Alarm Bells at Universities
At the University of California, San Diego (UCSD), they began to notice that something was not adding up in mathematics classes. More and more first-year students needed additional instruction before they could tackle university-level calculus. In 2020, 32 incoming students were placed in the most elementary mathematics courses. Two years later, there were 390. By the fall of 2025, the figure had risen to 921.
The scale of the problem could also be seen in the kinds of gaps being detected. An internal university report found that, between 2020 and 2025, the number of first-year students whose placement tests showed that they did not meet high-school mathematics standards had increased almost thirtyfold. In 2025, they represented roughly one in eight incoming students. And more than 70% of them did not even reach the level expected in middle school, roughly between the ages of 11 and 14. Altogether, one in twelve students entering UCSD was in that situation.
The issue has generated intense controversy within the University of California. In May 2026, professors of mathematics and other STEM disciplines signed an open letter warning about the gaps they were encountering in the classroom. At Berkeley, for example, diagnostic tests administered over three academic years revealed significant gaps in preparation among between 20% and 30% of the first-semester calculus students who took them. Some needed to review even fractions and exponents.
Similar warning signs are coming from the humanities. In 2024, journalist Rose Horowitch interviewed 33 university professors for The Atlantic. Nicholas Dames, a literature professor at Columbia since 1998, said that his students were finding it increasingly difficult to cope with the complete books required for his course. A possible explanation began to emerge when one student admitted that throughout high school she had never been asked to read an entire book. At Georgetown, the head of the English department described concentration problems even when students were faced with a fourteen-line sonnet. At Princeton, historian Anthony Grafton observed a more limited vocabulary and greater difficulties with language.
These are testimonies, not a representative assessment of university students. And complaints about the decline of younger generations have a history almost as long as education itself. But this time we have something else against which to test them: data. And very recent data at that.
Looking Back
PISA allows us to pause the film just before university. Every three years, the OECD assessment measures the extent to which 15-year-old students can apply what they have learned in reading, mathematics and science. In its 2025 edition, whose results have just been published, more than 760,000 students from 91 countries and economies took part.
The picture is hardly encouraging. PISA 2022 had recorded an extraordinary drop in performance coinciding with the pandemic: between 2018 and 2022, the OECD average fell by 15 points in mathematics and 10 in reading. It might have been expected that, once normal schooling resumed, a significant part of that loss would begin to be reversed.
That has not happened. In 2025, average OECD performance fell again in reading and mathematics, reaching the lowest levels ever recorded by PISA. If we extend the comparison back to 2015, students have lost 28 points in reading and 22 in mathematics. The OECD estimates that those differences are equivalent to approximately a year and a half of learning in reading and just over a year in mathematics. Science, by contrast, has remained stable since 2022, although it too is below the levels recorded a decade ago.
The decline, then, cannot be explained solely as a scar left by the pandemic. Part of it had begun earlier and has continued since.
There is another way of looking at the phenomenon that is especially relevant when considering what tools students possess by the time they finish school. PISA defines Level 2 as the baseline level of proficiency. In reading, it means, among other things, being able to identify the main idea of a moderately long text, locate information according to certain criteria, and reflect on its purpose. In mathematics, it means recognising how a simple situation can be represented mathematically and using that representation to solve it.
In 2025, one in five adolescents across OECD countries fell below that basic level simultaneously in science, mathematics and reading. In 2022, the figure was 16%. We are not talking about students who are unable to solve advanced equations or interpret especially complex literature, but about young people who reach the age of 15 without mastering some of the tools they will need in order to continue learning.
And we can go back even further.
In Spain, PIRLS assesses reading comprehension in the fourth year of primary school, when children are around nine or ten years old. In 2021, Spanish students scored an average of 521 points, below the 533-point average of participating OECD countries and the European Union average of 528. Five years earlier, Spain had scored 528.
It is a crucial age. During the first years of school, much of the effort goes into learning to read. Later, reading becomes a tool for learning almost everything else. Weak comprehension then begins to affect science, history, mathematics or any subject that requires students to extract information, connect ideas and follow instructions.
Gaps do not wait patiently for students to resolve them. They accumulate.
In 2025, average OECD performance fell again in reading and mathematics, reaching the lowest levels ever recorded by PISA. If we extend the comparison back to 2015, students have lost 28 points in reading and 22 in mathematics.
Spain: Reaching 15 with Learning Gaps
Ten years after starting primary school, PISA allows us to see how much of that accumulation remains. And here, the results published in 2025 paint a worse picture than the one we had until now.
Spain is not an exception among developed countries, but this time it ranks below the OECD average in all three traditional domains: reading, mathematics and science. Compared with 2022, results have declined in mathematics and reading and remained broadly stable in science. Compared with 2015, they are among the lowest levels Spain has recorded across the entire PISA series.
The change becomes clearer when we look at how many students reach the minimum proficiency level. In mathematics, 67% do so. Put another way, one in three Spanish 15-year-olds cannot consistently perform tasks such as recognising, without direct instructions, how to represent a simple situation mathematically, comparing the distances of two alternative routes, or converting prices from one currency into another.
In reading, 68% reach the baseline level. Again, roughly one in three students falls below it. The threshold requires students to be able to identify the main idea of a moderately long text, locate information according to explicit criteria, and reflect on its purpose and form when asked to do so.
This represents a notable deterioration compared with the picture offered by PISA 2022, when 27% were below the baseline level in mathematics and 24% in reading. But the perspective over a decade is even more revealing: since 2015, the proportion of Spanish students with low performance has increased by 11 percentage points in mathematics, 15 in reading and six in science.
