Is there really no best way to teach?

Whenever child-centred educationalists are in the ascendency, they use their authority to try to impose teaching methods. We have seen this in the U.K. with Ofsted, in North America through various iterations of the standards movements and initiatives such as Productive Pedagogies in Australia. I am even required by the Victorian state exam board to set my Year 12 students open-ended investigations.

To an extent, this is perfectly reasonable. Child-centred educators have a set of beliefs and, when they get the chance, they seek to put them into practice. Others might disagree with these beliefs and practices and so a debate may be had that will test these ideas. This is the stuff of democracy.

Unfortunately, not everyone sees it this way. There are those who would stifle debate. As I understand it, teachers in New South Wales government schools are effectively prohibited from engaging in these kind of discussions by their social media policy. And I’ve heard stories of complaints being lodged with teachers’ employers. 

A few child-centred educationalists remain proudly partisan yet what we see from others is perhaps a little less honest. These are those who, when presented with evidence that contradicts their positions, assert that there is nothing to debate: there is no best way to teach and it all depends on context.

Here are a few thoughts:

Context matters to an extent 

I don’t think anyone argues that we should teach in one mode only. I tend to refer people to Rosenshine’s Principles of Instruction as a concise account of explicit instruction: Students are led through a modelled then guided then independent set of tasks. The key factor in changing the mode of learning is the students’ increasing knowledge. 

Cognitive scientists will point to research that shows that experts tend to benefit from different tasks to novices. This means that accurate assessment is critical to instructional decision making.

So, yes, context does matter. But if all we mean when we assert this is that we must change our methods to reflect the increasing expertise of our students, then this hardly seems a point worth making because it is a point on which everyone already agrees. 

Apart from prior knowledge, learners are very similar

Although the pace of learning may vary, students learn in pretty similar ways. A task that is optimal for one novice learner is likely to be optimal for another and, with the aid of cognitive science, it is possible to draw-up some key principles for the design of these tasks.

If this were not the case, we would be able to point to evidence that showed that Child 1 learnt more from Task A than Task B but Child 2 learnt more from Task B than Task A. Apart from differences in prior knowledge between the children, I am aware of no body of evidence to show such different effects of tasks. In fact, it was a lack of such evidence that convinced the scientific community that learning styles was a junk theory.

How does context matter?

Let’s take this argument on face value. Let’s assume that there is no best way to teach and that everything works in a particular context. And let’s take a example from Productive Pedagogies – the description of tasks that involve ‘lower order thinking’.

“…which occurred when students were simply asked to receive or recite factual information or to employ rules and algorithms in repetitive routines. In such instances students were given pre-specified knowledge, ranging from simple facts and information to more complex concepts. Often this involves knowledge being conveyed to students through readings, worksheets, lectures or another direct teaching medium.”

What contextual cues would lead us to teach for lower order thinking? What signs would we observe in our students that would suggest that this is appropriate? How would we make this decision? In what way would students differ so that we would decide that one group needed lower order thinking and another needed higher order thinking?

Obviously, from the way it is framed, we are supposed to conclude that the teaching techniques associated with ‘lower order thinking’ are generally inferior to the ones that the authors prefer.

And that’s fine. Let’s just be honest about it. If you want to take teaching methods that I think are effective and label them as ‘lower order thinking’ then so be it. Bring it on. Let’s have the discussion and see whose ideas withstand the scrutiny.

But let’s be grown-ups. Let’s not pretend there’s nothing to discuss. Let’s not try to censor teachers. 

The truth will out.


PISA data on discipline 

Embed from Getty Images

So I’m still working my way through some of the findings from the 2015 Programme of International Assessmemt (PISA). PISA don’t just run assessments of knowledge and skills, they also survey students on a range of matters. I’ve already written about the association between inquiry learning and science performance and the association between perceptions of feedback and science performance.

Students were also asked about their perceptions of the disciplinary climate in science lessons and the results are pretty interesting. From a range of questions about levels of noise and disorder and whether students followed the teacher’s instructions, the statisticians created an index of disciplinary climate. Positive values reflect a better climate than the OECD average. 

