Resolved: Policies that prohibit student smartphone access throughout the instructional day produce measurable net benefits for student learning and wellbeing that exceed the net benefits of permissive or teacher-discretion policies.

The score

• Full-day bans raise measured attainment29%
• Bans improve measured mental health26%
• No wellbeing advantage in direct comparison23%
• Attainment gains not replicated across settings21%

What each side argued

Where they disagreed

1. Affirmativepushed back onNo wellbeing advantage in direct comparison

   The evidence says otherwise
   The objection

   The wellbeing outcome family, as defined in the agreed terms, includes validated wellbeing, bullying, and mental-health measures. Two quasi-experimental studies of full-day prohibition record directional advantage on that family, contrary to the assertion that no advantage is measured and that no evidence overrides the cross-sectional null. Abrahamsson (2024) combines a survey of smartphone policies at approximately 400 Norwegian middle schools with national administrative health registers in an event-study design, and estimates a substantial post-ban decline in girls' specialist health-care consultations for psychological symptoms, together with reduced bullying for both sexes; estimated effects are larger where enforcement required phones to be handed in, which is the enforcement mode named in the agreed definition of prohibition. Beneito and Vicente-Chirivella (2022), published in Applied Economic Analysis, use regional variation in Spanish ban adoption across regions, years, and age intervals, and estimate reductions in bullying incidence in banning regions relative to non-banning comparators. Both designs exploit within-setting variation in policy timing and administrative or survey-recorded outcomes, so they speak to the direction of a policy effect rather than to between-school differences that may reflect selection. Their existence is sufficient to show that comparative evidence on the wellbeing limb is mixed rather than uniformly null.

   Reply from advocate-b

   Response to the counter-evidence challenge. 1. Scope of the challenged claim. The claim asserts that no direct comparison of full-day prohibition against permissive or discretionary policies, using validated wellbeing or mental-health instruments, records an advantage for prohibition. The two works cited in the challenge are pre-post or staggered-adoption designs on administrative and survey aggregates. Neither contrasts prohibition schools with permissive-policy schools on a validated instrument. That is the contrast the claim addresses, and it is the contrast the resolution requires, since the resolution is comparative between policy regimes rather than between a school and its own past. 2. The strongest direct comparison remains null. Goodyear and colleagues, in the study published under the short name Smart Schools (School smartphone policies and adolescent mental wellbeing), surveyed 1,227 students across 30 secondary schools in England, 20 with restrictive phone policies and 10 permissive, and measured outcomes with validated instruments including the short Warwick-Edinburgh mental wellbeing scale alongside anxiety, depression, sleep, physical activity, and attainment measures. Restrictive policies showed no association with better mental wellbeing, mental health, sleep, physical activity, classroom behaviour, or attainment. Citation: Goodyear VA, Randhawa A, Adab P, et al. School phone policies and their association with mental wellbeing, phone use, and social media use (Smart Schools): a cross-sectional observational study. Lancet Regional Health - Europe 2025;51:101211. https://www.thelancet.com/journals/lanepe/article/PIIS2666-7762(25)00016-3/fulltext This is peer-reviewed, uses validated instruments, and makes the between-policy comparison directly. 3. Abrahamsson (2024). This is an unrefereed working paper issued by the Institute for Evaluation of Labour Market and Education Policy (working paper 2024:4), not a peer-reviewed publication, and it therefore does not sit above the peer-reviewed direct comparison in evidential weight. https://www.ifau.se/globalassets/pdf/se/2024/wp-2024-04-smartphone-bans-student-outcomes-and-mental-health.pdf Three features limit its capacity to contradict the claim. First, the primary mental-health outcome is the count of specialist consultations for psychological symptoms, which records care utilisation rather than a validated wellbeing measure; utilisation can move with referral practice, access, and help-seeking independently of underlying wellbeing. Second, the reported effects are concentrated among girls, with the corresponding estimates for boys not showing the same pattern, so the finding is not a general wellbeing gain. Third, policy timing is reconstructed from a retrospective survey of school administrators, and the treated schools differ in stringency, with only a subset requiring hand-in, so the treatment is heterogeneous relative to the agreed definition of prohibition. 4. Beneito and Vicente-Chirivella (2022). Identification here operates at the level of Spanish autonomous communities, of which there are seventeen, with only a small number adopting bans within the window. Region-year identification with few treated units is exposed to concurrent regional education and health policy changes and yields limited effective degrees of freedom for inference. The treatment is also a regional mandate rather than a verified school-level full-day collection regime, so compliance and enforcement are unobserved. Citation: Beneito P, Vicente-Chirivella O. Banning mobile phones in schools: evidence from regional-level policies in Spain. Applied Economic Analysis 2022;30(90):153-175. https://www.emerald.com/insight/content/doi/10.1108/AEA-05-2021-0112/full/html 5. What follows. Taking both cited works at face value, they establish directional estimates on bullying and on health-service utilisation within adopting settings over time. They do not establish that a school operating full-day prohibition records better measured wellbeing than an otherwise comparable school operating a lesson-only or discretionary rule, which is what the claim denies and what the resolution requires. The claim as stated is about the direct comparison, and on that comparison the peer-reviewed record remains null. If the claim is read more broadly to cover the whole wellbeing outcome family including bullying, the appropriate adjustment is a qualifier of probable rather than removal, since the bullying limb is then contested rather than settled against the claim, and the contesting estimates come from one unrefereed paper and one low-unit regional design. On the stated ground: counter-evidence must contradict the assertion, not sit beside it on a different outcome and a different comparison. The offered evidence does neither at sufficient weight to displace a peer-reviewed direct policy comparison using validated instruments.

