PISA 2025: CELE researchers call for stronger evaluation of school reforms
Sweden’s PISA results have fallen sharply in reading and mathematics. In an opinion piece in Svenska Dagbladet, researchers from the Center for Educational Leadership and Excellence (CELE) argue that identifying possible explanations is not enough: Sweden also needs to become better at evaluating which school reforms actually work.
The latest results from the OECD’s Programme for International Student Assessment (PISA), released on September 8, show a continued decline in the performance of Swedish 15-year-olds.
Between 2022 and 2025, Sweden’s average score fell by 21 points in reading and 18 points in mathematics, to 466 and 464 points respectively. Science performance fell by eight points to 485, although that change was not statistically significant. Swedish students now perform close to the OECD average in all three subjects, and the results are among the lowest Sweden has recorded since participating in PISA.
The decline is not unique to Sweden. Across OECD countries, average performance in reading and mathematics also fell, reaching the lowest levels recorded by PISA. OECD data nevertheless show that Sweden’s decline between 2022 and 2025 was larger than the OECD average in both reading and mathematics.
Looking beyond individual explanations
In an opinion piece published in Svenska Dagbladet following the release of the results, CELE researchers Ema Demir, Ebba Henrekson, Mats Jutterström, Axel Norgren, Rebecka Persson, Abiel Sebhatu, and Karl Wennberg argue that the new figures should prompt a broader discussion about how Sweden approaches school improvement.
The researchers point to several factors that may matter for student outcomes, including segregation, language conditions, changing reading habits, digitalization, student absenteeism, teaching quality, and school leadership and governance.
At the same time, they caution against treating a list of possible explanations as evidence of what is actually driving the decline.
The results also suggest that the challenge extends beyond traditionally disadvantaged groups. According to the researchers, performance has fallen substantially even among students from more advantaged socioeconomic backgrounds. Socioeconomic background nevertheless remains strongly associated with achievement; OECD data show that it accounted for 12% of the variation in science performance in Sweden in PISA 2025.
Building evaluation into school reform
The central argument of the opinion piece is that Sweden does not lack school reforms. Rather, major changes are too often introduced without simultaneously creating sufficiently strong conditions for determining whether they work, for whom, and under what circumstances.
The CELE researchers argue that evaluation and learning should therefore be built into reforms from the beginning. This could include testing interventions on a smaller scale before national implementation, making better use of school data, improving systematic data collection, following outcomes over time, and examining why some schools and school providers achieve better results than others.
Their argument is not that PISA itself can provide the answers. PISA can identify patterns, trends, and differences between student groups, but it cannot determine which policy response Sweden should choose.
Instead, the researchers argue that the education system needs a stronger capacity to learn from the reforms and interventions it implements.
From measuring problems to learning what works
For CELE, the results raise a broader question: How can Sweden build an education system that becomes better at identifying problems, understanding them, and acting on evidence?
The researchers argue that this requires systematic evaluation not only within schools and municipalities, but also among government agencies and at the political level. Major reforms should be designed from the outset with clear plans for how their effects will be evaluated and against which objectives.
When the next PISA results are released, they argue, Sweden should be able to assess not only whether student performance has changed, but also what has been tried in the intervening years - and which changes actually made a difference.
Read more
Read the opinion piece “Skolan behöver mer än nya förklaringar” in Svenska Dagbladet.
