Hattie’s effect sizes still tend to pop up in my blog posts and talks, though usually with the necessary caveats. By now, however, I’ve come to realise that his idea that anything above .40 represents added value has been leading us astray for years. A recent study by Vera Busse and colleagues on writing instruction in primary education is a good example of this. If I tell you the effect size is .11, you might already be inclined to tune out. But do keep reading. There is much more behind that number than you might think.

This is not a lab experiment under ideal conditions, but simply teachers working in their own classrooms, with real students, real time pressure, and all the messiness that comes with it. Exactly where it actually matters. And in those contexts, effect sizes tend to be smaller, but also far more realistic.

What did the teachers do? Nothing particularly extraordinary. On the contrary, they built on what we have known from research for quite some time:

• Writing as a process, not a final product.
• Letting students plan, write, receive feedback, and revise.
• Teachers thinking aloud and showing how to construct a text.
• Working with examples.
• Providing targeted feedback. Sometimes writing together, sometimes individually.

In other words, these teachers showed what evidence-informed writing instruction can look like. The control group, by the way, did not do nothing. They simply taught as usual. And that matters. In that “as usual” there are often already elements we know can work: a writing task, some attention to structure, occasional feedback. Such elements are rarely entirely absent from writing instruction. What happened in the intervention group was different. Those same principles were used explicitly, systematically, and in combination: planning, modelling, writing, feedback, and revising. Not as isolated ideas, but as a coherent approach. So the difference is not between poor and good teaching, but between more intuitive practice and deliberate, evidence-informed practice.

And the result? Yes, it works. Students write better. The quality of their texts improves, and that improvement persists for a while. But, and this is important, the effect is relatively small. No spectacular leaps. No sudden miracle. Just: a bit better. The control group did not stand still either, but the experimental group improved in a slightly more focused way.

And that is precisely where things often go wrong in how we look at education. We have somewhat convinced ourselves that “what works” must automatically mean “works a lot.” That an evidence-informed intervention should produce large effects. When that is not the case, it almost feels like a disappointment. But that expectation is misplaced.

Writing is complex. It requires students to think simultaneously about content, structure, language, spelling, and their audience. That does not suddenly become easy because you change one approach. What this study shows is that consistent, well-thought-out support does help. But step by step.

There is a second important lesson. The intervention consisted of multiple components at once: planning, modelling, feedback, and collaboration. That makes it powerful, but also less straightforward. We do not know exactly which component makes the difference. But it is likely the combination. Good writing instruction is not a trick, but a coherent whole.

Moreover, implementation was not straightforward. Not every teacher used all components. Time, experience, and familiarity play a role. That may well be the most underestimated factor in educational improvement: what works on paper still has to land in the classroom.

The researchers also looked at motivation. That did not increase dramatically in the experimental group. In some cases, it even declined slightly. That, too, is not surprising. Getting better at something difficult does not automatically make it more enjoyable. Sometimes it simply makes the difficulty more visible.

So what should you take away from this? Perhaps mainly this: stop looking for the one intervention that solves everything. Good writing instruction does not lie in a single trick, but in a set of fairly robust principles we have actually known for quite some time. Let students plan. Model your thinking. Work with examples. Provide targeted feedback. Take time for revision. Not spectacular, but well considered.

Just don’t expect miracles.