Smartphone Bans, Student Outcomes and Mental Health

477 middle schools linked to registry data (45% of 1,187 surveyed schools responded; 40% could be linked); 161,371 student-year observations in the baseline sample; 53,484 observations for externally graded mathematics exams, 49,657 Norwegian, 52,820 English. Bullying is measured at school level from the compulsory national Pupil Survey.

Norwegian middle-school students (grades 8-10, ages roughly 13-16), nationwide, linked to population registers including the Norwegian Patient Registry and GP claims. Norway has no national policy, so each school sets its own phone rule, which is what generates the variation.

The strongest data in this cluster by some distance: whole-population administrative registers (specialist and GP consultations coded ICD-10/ICPC, school-leaving GPA, externally marked exit exams, high-school track enrolment, compulsory bullying survey), linked to a purpose-built survey of school smartphone policies. Estimation is a non-parametric event study in the style of Bailey & Goodman-Bacon, with the year before adoption as baseline and periods beyond four years binned, run separately by gender and by paternal education. Pre-trends are flat for the main educational outcomes. Because externally marked exams are graded blind by non-teachers, the paper can separate real learning gains from teachers grading better-behaved pupils — an unusually clean test that most of this literature cannot run. WEAKNESSES, and there are several. (1) Adoption timing is school-chosen, the same endogeneity that keeps Beland & Murphy and Kessel off grade B; this is why the grade is C despite register-quality data. (2) The survey response rate is 45%, and only 40% of schools could be linked; responding schools have slightly higher parental earnings. (3) The treatment is coarse — only 45% of responding schools had what the author calls a "strict" policy (phones prohibited or collected into "phone hotels"); the rest merely restricted use during lessons, and commentators have questioned whether the lenient tier is a ban at all. (4) A large number of outcomes, gender splits, SES splits and event-time coefficients are estimated with no multiple-hypothesis correction reported, and many headline p-values sit between .05 and .10 (GPA p = .064, GP visits p = .076, bullying p = .058-.094). (5) The gender contrast, which is the paper's entire story, is never shown to be statistically significant: "I cannot reject the null hypothesis that the coefficients between girls and boys are equal, in any of these results." Independent academic work (Norwegian School of Economics / Norwegian Institute of Public Health), funded by the Research Council of Norway. Displacement was NOT measured directly — no time-use, sleep or attendance data — although the health-care and bullying outcomes are a partial window onto what the phone had been doing. Every number recorded here was read from the full text of NHH Discussion Paper SAM 01/24 (February 2024); the paper was subsequently published in the Journal of Human Resources (online 11 March 2026), whose abstract is more cautious than the discussion paper's, and whose tables were not consulted.

On the full sample the ban does nothing to GPA, teacher-set grades or the probability of entering an academic high-school track. The results the paper is famous for are all in the girls' subsample: girls exposed to a ban for the whole of middle school gain 0.08 SD in GPA (p = .064) and 0.09 SD in teacher-set grades (p = .05), and — the most credible number, because it is marked blind by external examiners — 0.07 SD in mathematics one year post-ban rising to 0.22 SD four years post-ban (p = .014). Girls become 4-7 percentage points more likely to enter an academic high-school track (p < .05, on a base of 49%). On health, girls' specialist consultations for psychological symptoms fall by about one visit after one year and 2-2.7 visits after three to four years (roughly 60% of the pre-treatment mean of 3.4), and GP visits for the same causes fall 0.22 (about 29%, p = .076). School-level bullying falls 0.25-0.35 SD two to four years post-ban for the whole sample, and 0.42 SD for fully exposed girls. Effects concentrate among girls whose fathers lack academic education: GPA +0.09 to +0.14 SD and bullying down 0.58-0.99 SD, versus nothing for high-SES girls. Nothing shows up for boys anywhere. The caveat that most changes how it should be read: the girls-versus-boys difference is never statistically established, and the average effect is zero, so this is a subgroup finding of the kind this archive discounts by default.

Medium. Within-school before/after comparison on population registers, so pupil genotype does not change at the ban date; but the SES split (paternal education) is heritable on both sides and describes which families respond rather than establishing an environmental cause.

This paper sits on both sides of the Beland & Murphy / Kessel disagreement at once, which is why the flag is `mixed` and why the popular reporting of it is misleading. Its FULL-SAMPLE results on GPA, teacher-set grades and academic-track choice are NULL — the author says so in a footnote and notes they "are similar to the findings of Kessel et al. (2020)" — so on the average effect this is a third independent corroboration of the Swedish null, not a vindication of the English positive. What is new is the gender split: everything positive is in the girls' subsample. That split has not been independently replicated, and the author herself cannot reject equality of the girls' and boys' coefficients in ANY of the educational results — a fact almost universally dropped when this study is cited. Figlio & Ozek (2025, Florida) find the opposite gender pattern (larger gains for boys), which is not reassuring for either subgroup story. The Journal of Human Resources version (published online 11 March 2026, doi 10.3368/jhr.0224-13403R2) leads with the honest framing that the discussion paper buried: "banning smartphones in middle schools has no average effect on education or mental health but masks important gender differences."