The first PISA 2025 results have been out for a while now. Two weeks later, there are still a few puzzles in them that keep occupying my mind. I have already discussed one of them with several people over the past few days. However, I thought I would share my musings here as well.

I have to admit I hesitated a little before writing this, because PISA critics may immediately use what follows to cast doubt on the PISA results as a whole. But I hope people will be intellectually honest enough to realise that, in light of PIRLS and other results, what I am about to discuss can never be the only explanation for the declines we have seen.

PISA does not only ask students to complete all kinds of tasks. Afterwards, participants are also asked how much effort they actually put into the test. And according to OECD education director Andreas Schleicher, never before have so many 15-year-olds indicated that they did not really put that much effort into PISA.

On average, students across OECD countries rated the effort they put into PISA 2025 at 7.16 out of 10. What is perhaps even more striking is that almost three-quarters of students across OECD countries say they would have tried harder if their PISA performance had counted towards their school grades. This is also a trend. The amount of effort students report putting into PISA has been declining just as the results. But there is more you need to know to understand my personal puzzle.

PISA 2025 also includes, for the first time, a separate domain called computational problem solving. How this is measured looks quite different from the more traditional PISA test. In this part of the test, students work on longer interactive tasks. For example, they can experiment with models and programming tools. They will also receive feedback and can adjust their approach based on that feedback. To put it in very unscientific terms, I think it simply looks like more fun than the more traditional PISA questions. Incidentally, the OECD itself emphasises the importance of motivation and task engagement in these tasks.

And this is where Flanders, the Dutch-speaking part of Belgium, becomes an interesting case. Flemish students score 523 points on computational problem solving, placing Flanders among the best European regions for this theme. Within the EU, only Estonia scores higher. Additionally, more than a third of Flemish students reach proficiency level 5 or 6, the levels of the top performers.

What makes this even more intriguing is that computational problem solving is not disconnected from the traditional PISA domains. E.g., in Flanders, performance in this new domain is strongly correlated with performance in reading, mathematics, and science, and these relationships are even stronger than the averages across EU and OECD countries. At the same time, this means that we perform extremely well in this new domain. On the other hand, our performance in the traditional domains has been declining over time. In the past, we used to be as great in math and reading.

And that brings me to a somewhat uncomfortable hypothesis. What if part of the story, and please note that I deliberately write part, is not simply that our young people are less capable? What if their performance has become more dependent on how much a task engages them?

To be absolutely clear: I’m not stating that the PISA data is demonstrating this. It doesn’t. What we know is that reported effort on PISA has been declining internationally. We also know that the new domain of computational problem solving is designed quite differently and is much more interactive. In addition, we have at least one intriguing case, Flanders, where students perform remarkably well in this new domain while their performance in the traditional PISA domains has been declining over time. I do think other countries could have the same situation on hand.

But… putting A, B and C next to each other does not mean that A or B causes C.

There are plenty of other possible explanations I can come up with:

• The skills measured by computational problem solving simply align better with what young people today can do.
• Maybe experience with digital applications matters.
• Differences in curricula still play a role, despite the OECD’s attempts to minimise their influence.
• …

There are probably other possibilities I have not even considered.

In recent years, declining PISA results have often been summarised as a decline in knowledge and skills. I agree that there is good reason for that. The negative trends appear across different countries, domains and measurement points, and they are consistent with results from other assessments, as I mentioned earlier. We cannot simply dismiss them as mere measurement noise. However, at the same time, the amount of effort students say they put into the test is also declining.

Perhaps, then, we are measuring two things that have become entangled: what young people can do and how much effort they are willing to invest in showing us what they can do. I’m pretty sure even that is still too simple. The strong results in computational problem solving raise a third possibility. It is possible that this willingness to invest effort may itself depend partly on the kind of task we put in front of students.

If that is the case, it would make little sense simply to dismiss declining test effort as an annoying methodological problem for PISA. Instead, if young people are indeed less willing to persist with tasks that they don’t care for, that may itself represent a relevant change. Education, and society more broadly, inevitably contain things that are not always fascinating. But if those same young people suddenly perform rather well when faced with other kinds of cognitively demanding tasks, that is at least as relevant.

For now, then, I mainly have a question rather than an answer. But I know that many clever people read this blog. Perhaps some of them have further insights to share.