Tests don’t make pupils better at exams, but they do improve their teachers

There is a never-ending argument about whether we test our children too much or too little. Why can’t we be fluffy like the Nordic countries, with no tests in primary schools? ask some. We need tests to support children and spot schools that are underachieving, say others.

It’s a permanent debate not only because people have different views on the purpose of education but also because analysis can be hard, not least because countries that don’t do much testing (such as Norway) obviously have limited data on pupil progress.

But new Norwegian research found a very clever route to get round the “no data in a low-testing country” problem. The researchers looked back at data from a random control trial originally done to assess small-group maths tuition, but used it for a very different purpose.

Eighty-odd schools got the tuition and the same number didn’t, but they all introduced a detailed maths test for kids aged seven to nine to measure the tuition’s impact.

So pupils in the 80-plus schools who didn’t get the tuition but did get the test can be compared with pupils across all the other schools in Norway – that had neither the test or the tuition – to assess the impact of an early maths test on later maths attainment.

So, are tests beneficial? The answer is yes... but maybe not quite in the way expected. Generally, testing has zero effect on later student attainment, but don’t conclude that all tests are bad. First, screening helps low-performers by ensuring that they’re noticed. Second, despite fears that tests stress out kids, there’s no evidence to suggest that. Instead, they improved teaching practice and student perceptions of teachers’ engagement. So, test for better teachers as much as for better students.

• Torsten Bell is chief executive of the Resolution Foundation. Read more at resolutionfoundation.org