As you might expect, countries such as Japan, Korea and Hong Kong are in the top ten. Perhaps surprisingly, so is the United States. In contrast, the United Kingdom and Australia both rank well below the OECD average. The U.K. result seems to be at odds with Ofsted evidence that generally finds few discipline problems in schools. We have to wonder which gives a more accurate picture, anonymous student self-reports or the results of a pre-announced, high stakes inspection. 

Not surprisingly, in virtually all countries, a higher score on the PISA index was associated with better science performance. 

PISA also surveyed principals about their views on school climate and it’s interesting to contrast the two sets of data. For instance, U.K. principals report a relatively low rate of students lacking respect for teachers. Perhaps different groups of stakeholders have different levels of expectations.

A culture of labels

Embed from Getty Images

It seems as if humans have evolved an innate desire to categorise. This is probably a reason for our cognitive power but it can have unfortunate consequences, especially when we start labelling each other. Politicians of the left and of the right have used this to the advantage of themselves and the detriment of society.

We need to recognise this dangerous tendency in education. The learning styles fad is not just an idea for which there is little evidence. It’s about labelling people. And once we label them, we tend to view any differences between us and them as relatively permanent: I may be an auditory learner but this boy from an ethnic minority background is a kinaesthetic learner so we cannot possibly expect him to sit still and learn to write. 

If we modify our expectations – if we try to accommodate the perceived difference rather than address it – then we compound that difference. A child who cannot write needs more work on his writing not a label that means we expect him to write less.

This is what may happen when we label a child ‘dyslexic’. I am not here to argue whether dyslexia does it does not exist but to point out that, at least in some cases, the label might not be helpful. 

We know, for instance, that a rigorous phonics-based intervention can help children with dyslexia. But the label is as likely to prompt us to try to accommodate reading difficulties by lowering our expectations for reading and writing or perhaps make us susceptible to spending money on unproven interventions such as coloured lenses. 

And this is why I feel queasy at the explosion of learning difficulties and disabilities. These are all diagnosed by the behaviours that a child displays. The fact that such diagnoses rocketed towards the end of the last century and the start of this one shows us that these behaviours used to be considered part of the spectrum of normal behaviours. A child will now attract a label that he would not have attracted in the past.

This is clearly well intentioned and reflects a growth in attention and resources aimed at children who are identified in this way. This is probably a little circular – once you have the resources in place, including staff with a role to coordinate those resources, you’re likely to identify ever more children with needs and disabilities.

The problem here is that these labels may prompt the wrong actions on the part of teachers. If a child behaves poorly because she has a learning disability then we might be tempted to try and accommodate this – to work around it. Indeed, this is what the law suggests we have to do once something has been labelled a ‘disability’.

However, it might be better for that child to address the poor behaviour. Punishment and rewards might be one way to address it but other ways could include the use of a restorative approach where the child reflects on her behaviour or a proactive approach where the child receives explicit teaching of the right behaviours. All of these approaches are far more optimistic than one that consigns a child to a box.

We need to be wary when we see labels in education. The labels don’t really exist. The children do.

How goals affect your teaching style

Two axis model

It can be a bit of  puzzle. There is no reason why a particular curriculum should imply a specific way of teaching it. For instance, Dan Willingham has written at length of the need for a knowledge-rich curriculum and yet his statements on teaching styles are relatively few, limited to cautioning us about flashy hooks or the overuse of pure discovery learning.

Yet, on social media and when examining the way that schools enact different educational philosophies, a knowledge-rich curriculum tends to be associated with explicit instruction whereas a knowledge-lite curriculum, such as the one embodied by England’s 2007 National Curriculum, is often seen to favour for exploratory, student-directed forms of learning. This seems likely to be the reason why the Times Educational Supplement (TES) recently seized upon what it mistakenly saw as a criticism of explicit instruction by knowledge advocate, E D Hirsch Jr. They knew that the headline would be clickbait.

I don’t think means and ends are as independent as we might think. If you have a body of knowledge that you wish to communicate then, as Hirsch suggests, research and even common sense tell us that explicit methods will be more efficient. You need to learn-by-doing if you have a skill that needs practising. To traditionalists, children must first learn a body of knowledge before practising its application whereas, to progressive educators, something like scientific inquiry is itself a ‘skill’ that can be practised and refined from the outset and that builds other more ephemeral ‘skills’ in the process, such as collaboration. This means that all teachers should, or at least intend to, reach the point of independent application but that traditionalists will pay far more attention to building a body of knowledge first and they believe the best way to do this is explicitly.

