How to put education policy to the test

Text: Richard Fisher
Illustration: Satu Kettunen

 A few years ago, an intriguing leaflet was posted to thousands of families all over Finland. ‘Congratulations!’ it said. ‘Your child is among those who have been selected for the test group for a two-year pre-primary education trial.’

It was an invitation: an offer to join one of the most ambitious and rigorous education experiments ever conducted in the country – one that is now well underway. 

Around 35,000 young children are now taking part in a €30m project called the Two-Year Pre-School Experiment. Its goal is to find out whether starting pre-school a year earlier than usual in Finland – aged five- rather than six-years old – will have positive long-term effects for the children, their families and the economy. 

The scale and design of this project offers lessons for researchers and politicians all over the world. Why? Unlike most education policies introduced by governments, this one will be analysed scientifically before it gets fully rolled out. 

‘It's about the government being willing to test a policy before full implementation,’ says economist Matti Sarvimäki of Aalto University's School of Business, who is one of the experiment's architects. ‘It's a story I love to tell, because nothing stops you doing the same for many other policies. Or other countries adopting similar approaches. If done properly, this approach can really improve public policy.’

A Nobel-winning approach

The roots of the Two-Year Pre-School Experiment go back to 2015. Back then, Finland's newly elected government wondered whether they might begin to test policies with experiments. This led to the internationally recognised Universal Basic Income trial, in which researchers gave 2,000 unemployed people a monthly sum of €560 whether or not they got a job. 

The project was unusually robust not only because of its scale but also because it was designed as a randomised control trial (RCT). In an RCT, participants are randomly assigned to be part of the ‘treatment’ group (in this case, the 2,000 people who got the basic income) or the ‘control’ group (173,222 people who did not.) 

RCT methodology is not new in itself – pharmaceutical trials have been done this way for decades – but applications in social sciences remained few and far between throughout the 20^th century. Since the turn of the millennium, the use of RCTs has increased rapidly in economics – an evolution recognised by the Nobel committee, which awarded the 2019 prize to researchers pioneering the use of RCTs in development economics. 

In a nutshell, using an RCT lets researchers compare a policy's impact and effectiveness against a parallel world where it isn’t introduced. 

Testing education

A few years later, the Two-Year Pre-School Experiment was established to explore a different policy question: what would happen if children in Finland started pre-school a year earlier than usual?

Children enter the education system at different ages across Europe. In Finland, children start pre-school aged six and go to school a year later, but schooling starts earlier in some other countries. Opinions differ about which age is best for a child's long-term prospects and whether investing in the facilities and teachers needed for an earlier start pays off pedagogically and economically.

To find out, the Finnish government passed special legislation funding an extra year of pre-school for thousands of children. Letters were sent to families in randomly selected pre-schools all over the country inviting them to participate; the vast majority did.

Overall, the curriculum content is similar in both treatment and control groups, says Marja-Kristiina Lerkkanen, a professor of education at the University of Jyväskylä who is part of the project. The main difference is that the children in the treatment group have twice as much time with pre-school educators. 

‘Yesterday I talked with a group of teachers, and they told me that having two years gives them peace and more time to support children's development and skills,’ she says. ‘There's no rush and no hurry compared to one-year pre-school education.’

In principle, having two years instead of one should mean more opportunity to get the children ready for comprehensive school. But the whole point of using an RCT is to test that idea with a robust analysis of data rather than relying on theory or anecdotes. Many types of data will be combined and compared: test scores, interviews, surveys, policy documents, background characteristics of the child, their guardians, their teachers and much more.

The results will be published in 2025 and will report on the impact of the programme on children at the start of school, as well as the experiences of the families, teachers and civil servants involved. In particular, the researchers would like to know if starting pre-school one year earlier has greater benefits for children of lower income parents or immigrant families. 

Looking beyond the 2025 report, the hope is to continue following the children and their families over the longer term – in fact, throughout their entire lifetime. 

Registering long-term effects 

Testing a proposed education policy experimentally isn’t the only way the Two-Year Pre-School Experiment stands out. The study is also special because of how Sarvimäki and the team will blend large-scale data collection with information derived from dozens of administrative registers. These data can give them details about the participants and their families that aren’t accessible in most other countries.

People living in Finland have personal ID numbers which makes it possible to (anonymously) connect and correlate information about them across various public services, from education and healthcare to employment and income. 

Years ago, this registry data sat in various, disparate databases, but over the past three decades, researchers in Finland have used it to get powerful insights into the work, health and decisions of people as they navigate their lives. This rich constellation of databases is growing all the time as new information is collected and archive data is digitised.

Spanning decades, the registry data enables long-term – even intergenerational – studies that are more difficult in most other countries. This is helpful because sometimes the consequences of a government policy don't emerge for years (see ‘The long-term effect of a 1960s agricultural policy’). 

In principle, the registry could allow the Two-Year Pre-School Experiment to carry on beyond 2025, as researchers follow the children over many years, through education, into work – and perhaps even for the rest of their lives. Sarvimäki acknowledges that it might not be him and his colleagues that do this – they will have retired or passed away by the time the five-year olds grow old – but their hope is that the next generation will pick up the baton.

The long-term effect of a 1960s agricultural policy

In the late 1960s, the Finnish government introduced a major policy without testing its consequences. To help tackle artificially high food prices and over-productivity, they paid farmers across the country to give up their farms and find other work. 

‘The policy itself wasn't really liked. It seems that the farmers really protested against it, and there was this political backlash in the very short-term election outcomes,’ explains economist Aapo Stenhammar of Aalto University.

But what was the long-term impact? Stenhammar found part of the answer in Finnish registry data, but he soon realised that a lot of the information he needed about the farmers was locked away in 300,000 physical documents in the National Archives. Using machine learning, he digitised all of the Agricultural Census records about the farms, the farmers and their families, and then he linked that up with broader Finnish registry data.

It turned out it was the children of the farmers who experienced the greatest change. ‘The outcomes are very long term, so you can only observe them 20-30 years after the policy,’ says Stenhammar. 

‘The children who were most exposed to this policy actually ended up getting more education, so they were more likely to finish secondary degrees. And they also moved into other sectors from agriculture, so they were more likely to work in offices or managerial positions, and they ended up earning more.'