No Simple Answer: Did Project Follow Through Really Prove That Direct Instruction Is Best?

There are some pieces of educational research that seem to take on a life of their own.

They are quoted, requoted and gradually reduced to a sentence that sounds so convincing that we stop asking what the research actually found.

Project Follow Through is one of them.

You may have recently heard the claim on Twitter. “Project Follow Through was the largest educational experiment ever conducted”. It compared different ways of teaching children and Direct Instruction came out on top. Not just for basic skills, apparently, but for cognitive and even affective outcomes.

Case closed then. Except it isn’t.

Before anyone reaches for their keyboard, I am not arguing against explicit instruction. Of course there are things that we should teach explicitly. I talk about this all the time in all my training and writing. There are times when the most helpful thing an adult can do is explain something clearly, model it, practise it with children and give feedback.

If I want to teach a child how to form a letter, I don’t need to hide the letter in the sand tray and wait for the children to “discover” it, that would be a total waste of everyone’s time!

Sometimes we need to say, “Let me show you.”

But that is a very different proposition from saying that because explicit instruction is effective for teaching particular knowledge and skills, it follows that it is the best way for young children to learn.

And that is where I think we need to look rather more carefully at Project Follow Through.

So what was Project Follow Through? And who on earth thought that was a great name for a research study?

Project Follow Through began in the United States in 1968 and continued through much of the 1970s. It was enormous. It involved children from disadvantaged communities across many different locations and compared a range of educational models.

One of these was Direct Instruction, associated particularly with Siegfried Engelmann.

And let’s be clear. Direct Instruction did well. We don’t need to pretend otherwise.

There is a substantial evidence base showing that carefully sequenced, systematic teaching can be very effective, particularly when we are trying to teach clearly defined knowledge or skills.

But Project Follow Through is often used to make a much bigger claim:

Direct Instruction was tested against alternative pedagogies and proved to be the best way of teaching children.

That’s where things become rather less straightforward. Because It wasn’t quite the experiment we are sometimes told it was. The phrase “largest educational experiment ever conducted” makes Follow Through sound rather like an enormous randomised controlled trial.

But it wasn’t.

Children were not simply randomly allocated to different pedagogical approaches so that researchers could compare like with like. The children, schools and communities involved were not necessarily equivalent at the beginning. Researchers therefore had to try to account statistically for differences between the groups.

That’s not a minor technicality. If groups differ before an intervention begins, establishing that any subsequent differences were caused by the intervention becomes much more difficult. These concerns weren’t invented recently by people who don’t like Direct Instruction, they were raised at the time.

In 1978, Ernest House, Gene Glass, Leslie McLean and Decker Walker published a critique of the Follow Through evaluation with possibly one of the best titles for an educational research paper ever:

No Simple Answer.

They challenged the conclusion that Follow Through had demonstrated the superiority of basic-skills approaches and raised questions about the sampling, classification, measurement and interpretation of the results. Perhaps we should have paid more attention to the title.

Because nearly fifty years later there are people (often with no direct experience of working in a classroom with 30 under 6s) who seem determined to turn an extraordinarily complicated project involving different children, teachers, schools, communities and educational philosophies into a league table of teaching methods. Education is rarely that simple and, despite what those with more limited experience of working in Early Childhood Education (ECE) may think, young children’s learning is never that simple!

What you measure matters

There is another problem. Suppose I have two groups of children. In one, I explicitly teach particular literacy and mathematics knowledge and skills because I know these are going to be assessed.  In the other, children spend considerably more time building, investigating, talking, making decisions, solving problems, pretending, negotiating, moving, creating and exploring. Adults support and extend their thinking, introduce language and knowledge and sometimes teach directly, but much of the children’s learning happens through active experience.

Then I give both groups a standardised test heavily weighted towards the academic knowledge explicitly taught to the first group. Which group would we expect to do particularly well?

Probably the first.

Those children have learned something and we shouldn’t dismiss it.

But what exactly have we proved?

Perhaps that explicit instruction is an efficient way of teaching material that is subsequently explicitly tested. That’s useful but it still doesn’t tell us that it is the best way to educate a four-year-old.

Because what about the things we didn’t measure, but that we know impact on children’s sense of self, self regulation, executive function and long term outcomes?

Curiosity, critical thinking , creativity, persistence, independence, collaboration, problem, solving, language development, the ability to plan, cognitive flexibility, the ability to change the plan when it doesn’t work. The ability to negotiate with another human being who has a completely different plan. These things matter too. In fact Whitebread and Bingham cite many of these skills as being necessary for higher order thinking, and identify that young children can demonstrate many of these when they are involved in pretend play. What we choose to measure inevitably influences what appears to “work”.