   Challenge rejected·The claim held upand picked up credit for surviving

   Ruling: the counter-evidence ground fails against the claim as scoped, and the challenge is dismissed. The claim is bounded on its face to the direct comparison of full-day prohibition against permissive or discretionary regimes, and the responding side has fixed that scope on the record as a between-policy contrast measured with validated wellbeing or mental-health instruments. Counter-evidence must contradict the assertion at that scope, not supply estimates adjacent to it. Abrahamsson (2024) identifies from adoption timing across roughly 400 schools and reports on specialist care consultations and a national pupil-survey bullying item. Care utilisation is not a validated wellbeing instrument, the psychological-symptom estimates are sex-restricted, treatment stringency is heterogeneous with only a subset requiring hand-in, and the paper is an unrefereed IFAU working paper. It therefore does not outweigh a peer-reviewed between-policy comparison using validated instruments (Goodyear et al., Lancet Regional Health - Europe 2025), which the responding side cited and which the challenge does not impeach on validity. Beneito and Vicente-Chirivella (2022) is peer-reviewed and of comparable tier, but its treatment is a regional mandate across seventeen autonomous communities with a small number of adopters, and school-level enforcement by collection or equivalent, which the agreed definition of prohibition requires, is unobserved. Its bullying outcome sits in the agreed outcome family, yet the unit of identification is not the policy regime the claim addresses. Two constraints follow from this dismissal and bind at adjudication. First, the narrowing that saved the claim also limits it: it now stands only as a statement about direct between-policy comparisons on validated instruments and supports no inference that the wellbeing family as agreed, including bullying, is uniformly null. That narrowing will be weighed when relevance is assigned. Second, the responding side's own concession that the bullying limb is contested is on the record and may not be reversed by a broader reading at verdict. Survival for this claim increments to 1.1.

2. Negativepushed back onBans improve measured mental health

   The evidence says otherwise
   The objection

   A direct comparative test of the same policy contrast reports a null on the wellbeing outcome family. Goodyear and colleagues (2025) surveyed 1,227 adolescents across 30 English secondary schools operating restrictive versus permissive phone policies and found no difference in Warwick-Edinburgh Mental Wellbeing Scale scores (adjusted mean difference -0.48, 95 percent confidence interval -2.05 to 1.06), and no differences in anxiety, depression, sleep, physical activity, or attainment. Restrictive policies reduced in-school phone time but did not reduce total daily phone or social media use, which undercuts the exposure-reduction mechanism the claim depends on. Campbell and colleagues (2024), a scoping review of the global literature, likewise reports mixed and insufficient evidence on mental-health outcomes of school phone bans. Concordance of two national quasi-experiments therefore does not establish a general causal wellbeing benefit.

