second year, the disciplinary spike subsided, well-being rebounded, and subsequent years registered positive trends.

This divergence produced a striking contrast between two leading scholars. Stanford education economist Thomas Dee, a co-author of the NBER paper, stated: “Early findings on school phone bans are largely disappointing and raise unresolved questions.” Harvard’s Kane, heading the CEPR study, countered: “Our answer is yes, phone bans do have a positive effect on growth. But it’s not a panacea.” The same country, the same schools, and two veteran education economists: one characterizes the results as disappointing, while the other affirms positive growth.

Parents and students learn about the pouch policy

Figure: Parents and students in the Lackawanna School District in New York watch an informational video on the pouch policy, August 19, 2024. Source: The Guardian / Carolyn Thompson, AP

The Discrepancy Lies in What Was Measured and Over What Horizon

Placing both studies side by side reveals no indication of manipulated data; both relied on genuine empirical records from the same group of schools. The divergence stems from three methodological choices.

First, how “strict enforcement” was defined. The NBER paper inferred the strength of a school’s policy from Yondr purchase records, but purchasing pouches does not guarantee consistent day-to-day implementation. CEPR, on the other hand, explicitly categorized schools that actively implemented strict pouch policies and evaluated them separately. Changing the definition fundamentally changes the denominator. Second, reporting averages versus subgroup distributions. NBER’s subgroup findings showed positive effects in high schools and negative effects in middle schools, which mathematically canceled out near zero when pooled together; CEPR reported the average gain specifically across pouch-adopting schools. One comparison yielded “close to zero,” while the other produced “positive but modest.” Third, academic outcomes versus holistic measures. NBER concurrently assessed student well-being, attendance, and bullying perception, uncovering negative friction during the transition year; CEPR’s design focused exclusively on standardized academic gains.

Both studies provided honest answers to the specific questions they framed. The ambiguity lies in the word “effective” itself. While the public clamors for a clean verdict on whether phone bans work, the data refuses to offer a simple headline — delivering answers strictly partitioned by definition, grade level, and timeline.

The Ban Stopped the Bleeding for a Vulnerable Cohort of Readers

The CEPR paper also highlights a critical finding that risks being buried beneath statistical averages. Kane’s team noticed that academic gains were distributed unevenly: a small subset of students who had previously experienced steep annual reading declines saw that downward slide halted under the ban. Kane hypothesized that these students belong to the subgroup most susceptible to digital distraction.

The 0.01 standard deviation average conceals this crucial nuance. For the broad majority of students, the ban translates to just one to two weeks of instruction per year; for that vulnerable subgroup, it caught an entire school year’s worth of lost reading progress. While the mathematical average remains accurate, the real-world value of public policy often resides in the tails of the distribution.

The policy landscape adds another layer of complexity: over 40 states have enacted legislation restricting phones in schools, yet most leave local enforcement rules to individual districts, creating a fragmented regulatory patchwork. For researchers, however, this fragmentation is welcome news: variations in enforcement intensity and grade-specific guidelines provide the exact variables required to unpack what truly works.

Taken together, the two studies offer a conclusion far more honest than either could deliver alone: phone bans do work, but their impact measures around 0.01 standard deviations, accompanied by halting severe reading declines for a small cohort of students. We crave a decisive answer, yet the empirical evidence offers only a modest figure. Whether this intervention is worth the effort depends on whether policymakers target the average score of all students or aim to rescue those in danger of falling a full year behind. That conclusion is more complex than a binary yes or no, but it represents the full truth the data currently allows.

Reference Links:

  • Harvard CEPR: States Leading States Initiative Inaugural Paper
  • NBER Working Paper 35132 (Allcott, Baron, Dee, Duckworth, Gentzkow, Jacob)
  • NPR Report: Pennsylvania District Student Demonstrates Yondr Pouches
  • The Guardian Report: Lackawanna School District Explains Phone Pouch Policy to Parents