Why Scott Alexander is wrong about schools

This is your periodic reminder that my new blog posts are over at Substack. You can sign-up for a free subscription.

Scott Alexander has been in the newspapers recently – specifically, the New York Times. Alexander (a pseudonym) was the author of Slate Star Codex, a popular blog that I used to dip into and that I have recently learnt was part of a some ‘rationalist movement’ about which I know little. Last year, the New York Times decided to write an article on Alexander and told him that as part of it, they would disclose his real name for lols. Alexander, a practising psychiatrist at the time, concerned about how the publicity may affect his work, complained about this doxing threat and closed down his blog and the New York Times then sat on the story. Now, Alexander has re-emerged on Substack under his real name and the New York Times have finally published their piece which turns out to be something of a hatchet-job, weakly attempting to link Alexander to the alt-right and everything that’s considered bad in their weird universe (you can read Alexander’s rebuttal here).

Continues at Substack


Academic wellbeing

This is your periodic reminder that my new blog posts are over at substack. You can sign-up for a free subscription.

A recurring motif in the world of the international education consultant is the diagram that opposes one set of ideas against the other – good versus bad.

I was struck by this when I read a new paper for the Australian Centre for Strategic Education by Michael Fullan, a Canadian educational consultant, and hit upon a diagram listing ‘drivers’ for whole system success. The right drivers are called, ‘the human paradigm’ and include, ‘wellbeing and learning’, ‘social intelligence’, ‘equality investments’ and ‘systemness’. How nice! The wrong drivers are apparently, ‘the bloodless paradiagm’ and include ‘academics obsession’, ‘machine intelligence’, ‘austerity’, and ‘fragmentation’. These wrong, rather pale-looking, drivers sound rather bad.

Continues at Substack


Paul Kirschner


Paul Kirschner is Emeritus Professor of Educational Psychology at the Open University of the Netherlands and Guest Professor at Thomas More University of Applied Science in Belgium. In this episode, Paul talks to Greg Ashman about his long career in educational psychology, the key distinction between epistemology and pedagogy and that Minimal Guidance paper he wrote with John Sweller and Richard Clark. Paul and Greg also discuss Paul’s new book co-authored with Carl Hendrick, How Learning Happens. Unfortunately, Paul and Greg run out of time before all of Greg’s questions are answered and so Paul has agreed to return in the future.


Filling the Pail has moved to Substack

Embed from Getty Images

From now on, new blog posts will appear on my Substack page. If you navigate to the page, you can sign-up for a free subscription which will alert you to new posts via email. If you follow my WordPress site via email then you will need to sign-up to the new Substack page if you want to keep up-to-date. Until now, I have been cross-posting new Substack posts to WordPress but I am going to stop that. The WordPress site will therefore remain as an archive of posts up to the end of 2020.

There also may be some disruption if you tend to follow my posts via Facebook or LinkedIn because I haven’t yet figured out ways of automating alerts – although I’m looking into it. I will continue to (over)promote posts via Twitter.

Why have I made the switch? I have grown dissatisfied with the (sometimes quite unpleasant) adverts on WordPress and the look of the site. Substack allows for paid subscriptions and then takes a cut of these as its funding source, so they don’t need to put up adverts. This means I can have a free site with no ads. It also means that in the future, if I want to, I can create a paid subscription for additional material for a few dollars a month. I am not sure whether I will do so or what that would look like, but it’s an option missing from WordPress at the moment.

As with all innovations, it may not work out and I may end up back here. Regardless, rest assured that the extensive archive on this page will stay live.

If you like my posts, please consider sharing them on social media and telling friends and colleagues about them. I am about to write a series of posts tackling my key themes. One advantage of Substack is that social media sceptics can receive my posts via email without having to lurk on Twitter.

And finally, if you haven’t checked out my podcast, consider giving that a go too. You’ll find a mix of well-known guests such as Dylan Wiliam alongside guest who should be better known. If you like it, please share and maybe leave a positive review wherever you get your podcasts.


Natalie Wexler


Natalie Wexler is an author and journalist who became interested in educational issues when she began to work with students in disadvantaged schools in Washington. Natalie is co-author of The Writing Revolution, with Judith Hochman and author of The Knowledge Gap. In this episode, Natalie talks to Greg Ashman about her journey into education, the Impact of The Writing Revolution and how its methods align with cognitive science. Natalie and Greg then discuss The Knowledge Gap, the reason why we need more of a knowledge focus in schools and some of the objections and barriers to this idea before discussing some possible solutions.