Context matters too

There was considerable variation within Follow Through. The same programme did not magically produce identical results wherever it was implemented. Why would it?

Children are unique, teachers aren’t interchangeable delivery mechanisms, communities aren’t laboratories.

Implementation matters, relationships matter, adult knowledge matters. Children’s previous experiences matter, language matters, poverty matters, health matters, housing matters. Family circumstances matter. (see my last blog).

This is where Darbyshire, Finn, Griggs and Ford’s (2014) research is useful, but I want to be careful about what it does and doesn’t tell us.

Their paper, An Unsure Start for Young Children in English Urban Primary Schools, is not a comparison of play and Direct Instruction. It would be wrong to use it as evidence that one works better than the other.

What it does do is challenge simplistic representations of the “achievement gap” and draw attention to the much more complicated realities surrounding children’s lives. That matters because if we define the problem simply as:

These children have lower academic attainment.

It becomes very tempting to decide that the solution must be:

Give them more academic instruction, earlier.

But children and development are more complicated than that.

So what does early-years research tell us?

If Direct Instruction had settled the question of how young children learn best in the 1970s, we might expect subsequent early-childhood research simply to have confirmed it. It hasn’t. Take EPPE and REPEY.

The Effective Provision of Pre-School Education project was a major longitudinal study involving around 3,000 children. It followed children from the age of three to seven and examined both the impact of preschool and the characteristics of effective provision. The related REPEY study looked particularly closely at pedagogy in effective settings.

And what did researchers find?

Not adults standing back while children simply got on with it, but not endless adult-directed teaching either.

Effective settings provided a balance between adult-initiated and child-initiated experiences.

Adults taught. They modelled. They introduced children to new knowledge and language.

But they also observed children, joined their activity, responded to their ideas and extended their thinking.

One of the most important ideas to emerge from this work was sustained shared thinking.

Adult and child thinking together.

“Why do you think that happened?”

“What could we try?”

“I wonder what would happen if…”

“How could we find out?”

That’s powerful teaching. It just looks different to teaching in Key Stage 3 and 4. Because the children have only been alive for 60 months at the most and they need a different approach to the teaching strategies that support 14 and 15 year olds.

And then there’s Perry Preschool

The Perry Preschool study is fascinating because it is frequently cited as evidence of the enormous potential impact of high-quality early education, rightly so!

Its long-term findings are remarkable, and James Heckman and colleagues’ later economic analyses demonstrated substantial long-term returns associated with investment in high-quality early childhood education. But sometimes we talk about Perry as evidence that “early intervention works” without stopping to ask:

What did those children actually do?

The HighScope approach that developed through Perry was based around active learning. Children made choices. They manipulated materials. They solved problems. They talked. They played.

And, crucially, they planned what they wanted to do, did it and then reflected on what they had done.

Plan. Do. Review.

Adults weren’t absent, quite the opposite. Knowledgeable adults observed, interacted, questioned, introduced ideas and extended children’s thinking. But children had agency.

So, one of our most famous pieces of longitudinal evidence about the potential benefits of preschool comes from a pedagogical approach in which active learning, exploration and children’s decision-making were central.

Children need first-hand experience

Some people with limited experience of working with young children seem in danger sometimes of confusing knowing more with being told more.

Roessingh and Bence (2018), writing about embodied cognition, describe the importance of integrating children’s sensorimotor, cognitive and linguistic development. Put more simply, children’s bodies are involved in their thinking. Young children handle, manipulate, carry, combine, move, compare and explore. Language develops alongside those experiences and helps children make meaning from them. Before heavy means very much, children need to have lifted things. Before capacity becomes an abstract mathematical concept, they need to have filled and emptied things. Before children really understand balance, they need to have built things that fall over. Context matters, real life experiential learning matters. Before children can think in the abstract, they need to have experienced the concrete.

Repeatedly. This is why schematic play is so important to children, and they need numerous opportunities to revise and revisit their learning.

A picture of a cube on a whiteboard will not help a young child to understand the properties of a cube. But playing in the blocks will.  A worksheet about capacity will not support the child’s understanding of full and empty if they’ve never filled and emptied buckets and felt the difference in weight .

This isn’t an argument against direct teaching. It is an argument against assuming that because we can tell children something more quickly, telling is necessarily the most powerful way for them to come to understand it.

What about school readiness?

Whitebread and Bingham’s (2012) critical review of the evidence around school readiness is important here too. The phrase “school readiness” sounds perfectly reasonable until we start asking exactly what we mean by it. Too often it appears to mean making younger children’s learning fit into a model which is more suited to older children. Sit for longer, listen for longer, complete more adult-directed tasks.