Rosenshine’s principles of instruction actually includes application in part of its definition of an explicit teaching method. The recent monograph by Andrew Martin adds a lot of flesh to this idea.

There are those who love to suggest that advocates of explicit instruction are only interested in students repeating back disconnected facts that they don’t understand. But there are two problems with painting this picture. Firstly, I am not aware of anyone who is arguing for this. Secondly, explicit instruction is consistently associated with better performance on tests of application, with PISA 2015 being just the latest example.

E D Hirsch Jr’s article on why specific knowledge matters

Embed from Getty Images

At the start of December, the Times Educational Supplement (TES) in London published an article by E D Hirsch Jr. The article was linked to Hirsch’s new book, “Why Knowledge Matters,” which I encourage you to read.

The piece is paywalled but the TES produced a teaser which caused a stir because this teaser claimed that Hirsch suggests mainstream science does not support the use of direct instruction. I am a promoter of both Hirsch’s ideas and direct instruction (which I tend to refer to as ‘explicit instruction’). Had I, and others with similar views, been made to look a little silly?

When I read the article itself, I wasn’t clear that Hirsch’s comments supported the line in the teaser.

One explanation which is consistent with Hirsch’s book is that the statements that he makes about Direct Instruction refer to specific literacy programs developed by Engelmann and colleagues and reflect Hirsch’s belief that the positive effects of these programs wash out over time due to their failure to address the issue of world knowledge. Note that I have capitalised the first letters of ‘Direct Instruction’. This is the convention that many use to distinguish Engelmann-style programs from explicit instruction more generally.

I decided to contact Hirsch to clear this up. He indicated that the teaser was inaccurate and offered the following explanation of his comments on Direct Instruction:

“The evidence is clear that explicit instruction, well conducted is more effective than inexplicit instruction. Lab research, Project Follow Through, common sense, are unanimous about that. My comment on Success for All and Bereiter-Engelmann in my book was that DI modes were great for teaching what they taught, did so more effectively than indirect methods, but what they taught was misconceived. And that’s why they had limited positive effects on mature literacy. That’s what I said in my recent book.”

So that clarifies Hirsch’s position. He also gave me permission to publish the unedited version of the article on this blog which I do below. Note the use of capital letters for ‘Direct Instruction’ which became lower case in the version published in the TES. 

Why Specific Knowledge Matters

E. D. Hirsch, Jr.

While I was writing my just-published book “Why Knowledge Matters” a controversy erupted in New York State over an early-reading program that I helped develop, called “Core Knowledge Language Arts.” The program had been selected as the official early-childhood reading program of New York State. Placed on the websites of the state and of the Core Knowledge Foundation, the program was downloadable for free by New York schools and anybody else. Such programs normally cost schools thousands of dollars. Even factoring in printing costs, a great deal of money can be saved over the years by making use of this free program — which had been shown to be outstandingly effective in a twenty-school, three-year trial in New York City. Nonetheless, many teachers resisted, saying that the new program was “developmentally inappropriate.”

“Developmentally inappropriate” is a term used freely by early-childhood teachers. They have been taught to believe that education should follow a natural path of growth, so that exposing a young child to adult material prematurely is fruitless, harmful and boring. That sentiment had already caught on by 1929 when Antonio Gramsci, the brilliant socialist thinker imprisoned by Mussolini, wrote the following in one of his “Prison Notebooks”:

“The new education has become a kind of church that has paralyzed educational thought, and has given rise to some curious inversions. It is almost imagined that the brain of the child is like a ball of yarn that the teacher should help to unwind.”  

This naturalistic “new education” has now held on for more than a century. It is still received doctrine that an untimely Imposing of “rote-learned” facts on children before they are “developmentally ready” will harm their “development” and kill their joy in learning. In the Core Knowledge controversy, the most celebrated instance of such untimeliness was a first-grade unit on Mesopotamia. Here’s a typical teacher’s internet comment on the unit: “It’s NOT age appropriate. Six-year-olds should be learning about their own communities and surroundings.”  

