The Digital Whale

Screens and young people

Technology in the classroom

Neither "technology improves learning" nor "technology harms learning" survives the evidence, and the best-known claim about handwriting failed a direct replication.

The strongest cross-national data shows that students who use computers more often at school do worse in reading and mathematics, and that school systems investing most heavily in classroom technology have not seen results improve. The strongest meta-analytic data shows positive average effects for one-to-one laptop programmes. Both are real findings and neither settles the question, because the first is cross-sectional and cannot separate the technology from the reasons a school deployed it, and the second rests on a small and mostly non-randomised subset of its own included studies. The one claim in this area that most people treat as settled — that handwriting beats typing for note-taking — failed a direct replication in the journal that published it. Phone restrictions have spread across education systems far faster than evidence about their effects. For a workplace-oriented comparison point on how digital activity is translated into metrics, see this guide.

Frequent classroom computer use tracks worse results

The OECD's analysis of its own international assessment data remains the most-cited negative finding in the field.

2015

Students who used computers more frequently at school performed worse in reading and mathematics, and school systems that invested most heavily in classroom ICT showed no appreciable improvement in results.

Direction: Decrease. Strength of evidence: Mixed.

OECD, Students, Computers and Learning: Making the Connection, 2015PISA data across 31 countries and economies

Caveat Cross-sectional PISA data in which schools with struggling students may deploy more remedial technology, so reverse causation is plausible; the OECD's own PDF was robots-blocked, so this was verified through the Hechinger Report's coverage.

Reverse causation is not a technicality here. Remedial software is deployed where results are already weak, and one-to-one devices are often given first to schools serving disadvantaged intakes. A cross-sectional comparison cannot tell that apart from technology damaging attainment.

Distraction is the mechanism students report themselves

The later PISA round asked students directly, and the answers give the negative finding a plausible route.

2023

66% of US students reported being distracted by their own digital device use in maths lessons and 54% by other students' devices, and those distracted by others' devices scored 15 points lower in mathematics.

Direction: Decrease. Strength of evidence: Mixed.

OECD, PISA 2022 Results, 2023PISA 2022, US student responses

Caveat Self-reported distraction is itself likely a marker of lower engagement, the OECD frames the relationship as an association, and the OECD source was robots-blocked so figures come through Education Week's reporting.

The more interesting half of that result is the second-hand effect. Being distracted by other people's devices is not something an individual student can opt out of, which is the strongest available argument for treating classroom device policy as a collective rather than an individual matter. The general evidence on interruption and recovery is covered in what interruptions do to focus.

Restrictions have spread faster than evidence about them

Policy has moved quickly, and the count of policies is well documented even though their effects are not.

2024

By the end of 2024, 79 education systems — 40% of those tracked — had laws or policies banning smartphone use in schools, up from 60 systems, or 30%, a year earlier.

Direction: Increase. Strength of evidence: Strong.

UNESCO Global Education Monitoring Report team, 2024Global count of education systems tracked by the GEM Report team

Caveat A count of policies rather than of enforcement or outcomes, and UNESCO advocates for restriction, which makes this an advocacy-adjacent source even though the count itself is factual.

What those bans achieve is unresolved. The two best studies disagree, and both are discussed in the evidence on adolescent mental health: an English observational study across 30 secondary schools found no wellbeing difference between restrictive and permissive policies, while a Norwegian working paper using the timing of ban adoption reported benefits concentrated among girls.

The handwriting result did not replicate

The claim that students who take notes by hand learn more than students typing them is repeated in teacher training, in newspapers and in device policies. It comes from three small laboratory studies.

2014

Longhand note-takers outperformed laptop note-takers on conceptual questions across three laboratory studies, attributed to laptop users transcribing verbatim rather than reframing content.

Direction: Increase. Strength of evidence: Weak.

Mueller & Oppenheimer, Psychological Science, 2014Three laboratory studies of college students

Caveat Small laboratory studies with immediate post-lecture testing and no study period, a design that bears little resemblance to real coursework.

A well-powered direct replication in the same journal seven years later found the behavioural difference but not the learning difference.

2021

Laptop users wrote more words and more verbatim text but performed no worse on the quiz, and exploratory mini meta-analyses across eight similar studies did not support the encoding explanation.

Direction: No detectable effect. Strength of evidence: Strong.

Urry et al., Psychological Science, 2021Direct replication with 142 participants, plus mini meta-analyses across eight similar studies

Caveat A single well-powered replication of the original Study 1 only, testing immediate learning rather than longer-term retention or note review, so it does not rule out delayed benefits.

The honest summary is that the mechanism people cite for the handwriting advantage — that typing encourages shallow transcription — was confirmed as a description of behaviour and not as a cause of worse learning.

One-to-one laptop programmes show positive effects on a thin base

The evidence pointing the other way comes from a research synthesis of school programmes that gave every student a device.

2016

One-to-one laptop programmes showed significantly positive average effect sizes in English, writing, mathematics and science across a synthesis of 96 studies.

Direction: Increase. Strength of evidence: Mixed.

Zheng, Warschauer, Lin & Chang, Review of Educational Research, 2016Synthesis of 65 journal articles and 31 dissertations published 2001–2015

Caveat Only about 10 of the included studies contributed to the quantitative meta-analysis, few were randomised, and the per-subject effect size values could not be verified from any accessible source, so they are not quoted here.

Data protection is the least measured part of the picture

The question schools are rarely asked to answer concerns what happens to the data their software collects.

2023

Only 16% of countries explicitly guarantee data privacy in education by law, and 89% of 163 edtech products recommended to students during the pandemic could collect data on children.

Direction: Increase. Strength of evidence: Mixed.

UNESCO GEM Report, Technology in Education: A Tool on Whose Terms?, 2023Global review of education systems and 163 edtech products recommended during the pandemic

Caveat The 163-product figure derives from a Human Rights Watch analysis of pandemic-era recommendations, which is a non-random sample, and UNESCO holds an institutional position favouring restriction.

The short version

  • Across 31 countries, students using computers more frequently at school performed worse in reading and mathematics, on cross-sectional data that cannot establish why (OECD, 2015).
  • In PISA 2022, 54% of US students reported being distracted by other students' devices in maths lessons, and that group scored 15 points lower (OECD, 2023).
  • Smartphone bans covered 79 education systems by the end of 2024, up from 60 a year earlier (UNESCO), which is a count of policies and not of outcomes.
  • The 2014 finding that handwritten notes beat typed notes failed a direct replication with 142 participants in the same journal (Urry et al., 2021), and should not be cited as settled.
  • One-to-one laptop programmes show positive average effects, but only about ten studies contributed to that estimate and few were randomised, so the finding is weaker than its frequent citation suggests.