Alternatives to differentiation

I have been writing about differentiation for at least six years. However, after my most recent post, I noticed a difference – people started asking me for alternatives. Perhaps that signals that we have reached a tipping point.

Firstly, we need to establish a few points, Differentiation is poorly defined and at least three of my suggestions could be described as forms of differentiation. Also, I am not claiming that evidence supports these options. The evidence base has always been thin.

My primary concern with differentiation has always been its logic. As understood in most schools, it usually means creating groups within a class and preparing different tasks for each group. This implies a heavy workload for the teacher. As Rebecca Birch pointed out on Twitter:

It also reduces the capacity for explicit teaching. If you have one class divided into six groups then in a one hour lesson, you have a maximum of ten minutes you can spend with each group. That’s before you take into account any admin time and time spent on classroom management – which tends to expand once you group the students because you have to keep intervening with the groups you’re not currently working with.

So, I am not claiming my alternatives have a strong evidence base drawn from randomised controlled trials, I am claiming that they are based on superior logic.

I also wish to make it absolutely clear that I think reasonable adjustments should be made for children with a disability. There is an issue with the quality and practicality of many proposed adjustments, as well as the issue of potential over diagnosis, but I won’t go into that here.

Finally, I am going to focus on approaches to maximise academic progress i.e. address students’ needs. There are occasions when we may simply want to accommodate them. For instance, for obvious reasons, we may want a young person who cannot write to still participate in a sex education lesson. In these circumstances, putting up a barrier by requiring that student to write notes would be perverse.

So, caveats out of the way, what are my bright ideas?

Don’t create gaps

This may initially seem like unhelpful advice, but I think we should avoid creating gaps between students in the first place. Yes, some gaps are unavoidable because students joining school will vary in terms of family background, working memory capacity and a range of other factors. However, I do think there are circumstances where we make it worse.

For instance, when compared with East Asian maths teachers, maths teachers in the U.S. and Australia tend to be more insistent that understanding must come before memorisation. East Asian teachers still think understanding is important, but they are more relaxed about it coming after memorisation,

Understanding is usually expressed by students through their language skills. This means that children with more advanced language skills will be more able to give teachers the cue that they understand and therefore move on to the next step. I have heard teachers say, for example, that students cannot move onto decimals until they’ve demonstrated a good enough understanding of fractions. But on what basis are such judgements made? How do we know these kids can’t cope with decimals?

Instead, we should focus on drilling in number facts, GPCs and so on, regardless of whether we get these understanding cues.

Similarly, ineffective teaching practices will have a differential effect. Students with the best internal resources will be able to cope, but those without will become lost, further increasing the gap. Instead, we should use effective methods such as explicit teaching.

Extension booklets

A rather prosaic form of differentiation, the power of the humble extension booklet should not be overlooked. More advanced students can become bored as the teacher re-explains something to the less advanced because they are able to more rapidly transition from ‘I do’ to ‘We do’ to ‘You do’. There are two main things to bear in mind.

More advanced students still need to be taught how to do things, so an extension book full of questions on content they have not been taught will simply provoke lots of questions.

Secondly, students themselves are not always a good judge of when they can move on to the booklet. With many maths skills, for instance, we want students to achieve mastery. That involves them practising past the point of being able to get the right answer and up to the point where they cannot get the wrong answer. Therefore, the teacher still needs to control when these students move on.

Response to Intervention

I have written about Response to Intervention (RTI) before. The model consists of three tiers. The first is high quality instruction that all children receive. After this, some sort of screening check is put in place and a group of students who have not made adequate progress are identified. These pass into Tier 2 – a small group intervention which consists of a more intensive version of the Tier 1 teaching. Finally, students who don’t make adequate progress in Tier 2 pass to Tier 3 for individualised support.

RTI is good for non-negotiables such as reading or basic maths. It is clearly resource intensive and so few schools would be able to afford to apply it to other areas. It also relies heavily on the quality of the Tier 1 instruction and the screening checks. If your Tier 1 is balanced literacy and your Tier 2 is reading recovery then you can forget it.

Ability grouping

Ability grouping – assigning students to different classes based on what they have demonstrated they can do – is a bogeyman among educationalists because they see it as inequitable. One group of researchers called it ‘symbolic violence‘.

Meta-analyses tend to show a small advantage for students assigned to the more able groups and a small disadvantage for those assigned to the lower groups.

However, these meta-analyses tend to suffer from all the worst features of educational meta-analyses. Few of the studies involved are randomised. Flexible ability grouping in different subjects is often conflated with ‘tracking’ where students are assigned to the same ability group for all of their classes and even with within-class ability grouping or what many people think of as differentiation.