But that view isn’t based on accurate child psychology, and it is socially regressive. The famous psychologist Jerome Bruner stated years ago that any topic, properly presented, can be taught at any school age. Moreover, from the standpoint of social justice, withholding knowledge from poor children that rich children routinely gain is socially regressive. Middle-class parents in the U.S. have been harmlessly teaching their first graders about Mesopotamia for the last twenty-five years, from my book “What Your First Grader Needs to Know” (1991). Bored by the nullity of “our house” and “our community,” first graders and their parents have been charmed by learning tongue twisters like the “Code of Hammurabi,” and the chance connection between Mesopotamia and a hippopotamus. By sticking to “our community” the teachers of less advantaged children whose parents lack the means, education, or leisure to read such things to their children, are further holding back children who have already been held back.  

Mesopotamia is a trivial example, of course. It happened to be the lightning rod for New York State. Yet the issues raised by the incident are anything but trivial. My new book shows that in the United States, the idea of natural development, that is, the idea of tailoring the content of education to the development of the individual child has, along with related misconceptions, resulted in the following misfortunes: the narrowing of the early curriculum, the over-testing of students for non-existent general skills, and the widening of the achievement gap between rich and poor students as they progress through the grades.  

The book’s critiques are consistent with consensus science in psycholinguistics, cognitive science, and developmental psychology. I argue that the scientific consensus in these fields is the nearest we come to the reality principle in education. That’s why I invited well-known researchers — Steven Pinker, Susan Neuman, Daniel Willingham — to blurb my new book rather than asking public figures or representatives of any particular educational policy. I’m gratified that these scientists have endorsed the arguments of the book.  

My claim of scientific accuracy is strategic for the book’s ideas, because they are heterodox ideas in current American education. My book claims that our guiding theories and policies in the schools do not comport with the current scientific consensus, and that this explains our frustrations. We have recently become disappointed in policies and programs that seemed experimentally promising such as smaller class size, Direct Instruction, and Success for All. They were all supported by carefully conducted experiments. But in the long run they have disappointed. We have been likewise disappointed by the charter-school movement.  

These disappointments suggest that mainstream science needs to be our guide, not confident slogans nor educational experiments. The uncontrolled variables in short-term educational research fly in every direction. Education is a long-range process.  

I therefore argue that it’s time to look at policies that flow from relevant scientific principles that are strongly supported by largescale international school data. The consensus in basic science represents the reality principle in education. Any educational proposal or policy that runs counter to that consensus is likely to disappoint. This intellectual corrective is especially important when it comes to two ideas in particular: the importance of developmentally-appropriate practice, and the existence of general, all-purpose skills like finding the main idea, problem-solving, and critical-thinking.  

In current scientific literature, skilled performance is rarely called a general “skill”. The current term of choice is “expertise.” The verbal contrast between skill and expertise is important because the term “skill” has allowed educators to adopt the convenient but incorrect principle that the specific content of education is less critical than the development of general skills like problem-solving, and critical-thinking and finding the main idea. It turns out that these supposedly general competences simply do not exist as all-purpose skills independent of specific knowledge domains. There are few all-purpose, free-floating skills. Skills are dependent on specific knowledge. A key term in the current scientific literature is “domain-specific.” Most of the competences that children need to learn in school are domain-specific competences.  

This finding has exceptionally strong implications for schooling. If skills matter, as everyone concedes, then domain-specific knowledge must matter a great deal. The specific contents of a school curriculum matter crucially. The claim that “you can always look it up” is an outmoded slogan. You can only look things up effectively if you already possess preparatory, domain-relevant knowledge. Our slogans about general skills and our brave new world of instant internet information all need to be re-examined. Psychologically, they are not reality-based.  

But above all we need to reconceive our faith in individual natural development – the view, as Wordsworth put it, that “Nature never did betray the heart that loved her.” In developmental psychology, emphasis is gradually shifting from the word “development” towards the word “culture.” The differentness of a child’s development within different languages and cultures has become a major theme in the developmental field. To his great credit, Piaget himself heralded that change in a speech of 1964 when he conceded that “maturation is not everything.”  

