# Related Research and Trends Back to [[Reverse Flynn Effect and EdTech]] · Sources: [[Research]] · Terms: [[Vocabulary]] ## 1. The Flynn effect and its reversal - **The rise:** - IQ raw scores climbed about 3 points per decade through most of the 20th century. [[People/James R. Flynn|James Flynn]] documented this in 1984 and 1987. - A meta-analysis of 271 samples (Pietschnig & Voracek 2015) puts the rise at about 0.28 points per year from 1909 to 2013, slowing in recent decades. - **The reversal:** - Since the 1990s, declines have been reported in Norway, Denmark, Finland, the UK, France and some other developed countries (Dutton et al. 2016 review: nine studies in seven countries, per IQ Metrics ◻). - The best evidence is Norwegian. In [[People/Bernt Bratsberg and Ole Rogeberg|Bratsberg & Rogeberg]] (2018), scores peaked for men born in 1975 and then fell, **including between brothers**. That rules out changes in who has children and points to the environment. - That series stops with the **1991** birth cohort, which is before Gen Z. - **Not a uniform decline:** - US WAIS-5 norms (Trudel, Winter & Kaufman 2025) show fluid reasoning and processing speed still **rising** in 16-year-olds, while verbal comprehension and working memory fall. - The US online SAPA sample (2006–2018) shows reasoning down, verbal flat and spatial rotation up. - The latest Austrian attention study ("No Age of Inattention…", Oct 2026) finds small, mixed changes in adults. - **Proposed causes** (all contested): changes in schooling and curricula, media and reading habits, test-content drift and test-taking motivation, nutrition and health. The Norwegian authors say the cause is environmental but do not identify it. - **Screens and EdTech are one hypothesis.** No study I found links a cohort's IQ decline causally to classroom technology. ## 2. Achievement trends (what really did fall) - **PISA 2025** (OECD, Sep 8, 2026): - Record-low OECD averages in math and reading. Reading fell 28 points and math 22 points from 2015 to 2025. - One in five students is a low performer in all three subjects, up from 16% in 2022. - Science was stable from 2022 to 2025. - On devices: "Students tended to perform less well when use became excessive or when devices were used for leisure at school." More than one in four students said classmates were distracted by devices in most or every science lesson. - OECD Secretary-General Mathias Cormann: "Technology and AI can strengthen learning… but only when used purposefully and not as a substitute for attention, effort and understanding." - **NAEP 2024**, released Sep 9, 2025: - 12th-grade math and reading each fell 3 points from 2019. 45% of 12th graders scored below Basic in math, the highest ever, and 32% in reading, also the highest. - 8th-grade science fell 4 points. - **The lowest-performing students drove most of the decline. This trend predated the pandemic and worsened after it.** - **The timing debate:** - Horvath dates the turn to "around 2010," his written testimony to the "mid-2000s," and [[People/Jean Twenge|Twenge]] to "around 2012." - The pandemic (2020–22) also sits inside every recent decline. ## 3. Screens versus paper, typing versus handwriting - **Reading:** meta-analyses (e.g., Delgado et al. 2018 ⚠️ from memory) find a small advantage for paper on comprehension, larger for expository text and under time pressure. - Horvath's book gives −0.15 overall and −0.29 for expository text (per Education Next). His testimony gives −0.16. - **Note-taking:** Mueller & Oppenheimer (2014) found laptop notes led to verbatim transcription and weaker conceptual answers. A larger replication (Morehead, Dunlosky & Rawson 2019) found the effect was **not robust** ⚠️ (from memory). - Horvath reports −0.21 overall for typing versus handwriting. - **Distraction:** off-task device use in classes is well documented, but the main study he cites (Ragan et al. 2014) covered university lectures. ## 4. Policy trends - **Sweden:** - Sweden reversed its digital-first push in schools and started buying textbooks again: SEK 685M (2023) and SEK 658M (2024) for textbooks, plus SEK 176M (2024) for books ([regeringen.se](https://www.regeringen.se/pressmeddelanden/2024/02/regeringen-avsatter-658-miljoner-for-inkop-av-larobocker/)). The school minister was Lotta Edholm. - ⚠️ "Banned most ed tech for primary years" (Horvath, Q&A) overstates this. I found funding and guidance changes, not a ban. - ⚠️ The "2 billion kronor" figure in the quote post is not confirmed. - **Norway and Finland:** Horvath said Norway "just followed suit" and Finland would within about 2 months. ⚠️ I did not check either claim. - **Australia:** a ban on social media accounts for under-16s took effect in Dec 2025 ⚠️ (date from memory). Horvath backed something similar for the US. - **US:** - Many states have adopted school-day phone restrictions or bans since 2023 ⚠️ (I have no verified count). - The Kids Off Social Media Act (KOSMA; Cruz and Schatz) was discussed at the hearing. - Horvath also urged independent efficacy standards for federally funded EdTech. - **AI tutors:** the next wave. PISA 2025 found that students who did not use AI to draft writing assignments outperformed those who did. Among frequent users, students taught to assess AI output did better. ## 5. The wider screens and youth debate - **Jonathan Haidt**, *The Anxious Generation* (2024): attributes the post-2012 rise in teen anxiety and depression to the phone-based childhood. - Chalkbeat reports that Haidt has promoted Horvath's work. - **Candice Odgers** (Nature review, 2024 ⚠️ from memory): argues the causal evidence is weak and the panic could crowd out better interventions. - **[[People/Jean Twenge|Twenge]]:** cohort-trend research (*iGen*, *Generations*). Her 2026 *Journal of Adolescence* paper links PISA declines to leisure device use at school. - **[[People/Jenny Radesky|Radesky]]:** pediatric and design-focused. She is concerned about engagement-optimized design and, at the hearing, about AI companions above all. ## Criticism of Horvath's case - **[[People/Elizabeth Tipton|Elizabeth Tipton]]** (Northwestern statistician, via Chalkbeat, Mar 17, 2026): - Re-centering effect sizes against Hattie's +0.42 benchmark is "wildly unrealistic." - Averaging meta-analyses of meta-analyses is "garbage in, garbage out": "We are so far from the original data…" - **Daniel Willingham** (UVA, via Chalkbeat): quoted on the concern about distraction from classroom devices. ⚠️ I have not checked his full position. - **The sign flip:** - Raw EdTech effects in the book are **positive**: +0.29 overall, +0.52 for tutoring systems, +0.16 for 1:1 laptops (Education Next). - The negative numbers shown in the testimony come from subtracting 0.42. - Hattie's "hinge point" has itself been criticized as an average of very different study types ⚠️ (general critique; I did not trace a specific paper). - **Correlation:** - Horvath conceded at the hearing that "this is all correlative." In PISA and TIMSS, heavy computer use is tangled up with selection (who gets devices, remediation, and weaker schools). - Carl Hendrick's otherwise favorable review in Education Next says attributing the Flynn reversal to EdTech "requires more careful causal analysis than meta-analytic correlations can provide." - **The IQ claim:** - His own Substack (Feb 6, 2026) concedes: "we cannot concretely say that Gen Z is less intelligent than Millennials." He reframes IQ as "school-ability." - The best IQ series ends before Gen Z. - **Cited evidence cutting the other way:** his ref 4 (Andrzejewski et al. 2024) is titled "…evidence for **better** test performance." - **Unsupported physiology:** the Q&A claim about "tachykinins… precursor to chemical depression" has no source I could find. - **Other critiques I found but did not read:** - eSchool News (Apr 1, 2026) ⚠️ - an Adam Sparks Substack ⚠️ - **His reply** (Chalkbeat): "We are much closer to proving that it's harmful than we are to proving it's helping."