The UK’s Education Endowment Foundation had a chance to sort this out via a randomised controlled trial, but they bungled it, possibly because the study involved some of the researcher who think ability grouping is symbolic violence.

In my view, ability grouping offers a chance to pitch content in a more targeted way without the many practical issues associated with within-class differentiation. The problem arises when the less advanced classes are given to the least experienced teachers or when the content is degraded under some of the mistaken assumptions I described above. Layer in a school context with an inadequate approach to discipline and teaching the less advanced class becomes about keeping students busy rather than ensuring they make progress.

I would therefore only advise approaching ability grouping if you have systems in place to mitigate these problems. You may also need to prepare your response to ideologically motivated attacks.


Differentiation: It is now time to think again

In January 2016, I wrote a blog post about differentiation. Part of the problem with differentiation has always been about what it actually means. In essence, differentiation is the attempt to cater to a wide range of students’ needs and abilities. Some people therefore class ability-grouping – placing students into different classes based upon their level of advancement – as differentiation, but in my experience, most people tend to restrict the meaning to catering to different students within the same classroom. All teachers do some of this, even just by setting extension work or answering students’ questions. However, the kind of differentiation advocated by education faculties and consultants tends to take this much further.

In the 2016 blog post, I referred to a graph I had generated in 2014 from PISA data showing that, if anything, greater levels of differentiation were associated with worse mathematics results and, referencing a piece of research from the U.S., I concluded that:

“Even when researchers have tried to make it work, they have complained that the teachers weren’t doing it right. So it is either something that works if you have particularly talented teachers who can implement it – although this has not been demonstrated – or it is an idea that doesn’t work at all. You decide. What is clear is that it is not an approach that is grounded in solid evidence.”

In May of the same year, I was surprised to find the same blog post referenced twice in an article in The Conversation by Linda Graham and Kathy Cologon titled, “Explainer: what is differentiation and why is it poorly understood?” The references to my blog post were not positive, they were used as examples to support the contention that differentiation is poorly understood.

In their article, Graham and Cologon start by noting that, “Differentiation is a long word that sounds complicated,” before suggesting that, “it just means teachers plan for the children who are actually in their class, instead of designing lessons for their idea of the ‘average’ child.” This implies that they do not think of ability-grouping as a form of differentiation and they explicitly criticise ability-grouping later in the piece. Oddly, they illustrate their understanding of differentiation with a clip from a Hollywood Movie before advancing a model of differentiation known as Universal Design for Learning or UDL:

“One particularly well-developed and internationally known resource is Universal Design for Learning (UDL). UDL is a framework informed by a large body of research investigating the many different aspects of teaching and learning involved.”

It seemed odd to focus on UDL because at least as late as 2014, the UDL website that Graham and Cologon linked to, in a reversal of the usual scientific method, was asking visitors if they could point to any evidence to support the various UDL provisions. And subsequent to the Graham and Cologon piece, a meta-analysis found that while UDL improves the ‘learning process’, the, “impact on educational outcomes has not been demonstrated.” Nevertheless, here were two education researchers confidently recommending UDL. On what basis?

In a sense, the evidence perhaps did not matter because Graham and Cologon issued a stern warning:

“Access and participation are the right of every student in Australia. The Disability Standards for Education are intended to support educators in understanding and implementing their obligations under national and international law. Differentiation is a requirement under these standards and necessary to provide quality education for all.”

I dispute that differentiation of the kind described in this piece is required by law. It is a conflation of the requirement to make ‘reasonable adjustments’ so that students with a disability may participate in class and the wider concept of differentiation. However, when I attempted to critique this piece, I was repeatedly referred to the Disability Standards (and sometimes even United Nations conventions) and Graham and her supporters interpreted my criticisms as a call to break the law. Graham and I blocked each other on Twitter around this time and Graham has since made a defamatory statement about this.

In the intervening years, my position has not changed, even as I have learnt of more evidence relevant to the debate. In 2018, I wrote a blog post for the Centre for Evaluation & Monitoring (CEM) in the UK where I made a number of points, including one on the ambiguity of what differentiation means:

“So ‘differentiation’ potentially covers a highly diverse range of practices; some we are pretty sure won’t work, some where the evidence is ambiguous and some that may be more promising. In this case, does the term have much utility?”

