We all start life from unequal positions. James Heckman once argued that being born is the greatest source of inequality. And, in a way, education is one of our attempts to undo some of this. Now suppose children receive a good preschool education at age 4 and, as a result, build an advantage over children who do not attend preschool. A few years later, you measure the children’s learning and development again. This raises a question that is more difficult than you might think: what would you hope to see?

Perhaps the answer is that the advantage is still there. Perhaps it has even grown. But if the difference had disappeared, that would quickly sound like bad news. The original effect of preschool education would apparently have ‘faded out’.

But… there is another possibility. Perhaps the first group did not fall behind at all. It could also be that the other children simply caught up with them, at least partly. That distinction is not always easy to make in research into the long-term effects of preschool education. A new US study by Kaitlyn Mumma and colleagues nicely illustrates why, and does a few other things besides.

A large sample (but not as large as the press release suggests)

The researchers followed no fewer than 27,814 children (interestingly, the press release mentioned 400,000 children) who, at age 4, had attended various forms of preschool or childcare. Most of the children in this sample came from low-income families. A large group attended public-school pre-K, while others went to centre-based childcare or were cared for by family childcare providers. The researchers then looked at how these children were doing several years later, in third grade.

Let me immediately clear up a possible misunderstanding. The study does not show that children’s preschool experiences ultimately do not matter. This is not an anti-preschool finding. Quite the opposite.

Children who had attended public-school pre-K still had, on average, better reading and mathematics scores in third grade and also achieved better school grades than children who had attended centre-based or family childcare. The differences were not large, but they persisted after the researchers had accounted for, among other things, poverty and cognitive skills measured when the children were four.

So far, you might summarise the study as follows: differences in the preschool years can still be visible years later. But the researchers were particularly interested in what happens after preschool.

What if the primary school is good too?

The team examined whether the differences between the groups depended on the quality of the primary school the children subsequently attended. To do this, they used the official ratings given to schools in this US state. Importantly, ‘quality’ here was determined largely by test performance and pupils’ progress, which is quite a limitation, imho.

You might now expect a good preschool and a good primary school to reinforce each other. An early advantage would, as it were, keep receiving fresh fuel, allowing it to persist and perhaps making the differences even larger. This resembles what the literature calls dynamic complementarity, an idea strongly associated with economist James Heckman: investments at different stages of development can reinforce one another. Early investments make later investments more productive. Yes, Heckman is an economist.

But… reality turns out to be more complicated. For mathematics and overall school grades, the quality of the later primary school made no significant difference to the size of the original preschool advantage. For reading, however, the researchers found something different. The gap between children who had attended public-school pre-K and comparable children from centre-based childcare became smaller as the quality of the primary school increased. You read that correctly: not larger, but smaller.

A small effect

Mind you, the effect was quite modest. Among children from poorer families, the original reading advantage was around 0.08 standard deviations. Moving up one step in school quality was associated with a reduction in that gap of around 0.03 standard deviations. So, to be clear, the original advantage did not disappear. But the direction of this development is fascinating.

One possible explanation is not that children who attended public-school pre-K learned less once they entered primary school. It could also be that better primary schools are more successful at helping children who start from behind to catch up. I am afraid I may have lost you by now because there are quite a few steps involved. But here is the point: the better later education is at reducing differences between children, the smaller an earlier advantage may appear in the long term. And that really is not bad news, not even for preschool education.

When is fade-out good news?

The researchers prefer not to speak of fade-out, but of convergence. Fade-out suggests that something children had learned gradually disappears. But when, years later, you simply compare the performance of two groups, you cannot actually conclude that this is what happened.

Take an extreme example. Suppose group A scores 60 after preschool and group B scores 50. A few years later, both groups score 80. The original difference has completely disappeared. But both groups have clearly learned.

There is quite a difference between concluding that ‘the benefits of preschool have disappeared’ and that ‘the other children have caught up’. Among children who had been in family childcare at age four, the researchers even see a striking pattern that fits this idea. At lower-performing primary schools, these children remain clearly behind the other groups, while at the highest-performing schools the difference virtually disappears. One important caveat: this group consisted of only 266 children, which is really too small to test the key interaction properly. More research needed…

Not a refutation of Heckman

You could read all of this as a correction to Heckman. That was actually my first thought, but I would no longer call it a refutation. This study simply is not suited to that. And I can see several reasons why.

First, this is observational research. Children were not randomly assigned to the different forms of preschool education or childcare, if that would even have been ethically possible. In addition, as I mentioned earlier, the measure of primary-school quality is rather crude.

The researchers also know relatively little about how good the different preschool settings actually were. They know the type of provision, but they do not have a direct measure of its quality. Nor does their dataset contain the children’s complete childcare history. And, perhaps unnecessarily to add, the research was conducted in one very specific US context. We cannot simply assume that the findings apply to the UK, the Netherlands, Flanders or elsewhere.

At the same time, the study makes something visible that can easily get lost when we talk about the long-term effects of educational interventions. We tend to regard a difference that is still present years later as evidence of success. And when that difference disappears, we quickly start talking about fade-out.

But perhaps we should always ask a second question: why did the difference disappear?

Because a difference can disappear when an earlier advantage is lost, but just as easily because later education succeeds in helping the other children catch up.

And those are two completely different stories.