“Yes, but the context is different.” It is a response I hear regularly and one I sometimes use myself when discussing research from another country. A study from England? Different context. A study from the United States? Different context. I sometimes hesitate to discuss research from China or Japan for the same reason. But before you know it, every research finding starts to look like a local curiosity with little relevance elsewhere. In fact, I described this very issue during a meeting this week as the Achilles’ heel of evidence-informed education.

At the same time, the opposite happens just as often. A finding from a single study or a single country is quickly translated into a general recommendation for everyone. That does not feel quite right either.

A new study in Studies in Educational Evaluation attempts to better understand this tension. Zin Oo and colleagues used data from PIRLS, the international study of reading literacy among fourth-grade students. Rather than focusing on one country, they examined dozens of countries across three assessment cycles: 2011, 2016, and 2021. They then asked a simple question: if a relationship is found in one country, how likely is it that we will find the same relationship elsewhere?

The answer turns out to be nuanced. Were you expecting anything else? Some relationships are remarkably stable. Socio-economic status continues to be associated with reading achievement in virtually every country. The same is true for reading confidence. Students who feel more confident in their reading abilities tend to perform better. The strength of these relationships varies from country to country, but the direction remains strikingly consistent.

Other factors behave quite differently. Enjoyment of reading, engagement during reading lessons, bullying, and school discipline appear to be much more context-sensitive. In some countries, the relationships are positive; in others, they are weaker or almost absent. What seems to be associated with better reading performance in one country may matter much less in another.

That is what makes this study interesting. Not because it tells us something revolutionary about reading, but because it forces us to think differently about context. The question is not whether context matters. We have known that for a long time.

Education always takes place within a specific environment. Schools differ. Teachers differ, even within the same school. Cultures differ, for example, in the expectations that parents have of education. Policy choices differ too, even though countries increasingly look to one another for inspiration. Nobody seriously doubts that.

The real question is much more interesting: For which findings does context matter a great deal, and for which does it matter much less? Because if context truly determines everything, science becomes a rather strange enterprise. We could never learn anything from a study conducted in another school, let alone another country. Every finding would remain trapped within the boundaries of its original setting.

But that is clearly not what this study shows. Some relationships seem to travel remarkably well across the world. Others remain much more closely tied to their local context. So it is not the case that context always determines everything, but it is not irrelevant either. That is why context should not be the end of the discussion.

“Yes, but the context is different” is often not an explanation. It is the beginning of a new question. Which mechanisms are so fundamental that they reappear across very different education systems? Which effects depend on culture, policy, or organisation? And when does a difference in context genuinely change a research finding? These are the questions that make evidence-informed education interesting. The issue is not whether research from another country is useful, but under what conditions it becomes useful.

Or, to put it differently: some insights travel surprisingly well around the world. Others prefer to stay at home. The challenge is knowing which is which.