A TAO webinar with 2E examined Spain’s results beyond the national average and considered what the findings mean for equity, digital habits, grade repetition and the future of assessment.
Spain’s results require context
PISA results tend to trigger a familiar cycle of headlines, rankings and comparisons. The webinar From Data to Decisions: What PISA 2025 Means for Spain took a different approach. Hosted by TAO with 2E Estudios, Evaluaciones e Investigación, the session examined what sits behind the national scores and how education leaders can turn the findings into practical priorities.
Spain recorded average scores of 477 in science, 457 in mathematics and 451 in reading. Each result was below the corresponding OECD average of 482, 463 and 461. Between 2022 and 2025, science performance fell by seven points, a change the OECD reports as not statistically significant, while mathematics fell by 16 points and reading by 23 points. OECD profile for Spain
The proficiency distribution makes the policy challenge more concrete. Around one in four students performed below Level 2 in science, while roughly one in three did so in mathematics and reading. At the other end of the scale, only a small share reached the highest proficiency levels. These figures point to a dual requirement: reduce the number of students who have not yet secured foundational skills and create more opportunities for advanced performance.
The national average also conceals meaningful regional variation. Several autonomous communities achieved stronger results across science, reading and mathematics despite differing socio-economic profiles, migration patterns and repetition rates. The webinar did not present these differences as a league table. It argued that high-performing regions should be studied for practices that may be useful elsewhere, while recognising that no single policy can be transferred without considering local conditions.
Associations are signals rather than proof of cause
PISA combines cognitive assessments with questionnaires about students’ backgrounds, attitudes and experiences. That contextual information helps explain how performance differs across groups, but it does not establish causation. The speakers repeatedly made this distinction when discussing the factors most closely associated with Spain’s results.
The strongest positive associations included students’ understanding of how scientific knowledge is built and revised, environmental awareness, enjoyment of science, socio-economic and cultural status, and expected occupational status. Students who understood that scientific claims depend on evidence and can change with new findings also tended to perform better across science, reading and mathematics. The result suggests that scientific reasoning is relevant across the curriculum, rather than being confined to subject knowledge.
Socio-economic status remains a major dividing line. The gap in science between students in the highest and lowest socio-economic quarters reached 71 points in Spain, compared with 84 points across the OECD. Spain’s gap is smaller than the OECD average, but it is still substantial. Family context does not determine an individual student’s result, yet it continues to shape the resources, expectations and learning opportunities available to them.
Digital habits need a more precise policy response
One of the webinar’s most discussed findings concerned students’ use of digital technology for leisure. Frequency of ICT use during the week and at weekends appeared among the strongest negative associations with performance in the analysis presented by 2E, behind grade repetition.
The detail matters. Thirty-five percent of students reported spending more than three hours a day on social networks, while half said they did not create or edit their own digital content. The webinar also compared changes in leisure screen use with changes in science scores across autonomous communities. Regions where recreational screen use increased often recorded weaker science performance, while some regions where it fell improved their scores.
This pattern does not show that screen use caused the score movements. It does, however, justify closer attention to what students do online, how long they do it and what those activities displace. A blanket debate about whether technology is good or bad misses the operational issue. Schools and families need to help students regulate recreational use while building the capacity to search, evaluate, create and solve problems with digital tools.
The broader OECD findings support this distinction: moderate educational use can be associated with stronger performance, while excessive use and digital distraction are associated with weaker outcomes. Read the OECD PISA 2025 report
Grade repetition remains a structural concern
Grade repetition showed the strongest negative association with performance in the webinar analysis. Twenty-two percent of participating students in Spain had repeated at least one grade, compared with 12.9 percent across the European Union and 7.9 percent across the OECD. The score gaps between repeaters and non-repeaters approached 100 points in science, reading and mathematics.
Those figures should be interpreted carefully. Students who repeat a grade are more likely to have experienced learning difficulties beforehand, so the gap cannot be attributed to repetition alone. The policy implication raised in the webinar was narrower and more practical: repeating the same year with the same methods is unlikely to solve the underlying problem on its own.
The speakers argued for earlier and more individualised support, including timely identification of learning gaps, tutoring, methodological adjustments, work with families and support within the classroom. Repetition may remain appropriate in specific cases, but it should follow stronger preventive measures rather than operate as a routine response.
Gender gaps can move
The gender results provided evidence that performance gaps are not fixed. Spain’s science gap fell from seven points in 2015 to zero in 2025. Girls continued to lead in reading by 26 points, while boys led in mathematics by 14 points. The mathematics gap narrowed slightly from 16 points in 2015, although the reading gap widened from 21 points.
The closure of the science gap matters because it shows that differences can change over time and across domains. The next step is to identify which classroom practices, expectations and student attitudes contributed to that movement, and whether they can inform work in mathematics and reading.
Assessment should connect knowledge with application
A question at the end of the webinar asked what should change in educational assessment beyond the goal of raising scores. The response was direct: assessment should continue to measure curricular knowledge, but it should also show whether students can apply that knowledge in unfamiliar situations. More competency-based tasks can reveal reasoning, interpretation and problem-solving that conventional recall questions may not capture.
This has implications for schools as well as national systems. PISA for Schools was discussed as one way for individual schools to compare their own performance and contextual data with national and international reference points. The value lies in connecting evidence to a specific improvement plan, not in producing another ranking.
The delivery model expanded what PISA could measure
Miguel Prieto, Vice President of Corporate Strategy at Open Assessment Technologies, then examined the digital infrastructure behind PISA 2025. TAO supported more than 900,000 assessment sessions in 54 languages, including right-to-left scripts. More than 800,000 sessions were delivered online and over 100,000 through an offline model, with peak concurrency of around 20,000 users and fewer than one percent of sessions reporting technical issues.
The online and offline models were part of the same controlled platform. Where connectivity was limited or local authorities wanted to review data before synchronisation, schools could deliver the assessment through a local server and send the completed sessions to the central system later. This allowed a common assessment design to operate across very different infrastructure conditions.
Centralised delivery also reduced the logistical dependency on physical media used in earlier cycles. QTI structured the assessment content for portability and reuse, while LTI supported interoperability with connected systems. Shared governance across technology, content, localisation and analysis partners kept countries informed throughout delivery.
The platform also supported capabilities that paper could not reproduce easily. These included multistage adaptive pathways, foreign-language tasks with listening and speaking components, and event-level interaction data that captured more than the final response. Such data may help researchers study how students approach a problem, provided that its interpretation is valid, transparent and governed responsibly.
Priorities after PISA 2025
The webinar translated the findings into five priorities for education leaders and assessment programmes:
- Strengthen preventive support before learning gaps lead to repetition.
- Teach digital self-regulation alongside purposeful digital learning.
- Direct resources towards students and schools facing the greatest barriers.
- Use assessments that combine knowledge with application and capture richer evidence of learning.
- Build assessment infrastructure around interoperability, flexible delivery and clear governance.
PISA 2025 does not provide Spain with a single policy prescription. It provides a stronger diagnostic base. The national scores show where pressure is building, while the contextual data identifies groups, behaviours and system choices that deserve closer investigation. The practical task is to use that evidence early enough to change students’ experience, rather than waiting for the next cycle to confirm the same pattern.
For assessment leaders, the delivery itself carries a second lesson. Large-scale digital assessment can preserve comparability while operating across languages, connectivity conditions and institutional structures. The next stage will demand more sophisticated adaptive design, a broader range of authentic tasks, responsible use of AI, sustainable operating models and continued attention to equity.
Continue exploring PISA 2025
Explore the PISA 2025 Hub for upcoming webinars, expert analysis and resources on the future of assessment.