Here’s the second part of what I said yesterday at the ResearchEd event near Birmingham.
From a hundred years of research, there are three deceptively simple insights that when applied well in the classroom, have very powerful effects. They are not quick wins, silver bullets or revolutionary innovations. Instead, they offer something more modest: a chance to focus our teaching and help pupils remember what they’re learning. And expert teachers have been doing them since time immemorial.
Here they are in three words: examples, practice and quizzes. Worked examples, extended practice and frequent quizzes are much-underestimated and under-valued, but there’s a huge volume of scientific research behind them.
Intuitively, using examples makes sense. Any time we learn an abstract concept, the mind yearns for concrete examples. For instance, when I was learning recently about the concept of reliability in assessment, I struggled to grasp it until I heard this analogy with a familiar device: weighing scales. The higher the reliability, the higher the likelihood that two weigh-ins of the same object will yield the same result on the weighing scale. Intuition tells us examples are needed. As Willingham says, ‘people find it hard to understand abstractions: they need concrete examples to illustrate what the abstractions mean.’
Counterintuitively, though, the worked example effect, replicated several times across several subjects since 1985, shows that learners required to solve problems perform worse on subsequent test problems than learners who study the equivalent worked examples. Studying and comparing lots of worked examples reduces cognitive overload. Working memory is freed entirely for the study of the problem and solution steps. In 1987, Zhu and Simon found in a series of long-term studies that a 3 year maths course was completed in 2 years due a focus on worked examples.
So here’s how I use worked examples in English. Take complex procedural knowledge that I want pupils to retain in their long-term memories, like analysis of a poem. Here’s a worked example of poetry analysis that I’d share with my pupils.
It is the insight from these experiments about how to make worked examples work best that most intrigue me. Here is the key insight:
Worked examples must make pupils think hard to identify critical insights by annotating them with what they are supposed to illustrate.
For instance, in English, and other subjects with a heavy writing load like humanities, this means getting students to compare worked examples of model paragraphs, to criticise and improve, to annotate and aspire to. If they haven’t seen an example of what they’re aiming for, how can they work towards achieving it? The best teachers write lots and lots and lots of example paragraphs, introductions, conclusions and essays.
Using this insight, I can improve my use of worked examples. I can provide two or three, and ask students to compare them, annotate them, and work out which analysis of the poem works better and why. Here’s an example of what an annotated worked example would look like.
Why don’t we do this more? It requires a lot of time – and no little expertise.
Intuitively, we recognise practice as vitally important. We can’t imagine anyone improving at tennis, golf or any sport, at piano, in an orchestra, or in any form of music, without extensive, deliberate practice.
The research shows how much better distributing extended practice over time, rather than massed, blocked cramming is for long-term memory retention. This has been tested and demonstrated in hundreds upon hundreds of replicable scientific and classroom experiments, across learning conditions (reading, listening, writing), student characteristics (age, ability & prior knowledge), materials (problems, texts and questions across subjects) and tasks (recall, problem-solving and comprehension). As Willingham says, the three key benefits of practice is that it ‘reinforces basic skills required for more advanced skills, it protects against forgetting, and improves transfer’.
It is the insight from these experiments about how to make practice work best that most intrigue me. Here is the key insight:
Compared to conventional problems, completion problems decrease extraneous cognitive load, facilitate the construction of schemas, and lead to better transfer performance.
Using this insight, I can design completion problems to hone in on what pupils find hardest in their practice. For instance, here is a completion problem I’d use with my pupils to guide their practice of analysing a poem before independent writing practice.
If extended independent writing is the ultimate aim, pupils must be given the opportunity to practice that with less guidance:
What’s exciting is how cognitive science dovetails with expert practitioners. Siegfried Engelmann held that pupils need ‘five times more practice than many teachers expect’. In Doug Lemov’s book Practice Perfect, based on decades of work with expert teachers, codifies how to improve practice. And my veteran colleague Barry Smith has always advocated pre-emptive planning that starts from what students find hardest.
Counterintuitively, longer lags and intervals between practice sessions improve retention. This is an insight more for curriculum design than for classroom instruction, which I’ll leave for another time. I just want to focus on what classroom teachers can do in their day-to-day teaching without changing the entire curriculum.
Intuitively, we know that being quizzed on something helps us remember it. That’s why actors memorising their lines don’t just highlight them or re-read them; they test themselves on them again and again and again until they’re automatic in long-term memory. We can remember poems off by heart if we quiz ourselves on them line-by-line.
Perhaps counterintuitively, the research shows that quizzing is better than studying for long-term memory retention. More than 100 years of research has yielded several hundred peer-reviewed, replicated experiments that testify to this. For instance, 4 blocks of study with practice tests outperformed 8 blocks of study without. Practice testing outperforms restudying. The advantage of practice testing with feedback over restudy is extremely robust. Distributed practice testing is better than distributed practice alone. So here’s an example of a quiz I give to my pupils to start a poetry lesson:
The insights about how best to use quizzes or practice tests are fascinating.
In terms of dosage, it turns out more is better. Shorter and more frequent tests (one a week) are more effective than longer and less frequent (once every six weeks).
On timing, it’s interesting to note that longer is better: sizable benefits are observed when repeated tests are spaced: longer lags produce greater benefits.
To improve on this, I prepare weekly homework quizzes. The more practice testing, the better the impact on long-term memory retention.
Lots of questions remain. Here are a few: how do we create enough worked examples and completion problems, given time constraints? How do we work out how to best space and interleave practice and practice tests? There were some excellent questions from the audience, particularly from David Weston about evaluating impact, and how all this applies to teacher training. David even adapted his own ResearchEd talk in the light of this, which I’m looking forward to seeing – and will hopefully write about.
I ended with an anecdote. My dad’s a prostate cancer surgeon – and he got prostate cancer last year. When I ask him what’s changed most over his career in medicine, he says without hesitation, scientific research has changed surgery. And it saved his life. The screening and operation that cured him were based on applying scientific research to the medical profession. Perhaps when we look back on our careers in education, we’ll say, scientific research has changed teaching.
The scientific revolution has brought us mastery over the world and our bodies: vehicles that permit flight and surgery that allow us to extend life itself. The promise of cognitive science – the science of the mind – is that it could bring us similar mastery over how we learn.
A hundred years of replicated scientific research can begin to tell us not only what works, but what works best, and why.
Many of the references I’ve collated above are from the peer-reviewed paper of five cognitive psychologists (Dunlosky, Rawson, Marsh, Nathan and Willingham) that synthesises over a century of scientific research evidence.