This points up a paradox concerning “natural” development. It’s now understood that letting a child unfold naturally according to its own time clock like a ball of yarn is quite an unnatural proceeding for human beings. What is natural for a child is to be inducted into the particular language and value-orientations of a particular culture. And these cultures are quite varied. The word “culture” is now understood to be a co-determinant of “development.” This leads to the paradox that so-called “unnatural” education is inherently natural, whereas naturalistic education is inherently unnatural. This shift in scientific understanding has implications for educational policy and practice: Truly natural development for human beings includes acculturation. In our post-agrarian era, with the rise of cities and nations, when parents have partially relinquished education to schools, a chief responsibility of schools is acculturation. The abstract, naturalistic approach does not work.

The story that I tell in my book thus has two subplots: the story of policies that do not work, and the story of the underlying ideas that have caused us to persist in those policies. It’s possible that the intellectual-history sub-plot could be the main contribution of the book. Absent changes in our underlying convictions about individualism and natural development, we are unlikely to make fundamental changes in our practice.  

After many years in educational discussions I think I can now identify the chief underlying idea that needs changing. I call it providential individualism – the focus on the unique individual rather than on acculturation, combined with the belief that some supervening providence like nature or the free market can guide our educational policies. On the contrary it’s neither providence nor nature but we adults who need to decide quite specifically what our children shall know and be able to do. The book thus accuses both the naturalistic left and the market-oriented right of misplaced providentialism. The critique thus includes most of my colleagues in the school world. My main aim has been clarity. To offend everybody is one of the few prerogatives left to old age.

Thoughts on the journey so far

Embed from Getty Images

The first month of this blog attracted nine hits. Back then, it was hard to imagine that it would eventually be viewed over 300,000 times. Yet I am also aware that this is pretty minor in the world of blogging. I’m niche.

The arguments that I construct here are based upon evidence. This evidence is contested and there is evidence to support views that are at odds with my own. Nevertheless, I am no conspiracy theorist or climate change denier. My views are based in pretty conventional research.

Which is why I find it so surprising that this is one of the few spots where you will read these ideas. I hesitate to use the term, but the ‘mainstream media’ seems to have a systematic bias towards ed tech and progressive teaching methods. The recent finding that inquiry learning had a negative association with PISA science scores was worse than ignored. In some instances, the papers offered inquiry learning as the solution to declining scores.

I do understand that the PISA finding is only a correlation. I am personally convinced by the cognitive science rather than one piece of survey data although I see it as a tentative confirmation and I wonder what would have happened if the result had been the other way around. But is this the reason the media ignored the story? Have they suddenly adopted What Works Clearinghouse standards of evidence that they don’t apply when writing breathless articles about mindfulness?

So that’s why I’m here, plugging away. There are a few of us around – our numbers are growing and I am keen to welcome you to the club.

But here’s some advice. There are people out there who don’t like bloggers. They threaten to report bloggers to their employers or anyone else they perceive as having the authority to shut bloggers down. Make sure that anything you write is defensible. Do not divulge confidential information. Distinguish between facts and opinion. Avoid personal attacks – the only personal comments I make are to highlight the behaviour of those who try to silence me.

Feel free to go after the ideas. Go after them with gusto. However you write, someone will take issue with your tone. Make your best attempt to be respectful but recognise that you will draw this kind of criticism. I don’t tend to respond to it much. It’s the ideas, stupid.

Most importantly, the internet is your space. If you have enjoyed spending some time in this little corner of it then you might be ready to carve one out of your own. Who knows, you may be the one to break through and take your thinking out into the mainstream.

Reducing cognitive load in order to boost motivation

Embed from Getty Images

When I first started writing about cognitive load theory and the need for explicit instruction, a lot of constructivists responded to me by suggesting that I was ignoring the role of motivation. Perhaps explicit teaching is the most efficient way for students to learn but if it turns them off your subject for life then what has been achieved? Instead, we need to give them relevant and interesting challenges. Maths education expert, Dan Meyer, suggests that we need to ask students questions that induce a state of “perplexity” in order to provoke interest and promote engagement. But I doubt that being perplexed is very motivating at all.