Children do not suddenly stop being young children because somebody changes the sign on the classroom door from Nursery to Reception. Development doesn’t reorganise itself conveniently around our education system. Children continue to develop physically, socially, emotionally, linguistically and cognitively and their needs should be met, otherwise they will have gaps in their foundational knowledge which will take more time to plug later. Ofsted reconised this in their “Strong Foundations” report in 2024.

If we become so focused on producing the visible outcome earlier that we neglect the developmental architecture underneath it, we may get children to perform something without truly understanding or having the underpinning foundations for what comes next.

And what does more recent research say about play versus instruction?

Skene and colleagues (2022) conducted a systematic review and meta-analysis looking specifically at guided play compared with free play and direct instruction.

They reviewed 39 studies involving children aged approximately one to eight, with 17 studies included in the meta-analysis. And the results weren’t Play wins. But nor were they Direct instruction wins.

Guided play produced better outcomes than direct instruction for some measures, including early mathematics, shape knowledge and task switching. Across a number of other outcomes there were no significant differences.

There was also considerable variation between the studies, and the researchers themselves were appropriately cautious about what could be concluded. Good. We should be cautious.

 It’s pretty tiresome (not to mention downright dangerous) when people decide that one research study = a discovery of the one true way to teach for children aged from birth – adulthood across all subjects. The reality is, it’s a bit more complicated than that.

There is no simple answer.

For some learning, direct instruction may be highly effective. For some learning, guided play may be equally effective. For some outcomes, guided play may be better.

The interesting question isn’t which pedagogy wins? Surely it is What are we trying to help this child learn, and what is the best way to help them learn it?

Play isn’t the opposite of teaching.

This is the false dichotomy I think we desperately need to get away from, you can value play and teach explicitly. I have been doing just that for over 35 years.

You can believe in child-initiated learning and have a carefully planned curriculum. I’ve written several (and even had some published, thanks OUP!)

You can follow children’s thinking and introduce them to things they have never encountered before. You can teach phonics systematically and still believe children need significant periods of time when they can build, move, imagine, talk, negotiate and explore. These things are not mutually exclusive.

Sometimes the adult needs to say “Let me show you.” Sometimes “Let’s do it together.” Sometimes “What do you think?” Sometimes “I wonder how we can find out?”

And sometimes the most pedagogically skilful thing an adult can do is watch, wait and resist the urge to take over. That’s not doing nothing. That’s knowing children, understanding child development, knowing your curriculum and knowing when your intervention will move learning forward and when it will simply interrupt it.

So, did Project Follow Through prove that Direct Instruction is best?

No. Did Direct Instruction perform well in Project Follow Through? Yes. Is there good evidence for explicit instruction? Yes. Should we explicitly teach young children some things? Absolutely. But none of those statements leads logically to

Therefore young children need more adult-directed instruction and less play.

That’s the leap I think we need to challenge. The evidence base for early childhood education is much broader than Project Follow Through. Nearly fifty years ago, House and colleagues called their critique of the Follow Through evaluation No Simple Answer. Perhaps that’s the bit we should remember.

There isn’t one best way to teach everything to all children of all ages. Excellent teachers (and I am thankful to have worked with thousands of them) have a pedagogical repertoire. It’s time to bin the binary because young children’s learning is a lot more complex than it might first appear.

References

Darbyshire, N., Finn, B., Griggs, S. and Ford, C. (2014) ‘An unsure start for young children in English urban primary schools’, The Urban Review, 46(5), pp. 816–830.

House, E.R., Glass, G.V., McLean, L.D. and Walker, D.F. (1978) ‘No simple answer: Critique of the Follow Through evaluation’.

Roessingh, H. and Bence, M. (2018) ‘Embodied cognition: Laying the foundation for early language and literacy learning’, Language and Literacy, 20(4), pp. 23–39.

Siraj-Blatchford, I., Sylva, K., Muttock, S., Gilden, R. and Bell, D. (2002) Researching Effective Pedagogy in the Early Years. London: Department for Education and Skills.

Skene, K., O’Farrelly, C.M., Byrne, E.M., Kirby, N., Stevens, E.C. and Ramchandani, P.G. (2022) ‘Can guidance during play enhance children’s learning and development in educational contexts? A systematic review and meta-analysis’, Child Development, 93(4), pp. 1162–1180.

Sylva, K., Melhuish, E., Sammons, P., Siraj-Blatchford, I. and Taggart, B. (2004) The Effective Provision of Pre-School Education (EPPE) Project: Final Report: A Longitudinal Study Funded by the DfES 1997–2004. London: Institute of Education, University of London/DfES/Sure Start.

Whitebread, D. and Bingham, S. (2012) School Readiness: A Critical Review of Perspectives and Evidence. TACTYC Occasional Paper No. 2.