Today the OECD published the PISA 2025 Results, and Spain posted the worst result in its history in reading, mathematics and science all at once. Within hours, much of the Spanish press already had a culprit: screens. "Spain crashes in PISA, and students struggle with texts over 400 words — because of screens," read one elDiario.es headline. Other outlets talk about "curbing screens" as if that were the report's obvious conclusion.
It isn't. And it's worth pausing to read the data calmly before signing on to the headline.
An Imperfect Framework, But the Only One We Have
Let's start with the obvious: PISA is not a perfect instrument. It reduces entire education systems — their culture, funding, history, regional inequality — to three numbers per subject, from a two-hour test given to a sample of 15-year-olds. A system can have real strengths PISA doesn't measure (creativity, social cohesion, wellbeing) and weaknesses it overstates. No serious expert treats PISA as the final word on a country's education.
That said, it's also the most rigorous, systematic international comparison that exists: close to 760,000 students across 91 countries and economies, using the same methodology for more than two decades. Rejecting the framework outright because it's imperfect would be just as unrigorous as accepting it uncritically. If we accept its rules of the game — the same rules the entire public debate of the past few days has already accepted — some conclusions do hold up against the data. Let's get to them.
The Usual Suspect
We've written about this pattern before at Papitek. When we asked whether banning social media for minors was actually a good idea, or when we reviewed the silenced debate over screens in the classroom, we reached the same conclusion: in Spain there's a dominant strand of opinion, with plenty of media amplification, that turns technology into the favorite scapegoat for any educational problem. It's a comfortable explanation — an external, visible enemy, one we were already uneasy about for other reasons (mental health, addiction, social media) — instead of more uncomfortable explanations that would require looking inward.
And here's the trap: the report doesn't have to say "screens are the cause" for the headline to say it for them. We reviewed several Spanish outlets published today, and the pattern repeats: the headline oversimplifies far more than the body of the article does. elDiario.es itself quotes a senior OECD analyst, Tue Halgreen, in far more cautious language than its own headline — "there are a lot of items pointing in that direction" — nowhere near a demonstrated cause. Spain's Secretary of State for Education, quoted in the same piece, says the poor results "may possibly be driven" by rising screen use — a hypothesis, not a verdict. And according to Cronista.com, Spain's education unions (CCOO, UGT, CSIF, STEs-i, ANPE) point to teacher shortages, high student-teacher ratios, absenteeism and constant legislative changes as factors just as relevant as technology — which shows up as one factor among several, not the cause.
What the Numbers Actually Say
Let's get to the data, because it's stark and worth keeping in view amid all the debate.
Spain scored 451 points in reading, 457 in mathematics and 477 in science — its worst-ever result in all three subjects since the series began. It sits below both the OECD average (461 / 463 / 482) and the EU average (459 / 463 / 481) in all three. The drop in reading between 2022 and 2025 is 23 points — and 45 points since 2015. That's also larger than the EU's average decline over the same recent period (roughly 16 points).
That said, Spain's exceptionalism shouldn't be overstated: looking at the full results table, Spain is nowhere near the country that fell the most. Latvia lost 45 points in reading between 2022 and 2025 alone, Denmark 29, Malta 31. Spain belongs to a broad group of countries with steep declines, well above the OECD average (−14) — but it isn't a unique case, and treating it as one distorts the diagnosis.
This isn't a uniquely Spanish phenomenon: the OECD average has fallen 28 points in reading since 2015, its lowest level ever recorded in PISA's history, with one in five 15-year-olds now classified as a low performer across all three subjects at once. But it isn't an inevitable fate either: Türkiye improved more than any other OECD country this cycle, rising in all three subjects since 2022. Widespread decline isn't a climate-scale fatality — some countries are doing something different, and it's working.
It's How, Not How Much
This is the heart of the article — and also where the dominant narrative falls apart.
The report itself avoids talking about causation, and this no longer needs to be inferred from press coverage — it's in the text itself. In the preface signed by OECD Secretary-General Mathias Cormann: "In reading, increased digitalisation, longer screen time and falling rates of reading for enjoyment have coincided with weaker literacy outcomes." Coincided — a temporal overlap, not a demonstrated cause. On the relationship with performance, the same preface adds: "Digital and AI tools are increasingly embedded in students' learning, but their relationship with performance depends on how they are used. Across OECD countries, moderate use of digital devices for learning is associated with stronger science performance, while excessive use and digital distraction are associated with weaker outcomes."
What the report does draw a clear line on is quantity versus type of use. And there's a finding that directly undercuts the most popular solution in the Spanish debate — banning phones in the classroom. According to PISA 2025's own data, changes in cell phone bans between 2022 and 2025 are not associated, at the system level, with changes in science or reading performance. If anything, countries with more schools banning phones tend to have lower, not higher, average science performance. The report itself rushes to add an important qualifier, worth quoting in full so it isn't misread: "The PISA data do not allow for causal inference and therefore do not necessarily indicate that the bans themselves lead to lower performance. Rather, the findings may suggest that systems with lower performance are more likely to have introduced such bans as mitigation measures." In other words: it's probably not that banning phones makes results worse — it's that systems that were already struggling are the ones banning the most, as a reaction. But if banning were the fix that so much of the public debate promises, we'd expect to see improvement where bans are most common. We don't.