Instead, I have the view that it is the process of getting better at something that is motivating. In an interesting study from Quebec, students in Grades 1-4 reported their level of motivation towards maths and this was matched with maths test scores. Levels of maths achievement predicted motivation in later grades but motivation did not predict later levels of achievement. By assuming that we must provide specific activities to motivate students we place the cart before the horse. We need to focus on the quality of learning instead.

In an excellent monograph available via ResearchGate (thanks to @RonMarchetto for the tip-off), Andrew Martin of the University of New South Wales draws a number of threads together to develop an overarching theory that he calls “Load Reduction Instruction” (LRI). This carefully charts the evidence in favour of avoiding cognitive overload for students at each stage of learning. Those readers who are familiar with cognitive load theory will recognise his argument that working memory is limited and easily overwhelmed and that many teaching methods don’t take proper account of this. They will also recognise the idea of learning as a change in long-term memory (see here).

However, Martin does not stop here. Levels of guidance need to be gradually reduced as students move from novices to experts. I’d argue that instruction must not be inquiry-based (or project-based or problem-based) because this kind of teaching overloads working memory early in learning but there certainly is a role for more open-ended inquiry once students have developed sufficient automaticity and fluency. For instance, we know that solving problems becomes more effective than studying worked examples for students with sufficient levels of expertise. The key principle of LRI is therefore to reduce load during initial learning.

Martin acknowledges that more work needs to be done to tease-out the precise ways in which different components of LRI – such as direct instruction or worked examples – affect motivation. However, he does present evidence that a sense of success is important and states, “novices and academically at-risk students can have difficulty in early phases of learning and this is likely to impede their sense of efficacy throughout the learning process” [reference omitted]. The kinds of strategies he suggests will be familiar to those who have read some of the process-product research of the 1960s and 1970s; pre-training (e.g. teaching the names of the parts of a motor before teaching how a motor works), segmenting (breaking tasks into small, manageable chunks), retrieval practice and modelling.

The opposite kinds of strategies come later in a learning phase and again might enhance motivation. For instance, Martin suggests the following about ‘integrating’:

“For example, punctuation is often taught in isolation from students’ editing of their own essays and assessment tasks. In such cases, an opportunity to build a sense of relevance with regards to punctuation is lost. Integration might involve students being presented with an explicit punctuation check list (e.g. capitalise the start of a sentence, end questions with a question mark etc.) that they work through after they have written an essay. Thus, there is structured and scaffolded support for punctuation built into the student’s own essay writing activity that increases the perceived relevance and personal meaning associated with the punctuation activity.

Notably, integration is the reverse of some approaches to pre-training and segmenting described above, especially with regards to the ‘isolated elements effect’ . Whether elements should be isolated or integrated depends on available working memory resources that in turn depend on levels of knowledge – further underscoring the importance of pre-training if and when needed. Notwithstanding this, as a general principle, integration of information, materials, and/or activities allows students to better appreciate important connections in learning and thus the value of the relevant information, materials, and activities for other parts of their learning.” [reference omitted]

I think that this is a useful set of ideas that add much needed nuance to the current debate around teaching methods. It is not the case that segmenting is superior or inferior to integrating. But it is also not the case that both are equally valid strategies that different teachers can draw upon as they see fit. Instead, they are both valid at different stages in a phase of learning. In a sense, this gives justification to more traditional teaching sequences that work from the parts-towards-the-whole rather than the other way around. I discuss this in my book as ‘bottom-up versus top-down’.

Martin suggests that guided discovery learning may be effective when students are more skilled and knowledgeable – which implies a place in the learning path – and he attempts to clear-up some of the confusion around the appropriate level of guidance. Guided discovery may be considered part of LRI when it sufficiently reduces cognitive load. However, “If too much of the process remains undefined and uncertain, too much of working memory must then be directed to potentially distracting and irrelevant processes that have the capacity to lead to misinterpretation, inaccurate conclusions, and inadequate skill development.”

Martin’s monograph is fascinating. I’m not sure whether I yet agree with all of it. I need to read more about some of the ideas. However, I highly recommend it if you want to read an accessible article that will make you think about the science behind different teaching methods. One of his conclusions – that we should consider LRI to be ‘student centred’ – is an important counter to those who claim the moral high-ground for more constructivist teaching approaches.