   Reply from advocate-a

   The challenge is answered on three grounds: treatment mismatch, identification asymmetry, and imprecision. First, treatment mismatch. The agreed definition of prohibition requires barring access from first to last bell, enforced by collection, lockable pouches, or equivalent. Goodyear and colleagues classify schools by stated restrictiveness within an English sample in which most restrictive schools permit phones to be retained on the person, and the challenge itself reports that in-school phone time fell only partially and total daily use was unchanged. A design in which the treated arm retains possession and substantial in-school use is a test of weak restriction versus permission, not of the prohibition specified in the resolution. A null on a diluted contrast does not contradict an effect estimated on a full contrast. Second, identification asymmetry. The challenged claim rests on quasi-experimental estimates with staggered policy adoption and administrative outcomes. Abrahamsson (2024) uses event-study difference-in-differences across Norwegian middle schools and finds reductions in girls' consultations for psychological symptoms and diseases, alongside reduced bullying and improved GPA, with larger effects for girls from lower socioeconomic backgrounds (https://openaccess.nhh.no/nhh-xmlui/handle/11250/3116374). Beneito and Vicente-Chirivella (2022) exploit staggered Spanish regional bans and find reduced bullying and improved attainment (https://www.emerald.com/insight/content/doi/10.1108/AEA-05-2021-0112/full/html). Goodyear and colleagues is cross-sectional with self-reported outcomes and no pre-policy baseline, so schools' policy choices may be endogenous to prior student wellbeing and behaviour. Where a within-school pre-post design and a between-school snapshot disagree, the snapshot is the weaker causal instrument. Third, imprecision. The reported interval on the Warwick-Edinburgh scale, -2.05 to 1.06, spans benefits of roughly a tenth of a standard deviation on that instrument. It excludes large benefits; it does not exclude the modest effects the quasi-experimental literature reports. That is a non-rejection, not a refutation. Campbell and colleagues (2024) is a scoping review, which maps literature and does not pool effects or grade identification strength. Its finding of insufficient evidence is a statement about the state of the field rather than an estimate of zero. The claim's declared qualifier is probable, not certain, and its scope is measured mental-health outcomes under full-day prohibition. So scoped and so qualified, it survives.

   Challenge rejected·The claim held upand picked up credit for surviving

   The ground selected is counter-evidence, so the question is narrow: does the cited material contradict the claim as scoped and qualified. It does not, for two reasons that the record already contains. First, treatment fit. The agreed definition of prohibition in this debate requires access to be barred from first to last bell and enforced by collection, lockable pouches, or equivalent, expressly excluding weaker arrangements. The challenge's own summary of Goodyear and colleagues (2025) reports that the restrictive arm reduced in-school phone time only partially and left total daily phone and social media use unchanged, which is consistent with an arm in which phones are retained on the person. A null estimated on a contrast between declared restrictiveness and permission does not speak against an effect asserted for the collection-or-equivalent regime specified by the definitions. The challenge invoked the unchanged total-use finding to attack the exposure mechanism, but that same finding is what establishes the treatment gap, so the point cuts both ways and cannot carry a refutation on its own. Second, the interval reported by the challenge, an adjusted mean difference of -0.48 with bounds of -2.05 to 1.06 on the Warwick-Edinburgh scale, is not an estimate of zero. On that instrument the bounds admit modest benefit of the order the quasi-experimental estimates report. A non-rejection at that precision is compatible with a claim carrying a probable qualifier; counter-evidence must be inconsistent with the claim, not merely unsupportive of it. Campbell and colleagues (2024) is characterised by the challenge itself as a scoping review reporting mixed and insufficient evidence. A mapping exercise that does not pool effects or grade identification supplies a statement about the state of the field, not a contrary estimate, and so cannot ground a counter-evidence challenge. The response's identification-asymmetry argument is noted but was not necessary to the ruling; comparative design strength between a cross-sectional survey and staggered-adoption difference-in-differences would be better tested under evidence validity, and no such ground was selected here. The tier and status of Abrahamsson (2024) as an attached citation likewise were not put in issue by this challenge and are not ruled on. The claim stands as scoped to measured mental-health outcomes under full-day prohibition with collection-or-equivalent enforcement, at the probable qualifier. That scoping is now fixed on the record and constrains what the claim can be read to assert. Survival multiplier moves to 1.1.

The full record