This post is unique on this blog. Yesterday, I simply could not find anything ‘bloggable’ in my RSS feeds. There was nothing among my Scholar suggestions and nothing on EurekAlert that I had not already blogged about. So, in the end, for the first time I asked AI whether there were any new studies that might be useful. I got four suggestions. One study I already blogged about. Two seemed less relevant to me. But there was one I would like to discuss with you. The study is about peer and self-assessment in maths lessons, and there are some rather practical aspects to the story! Oh, and it turned out I had overlooked this study in my RSS feed!

The essence is actually quite simple: peer assessment and self-assessment sound fairly straightforward at first. Have pupils or students assess each other’s work, or ask them to look critically at their own work. This way, they not only receive feedback, but also learn to think about what good and less good work actually looks like. Both approaches can lead to more learning. But… an effect size tells us that something can work. The next question is more important: how?

The study by Karolin Maskos and colleagues, published in Teaching and Teacher Education, is interesting for precisely that reason. The researchers conducted a randomised study in Switzerland involving 96 teachers and 1,525 pupils in Grades 4 and 5, across 53 schools. The classes were divided into three groups:

• peer assessment,
• self-assessment,
• control group.

What happened next was quite substantial. The teachers in the two intervention groups received eight hours of professional development spread over three weeks. The training focused on two things: multiplication and division, and how to use peer or self-assessment effectively when teaching these topics. Afterwards, the teachers taught around twelve lessons on this content and used the approach from the training in at least six of them.

Importantly, teachers in the control group taught the same mathematical content and could also use peer or self-assessment. What they did not receive was the training or the accompanying materials. So you could say that this study is actually also about professional development.

And that professional development was strongly subject-specific. Teachers explored different strategies pupils can use to solve multiplication and division problems, examined typical mistakes, and considered how to make different solution strategies visible. They watched videos of pupils assessing each other’s solution methods and of teachers leading classroom discussions about different strategies. They also worked with an interesting principle: give pupils fewer problems, but ask them to solve the same problem in different ways. This principle becomes important later in the study.

In all of this, the researchers divided the teacher’s role into four steps:

• planning,
• instructing,
• supporting,
• using.

Following these steps, the teacher selects suitable tasks and thinks in advance about possible solutions, makes learning goals and assessment criteria clear, supports pupils while they are assessing, and then uses what emerges to take the learning further with the whole class.

In practice, this meant that pupils were encouraged to compare solution strategies, explain them, and experiment further with them. On the one hand, in peer assessment, they compared their approach with those of other pupils. In self-assessment, they did this, among other things, by comparing their work with worked examples.

These different approaches also formed the basis for measuring the intervention’s effect. The researchers gave the participating pupils six multiplication and division problems to solve. For each problem, they could provide up to three different solution methods. The researchers then looked at the number of different correct strategies a pupil could use. So they did not just look at whether pupils arrived at the correct answer, but also at the breadth of the repertoire they could use to get there.

Before the intervention, there was hardly any difference between the three groups. After the series of lessons, however, pupils in both intervention groups used more correct solution strategies than pupils in the control group. Not entirely surprising, of course, but after adjusting for their previous performance, the difference corresponded to an effect size of d = 0.36. More importantly, five weeks later the effect was still present: d = 0.29. The researchers conducted several sensitivity analyses, which confirmed this picture.

To put this more concretely, before the intervention only around four to five per cent of pupils in all groups could provide, on average, more than one correct strategy per problem. After the intervention, this had risen to 27.9 per cent in the self-assessment group and 26.9 per cent in the peer-assessment group. In the control group, by contrast, it was 15.9 per cent. You can see that the choice between the two forms of assessment made little difference.

You could easily conclude from this study that we should have pupils assess each other’s work more often. But here comes my nuance again. I do not think that is quite what this study shows. The intervention was a complete package. Teachers did not only learn about peer or self-assessment. At the same time, they received input on mathematical solution strategies, clear learning goals and criteria, diagnostic questions, worked examples, scaffolding and leading classroom discussions. They were also given concrete materials. What was measured was the effect of this entire package. The researchers therefore do not know which element, or which combination of elements, was responsible for the effect they found.

But one thing can be concluded, and that is the title of this blog post. With these forms of assessment, the teacher clearly remains crucial. Peer assessment is sometimes jokingly portrayed as if the teacher were simply handing over part of their work to the pupils. Instead of giving feedback yourself, the pupils do it for each other. OK, I admit that I once thought this myself as a pupil. Sorry, Mr Standaert. But in this overall package, the teacher actually had an enormous role. This is also in line with research on peer tutoring, where it is precisely the structured approach of the teacher that makes a difference.