The country comparison backs this up. Japan has one of the lowest shares of schools banning phones (22%, versus a 49% OECD average) and, despite that — or precisely because of how it manages use, not because it bans it —, it's among the OECD countries with the lowest use of devices during class time, with only 5% of students reporting frequent distraction. It has near-universal 1:1 device infrastructure (the GIGA School program) and still keeps in-class use tightly scoped and directed. South Korea, at 23% ban prevalence, remains among the best in the world despite losing 15 points in reading. Singapore uses more device time for learning than the OECD average (2.1h versus 1.7h) and is among the top two education systems on the planet. And Estonia reports more digital distraction than the OECD average — and yet that distraction isn't significantly associated with worse science performance there, and the country remains among the OECD's top three in all three subjects.
Andreas Schleicher, the OECD's Director for Education and Skills, sums it up with a very concrete example, right in the report itself: "In Singapore, despite its high-tech classrooms, students still spend substantial time with print texts, and reading remains a centrepiece of early education." That's not a contradiction — it's proof that technology and reading can coexist when a system decides, with judgment, what each one is for.
That these countries post strong results without restricting technology, and that Spain posts poor results without being an especially hyper-digitalized classroom environment by comparison, should be enough to dismantle the "more technology equals worse reading" equation. The real pattern is different: it's how it's used that matters, not whether it's used.
This repeats, with some nuance, for artificial intelligence — the biggest new addition to this PISA cycle, which for the first time asks students about their use of generative chatbots. Spain sits, according to various accounts of the report, above the OECD average in weekly use. And yes, there's a real gap: students who don't use AI for specific schoolwork tasks (summarizing, research) tend to score better in science than those who do. But the relationship isn't as simple as "more use, worse score": for general-purpose AI use as a learning aid, the report itself finds that weekly users perform similarly to non-users — it's daily use and very occasional use, the two extremes, that score worse. And among students who receive classroom training to critically evaluate what AI produces, performance improves relative to those who don't get that training. The OECD sums it up with a sharp image: technology should be scaffolding, not a crutch — something that supports a student's own thinking, not a substitute for it.
The Self-Criticism We're Missing
If the problem isn't simply "how much technology there is," then the problem lies, in large part, in how it's being integrated — and that's exactly where real self-criticism from the education community is missing, instead of pointing outward.
No one should downplay teacher shortages, high ratios or absenteeism — they're real factors, and the unions themselves rightly put them front and center. But that explanation shouldn't be allowed to cover for a more uncomfortable one: the data itself points to a pedagogical knowledge gap in how to use educational technology effectively. The performance difference between students who receive training to think critically about AI and those who don't isn't a hardware problem — it's a methodology problem, about knowing how to teach the effective use of a tool that's already in every classroom, whether we like it or not. And the finding that "hasty readers" (students who answer quickly but incorrectly) rose 5 percentage points since 2018 — from almost 7% to 11% — points to something that isn't fixed by taking away the phone either: how attentive reading is taught in an information-saturated environment.
The report itself devotes an entire section to explaining why reading has declined, and that's where the self-criticism becomes unavoidable. It identifies four forces, not one — students (less persistence, more rushed answers), schools and classrooms, the education system as a whole, and broader societal trends (where technology enters, last on the list, not first). On schools, there's a data point that should sting more than it currently does in the Spanish debate: according to the TALIS 2024 teacher survey, the share of class time spent maintaining order — not teaching — rose from 13% in 2018 to 16% in 2024. And the report itself is a reminder worth keeping in mind for Spain: spending more doesn't guarantee anything on its own. Luxembourg is, by far, the biggest per-student spender in the entire OECD — and posts lower science scores than Ireland, France or Italy, all of which invest less than half as much. Money matters, but it doesn't substitute for judgment.
None of these four factors is an argument against technology. They're arguments about methodology — about whether an education system knows how to manage a classroom, train its teachers, and use its own data to improve. And they're the ones getting the least space in the headlines.
Stripping technology out of the classroom, as much of the dominant narrative proposes, isn't free: it would mean giving up what actually works when used well — and judging by Japan, Korea, Singapore or Estonia, it's entirely possible to use it well without results suffering. Spain's problem doesn't seem to be that there's too much technology in its classrooms. It seems to be that there's too little judgment about how to use it — and that's a teacher-training and methodology problem, not a hardware one.
Main sources: "PISA 2025 Results (Volume I)," OECD 2026 — full official text, verified directly against the report's PDF (preface, foreword, executive summary, results table and the digital resources/AI chapter), not just press coverage. Supplemented by coverage from El Español, elDiario.es, Infobae España and Cronista.com. All direct quotations from the OECD report above are given in their original English, taken from the primary source text — not translated back from a Spanish rendering. A small number of figures (exact phone-ban percentages by country; Spain's exact weekly AI-use rate) still rely on secondary press coverage, as they live in report charts not extractable as plain text — see research/pisa-2025-informe-lectura-espana.md for the full sourcing detail.