I had come to understand that completely opposite approaches, such as allowing a child to record audio rather than write an explanation versus giving the child intensive writing support, could both be described as differentiation. I expanded upon this argument for a 2018 Keynote at the Making Shift Happen conference in Amsterdam. And I revisit the argument for a chapter in my new book, The Power of Explicit Teaching and Direct Instruction.

I was also aware that Graham and colleagues were working on a meta-analysis that would, once and for all, demonstrate the effectiveness of differentiation. In 2018, in another piece for The Conversation, Linda Graham, Ilektra Spandagou and Kate de Bruin again advanced the cause of differentiation. This time it was in response to the “Gonski 2.0” review and the review’s recommendations that were supportive of differentiation. This time, the evidence cited was a study involving Middle School Science (which I have critiqued in my new book). When I questioned this evidence in the comments, asking why the authors relied on a study of this kind rather than a systematic review, Graham replied, “We are in the process of writing a systematic review ourselves.”

However, no such systematic review has materialised. According to the references in this study, a paper titled, “What exactly is Differentiation and why is it so poorly understood? A systematic review,” by Graham, Davis and Spandagou was presented in 2018 at a conference, but I cannot find the paper itself. A European Educational Research Association session also refers to a forthcoming paper, but interestingly also explains that there are many definitional issues in researching differentiation and a paucity of certain types of research.

Which brings us to a new paper that has been published, written by Graham, de Bruin, Lassig & Spandagou. It is labelled as a ‘scoping review’, although it uses a systematic method that covers papers published in the 20-year period from 1999-2019 from which the authors find 34 studies that meet their inclusion criteria. The abstract begins:

“The use of a pedagogical practice known as ‘differentiation’ has become more common over time as educators have sought to respond to increases in the diversity of students enrolling in their local school. However, there are now so many misperceptions and definitional inconsistencies that it is difficult to know what is being enacted in the name of differentiation or indeed what is being researched internationally.”

So, the definitional issues I have long identified are now being centered. And exactly how can we claim people have ‘misperceptions’ if the definition is all over the place? Who are the true holders the right perceptions? The authors?

The paper continues, telling a sorry tale of studies that don’t really demonstrate much about the effectiveness of differentiation. Most of the studies the authors could find are just surveys. Hardly any involve secondary school students. There are unsurprising findings such as that, “teachers who believe in the value of differentiation and its role in addressing equity concerns are more prepared to engage in it,” which do not address the question of whether differentiation works. Even then, the studies are, “undermined by methodological weaknesses.” The authors note that the, “…gap in the research on differentiation signals a potential problem, given the ubiquity of differentiation as a concept and its adoption by education systems and accreditation systems.” Well, yes.

The authors conclude that, “The diversity of focus… prevents comparison of findings and weakens the evidential basis to make claims of either differentiation’s effectiveness or indeed its ineffectiveness.” OK, so we haven’t found evidence of the Loch Ness Monster but neither have we found evidence that the Loch Ness Monster does not exist. So there’s that to hold on to.

This sequence of events, discussions and debates raises a number of points.

Firstly, we have gone from an ‘explainer’ about differentiation written by researchers to a point where the same researchers admit that there are actually a plethora of interpretations of what it means. Given their prior commitment to differentiation, Graham and colleagues should be commended for the intellectual honesty and integrity with which they have researched this question.

We also need to move on from the pantomime where anyone who questions differentiation is accused of incitement to break the law. It is now established that there may be legitimate questions and concerns.

And finally, why are we pinning so much on differentiation at the policy level when the evidence base is so thin? And why are we teaching trainee teachers that must differentiate? Why is it in our teaching standards?

I’ve been involved in this debate for at least six years. It is now time to think again.


Explaining Finland’s educational decline

Regular readers will be familiar with the phenomenon of Finland’s continuing decline on international assessments such as the Programme for International Student Assessment (PISA) run by the OECD:

Commentators seek to explain this decline in a number of ways. Pasi Sahlberg, former policy advisor in Finland and now with the Gonski Institute in Australia, attributes it to the distraction of digital devices (In August I debated Sahlberg on the educational lessons we may draw from COVID-19 and you can catch that debate here). And in this post, I critically examined the suggestion that a decline in effort from Finnish students is the cause.

A newly published PhD thesis by Aino Saarinen sheds more light on the possible causes. Saarinen examined data on Finnish students from the 2012 and 2015 rounds of PISA with a few questions in mind: Were scores correlated with self-directed learning practices, use of digital learning materials in schools or participation in an early learning programme? Also, was variance in learning outcomes, which is apparently increasing in Finland, associated with any of these factors?