We are not simply looking at a decline in average scores. The group approaching the end of compulsory schooling without having consolidated the learning needed to support what comes next is growing.
Here too, where each student starts matters. In science—the main domain in PISA 2025—students in the most socioeconomically advantaged 25% outperform those in the most disadvantaged 25% by 71 points. It is a considerable gap, although smaller than the OECD average of 85 points. Spain’s gap has narrowed over the past decade, but there is an important qualification: it has done so largely because performance among advantaged students has worsened more than because that of the most vulnerable has improved.
PISA 2025 therefore leaves an uncomfortable picture. Spain maintains basic proficiency levels close to the OECD average, but does so within an international context that is itself deteriorating. Being close to the average offers some comparative reassurance; it says much less about whether students are acquiring the knowledge and abilities they will need later on.
The OECD’s adult skills assessment, PIAAC, offers a glimpse of the next stage. Spanish 16- to 24-year-olds scored below the OECD average in 2022-2023 in literacy, numeracy and adaptive problem solving. They are not the same students who took part in PISA, and the data do not allow a causal line to be drawn from one assessment to the other. But both snapshots point in the same direction: completing the expected years of schooling does not necessarily mean having consolidated all the learning those years are assumed to provide.
Latin America: The Gaps Begin Much Earlier
In Latin America and the Caribbean, the same story takes on another dimension. UNESCO’s ERCE 2019 assessed more than 160,000 children in 16 countries in the third and sixth grades of primary school. Its results, obtained before the pandemic, showed that the region had been virtually stagnant for years: between 2013 and 2019 there was no significant progress in the regional average.
Already in third grade, around 40% of students failed to reach the expected foundational competencies in reading and mathematics. By sixth grade, the proportion was around 60%. In mathematics, only 17.4% of sixth-grade students reached at least Level III, which involves being able to solve problems using several operations, interpret tables and graphs, or work with simple fractions.
The World Bank uses another particularly graphic measure: learning poverty, which identifies children around the age of ten who are unable to read and understand a simple text. Before the pandemic it affected 52% of children in Latin America and the Caribbean; subsequent estimates placed the risk at around 79%.
PISA 2025 now makes it possible to see what happens a few years later. Thirteen countries in Latin America and the Caribbean took part in this edition: Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Mexico, Paraguay, Peru, the Dominican Republic and Uruguay. None reaches the OECD average in all three assessed areas.
Nor is the post-pandemic recovery clearly visible across the region. Brazil remains practically stagnant compared with 2022; Chile declines in mathematics and reading; Mexico records its lowest mathematics score since 2006; and Argentina posts its worst result in that subject since it began participating regularly in PISA.
The percentages give an idea of the distance involved. In Mexico, only 30% of students reach the baseline level in mathematics; in Colombia, 29%; in Argentina, 24%. In Chile, which remains among the region’s best-performing systems, 41% reach it. The OECD average is 65%.
Rather than producing a ranking of countries, these figures reveal the scale of the problem. In several of Latin America’s largest education systems, more than half of young people reach the age of 15 without confidently handling the mathematical tools that PISA considers necessary to deal with simple situations in everyday life and continue learning.
There is also an important nuance. A decline in average scores does not always mean that an education system is performing worse with the same students. Colombia and Costa Rica, for example, have increased over the past decade the proportion of adolescents who remain in secondary education. That means PISA is now assessing young people from historically more marginalised groups who might previously have left school before the age of 15. The OECD itself warns that part of Colombia’s apparent deterioration in mathematics and reading is related to that greater inclusion.
This distinction is important for interpreting the broader story. Keeping more students in school is progress. The next problem appears when being in the classroom does not guarantee that they will acquire the learning they need in order to keep moving forward.
Being Prepared Begins in Primary School
It is important to avoid an overly easy conclusion. The data do not allow us to say that an entire generation is less capable than those before it, nor do they point to a single cause behind the phenomenon.
The pandemic caused significant learning losses, but PISA 2025 reinforces something that was already beginning to emerge in 2022: the trend began earlier and has continued since. Across the OECD, reading performance has fallen by 28 points since 2015 and mathematics by 22. The explanation will therefore also have to be sought in deeper changes in the school experience and in adolescents’ lives.
The professors interviewed by The Atlantic also point to changes in reading habits, the growing use of excerpts rather than complete works, and difficulties sustaining attention over long texts. PISA 2025 also brings the role of technology into the discussion: the OECD finds associations between certain intensive uses of digital devices and poorer results, although correlation alone is not enough to show that screens are causing the decline. The expansion of secondary education, curricular differences, teacher shortages in some systems, and schools’ capacity to detect and address unconsolidated learning all form part of a much more complex picture.
What the different assessments do allow us to observe is the continuity of the problem. PIRLS and ERCE identify difficulties in primary school. PISA finds them again at age 15. PIAAC later reveals deficits among a share of young adults. They do not follow the same students and therefore cannot trace a causal line from age ten to age twenty, but the snapshots taken at different stages are difficult to ignore.
Foundational learning is also cumulative. A child who does not understand well what they read at age ten will find it harder to learn from texts at twelve. Someone who has not mastered certain numerical relationships and basic operations will find algebra more difficult; and without algebra, university-level calculus becomes much more distant.
Universities are making the problem visible because they sit at the end of the journey. But waiting until the first year of university to discover it means arriving very late. PIRLS, ERCE and PISA figures show that many of these difficulties emerge years earlier. Detecting them does not appear to be the main problem. Ensuring that a student does not keep moving from one grade to the next while those gaps accumulate is much harder. After all, arriving at university prepared begins many years before setting foot there.