Similar to findings made by the OECD themselves, as well as by researchers using OECD data to examine a number of different countries, Saarinen found:

“Frequent use of self-directed teaching practices or digital learning materials at school were associated with students’ weaker learning outcomes in several knowledge domains. Instead, frequent teacher-directed practices were related to students’ higher learning outcomes. Moreover, frequent use of self-directed teaching practices or digital learning materials had more negative impact on students’ learning outcomes in students with (vs. without) risky background.”

Saarinen found little impact from participation in early learning.

Of course, these results do not directly address the question of what has caused the decline. It is possible, for instance, that there was even more self-directed learning in, say, 2006. However, this seems unlikely given the qualitative accounts we have of how Finnish education has changed over the last 20 years or so (see e.g. here and here) and the more recent turn toward approaches such as ‘phenomenon-based learning‘.

And there are limitations on drawing too many inferences from the questionnaires that PISA uses to assess the practices that students are exposed to. They strike me as a little eccentric, with potentially overlapping concepts often present in different constructs. But then, nothing is ever perfect. It would be great to run a statewide randomised controlled trial, but that’s not going to happen and so we have to make our inferences as best we can from the data as it stands.

If we do consider the data as it stands, the discussion highlights – again – that constantly pointing to Finland as an example of educational excellence is flawed. It is a sign either of ignorance of the last fifteen years of data, self-deception or worse.


Emina McLean


Emina McLean is an Australian language and literacy expert. In this episode, she talks to Greg Ashman about her training as a speech language pathologist, her imminent career move to a brand new primary school in Melbourne, the science of reading and teacher professional development. Along the way, Emina and Greg discuss the controversies that surround literacy teaching, such as the recent blog post by Diane Ravitch complaining about the term the ‘science of reading’ and New South Wales’ decision to roll-out a phonics screening check in all public schools, as well as the best and worst ways of changing people’s minds and hopes for the future.


Absolute classic

Yesterday, there was an extraordinary rant on Twitter full of random CAPITALISATIONS by an American poet, translator, author and education professor called David Bowles. Here is a link to the thread. It is not viewable at present because Bowles seems to have protected his tweets, but you can see the river of effusive praise he received for taking such a bold stance. If you are wondering what it is all about, I had the following still cached on my phone (I have covered up the swears in consideration of those of you with a nervous disposition):


I think it is important to point out that the thread did not appear to be in response to anything in particular and the object of Bowles’ shoutiness appears to be all those of us who think that there is value in a canon of classic literature and that children should be expose to this canon at school.

I have a number of thoughts on this. Firstly, if we are proposing that there is some choice to be made between good contemporary literature aimed at young adults (YA) and children reading the classics, then this is a false choice. In my view, children should read the former for pleasure and the latter for class. Yes, we have a big problem with encouraging reading in young people. This is not solely the result of ignoring the evidence on systematic phonics, but paying attention to this evidence would be a good start.

I also think it is wrong to assume that children can only be interested in stories about people who look like them, are their age, live in similar neighbourhoods and have similar problems. Such books may be a gateway for some, but this idea ignores the wild popularity of escapist fantasy fiction. Children are capable of empathising with a wide variety of characters because children are human beings and empathy is our thing. If you ask a child whether they want to read Jane Austen, they will probably say, ‘Who’s Jane Austen?’. It is our job to open the world up to them and give them possibilities. Sure, they may dislike some or even most of what we serve them, but education is about expanding minds, not narrowing them.

And finally, there is the argument about core knowledge. Think of a book like 1984 and imagine being unfamiliar with its concepts. You wouldn’t understand what a person was talking about if they referred to a ‘Big Brother’ state. Some key texts function like this in our modern culture. A lack of such knowledge can be exclusionary. The kid who goes to university knowing nothing of Orwell will feel out of place in a first year lecture where such knowledge is assumed. I suppose the next logical step is to burn down the canon at the academy too.

And I do understand that for historical reasons, when you attempt to construct a canon of key literary works in a modern liberal democracy, your list tends to be overrepresented by white European men. So, yes, I think there is an argument for consciously trying to balance this, as I outline at the end of Chapter 1 of my new book, along with the mechanisms I propose for doing so.

However, there is perhaps a more fundamental point. Clearly, there are those who believe with capitalised, sweary fervour that it is far better to feed kids literary candy than proper books. I would not want them imposing this view on my children and I imagine they would not want me imposing my view on their children.

When positions on education are so diametrically opposed, when any possibility of rapprochement is eradicated by uncompromising rhetoric, is there any alternative to school choice?