You know what differentiation is. How could you not? It’s been a fundamental part of teaching for as long as most in the profession can remember. It’s what we do, day in, day out. It’s how we ensure that all our students manage in school and all our students excel. It’s been a topic in most observation feedback sessions, most departmental meetings, most planning workshops and many, many social media debates. It’s obvious what it is. It’s …

You give your definition and explain what it might look like in practice. And the teacher you are talking to screws up their eyes slightly, frowns a little and says, “Oh, really? Well, for me, it’s more …”

Puzzled, you both ask a third teacher. A fourth. A fifth. You ask your line manager. You ask their line manager. You ask your Facebook friends and your Twitter followers and your WhatsApp group. You ask any teacher you can find within shouting distance.

Suddenly, you realise something very disturbing: all this time, all those discussions, all that planning, and no one really agrees on a single definition of differentiation or what it should look like.

So, what the hell do we do now?

First, we must ask ourselves whether it really matters if we don’t agree what differentiation looks like.

Your immediate answer is likely to be “No, we all pretty much know roughly and we wouldn’t want anything too prescriptive, would we?”

But let’s say you are receiving feedback following an observation and you are told to differentiate more. Next time you are observed, you are doing exactly that, and yet the observer again picks you up on it, because the observer and you have entirely different ideas about what differentiation means - and both appear to be valid. This will only end badly.

Or what if you are looking, as a department, at how you could differentiate more? The head of department sends you all off to try to do just that. If you don’t all agree what differentiation is or what it looks like, that exercise would be entirely pointless and implementing any meaningful changes as a result would be challenging.

Finally, it matters because, under the umbrella of differentiation, much bad practice can find shelter. We need to make the umbrella transparent so we can see all the potential teaching ghouls lurking beneath, and we can begin to unpick what works and what does not under the differentiation banner.

So, yes, not agreeing matters. But do we really disagree as much as my initial foray into the topic indicated? I decided to find out.

I’m on Twitter. I have a reasonably sized network of teachers within my reach. So I asked my fellow tweachers what they thought differentiation was. The responses could be grouped into four ideas:

• Ensuring access to learning.
• Providing high challenge, low threat in a way all learners can succeed.
• Scaffolding without reducing expectations.
• Inclusive teaching for inclusive learning.

At first glance, these seem quite cohesive as a set of definitions. You could contain these four points under the central idea of knowing your learners, and enabling them to access learning and be successful. It’s simple, it’s snappy, it’s as good a definition as any.

But when you try to roll this out in practical terms, you realise you can do pretty much anything under that definition and those four tenets, and call it differentiation.

A range of worksheets aimed at different grades within an objective? Tick. Giving students a choice of difficulty via some “chilli challenges”? Tick. Marking every student’s book in their favourite coloured pen for more personalised feedback? Tick. Letting Toby demonstrate his understanding of Newton’s third law using mime as he’s recently been researching Marcel Marceau? Erm, well, strictly speaking, tick.

So, what exactly are we asking teachers to do when we say “differentiate”, if we work with this definition? Everything, and almost anything. When you look through the copious research on the topic (which I have done on your behalf), you find the same issues. Whether it is the work of Carol Ann Tomlinson - the apparent oracle of differentiation - or the countless other experts on it, you don’t find any real agreement and you end up coming away with more questions than answers.

This troubled me. I had spent nearly a week looking into it, and the very word “differentiation” had morphed into something nebulous and almost meaningless.

I couldn’t quite believe it to be true. So I went bigger. I put out a survey. And 5,000 lovely people responded.

I asked them what techniques they had used as vehicles of differentiation in the past year. I hoped I would get a small collection of interventions that were backed by research, that were sensible and that could be grouped under a single definition.

This did not happen. In the name of differentiation, much wrong is, probably unintentionally, being done. Here are the top five (of many) and my own thoughts on each.

1. Differentiating oral questioning

First up is the targeting of individual students with questions differentiated by difficulty, which about 80 per cent of respondents reported doing.

Pros: The goal of providing stretch and challenge for the most able can be satisfied by targeting those students in this bracket with the most devilish and challenging of questions. It will force them out of their comfort zone. Equally, you can ask students with lower prior attainment easier questions, as you methodically work your way through Bloom’s taxonomy, to help build their confidence and aid motivation.

Cons: Quite simply, doing this risks holding children back through low expectations. Asking different questions to different individuals risks widening the attainment gap if all students aren’t expected to perform the same thinking. We should ask every question to the group - cold calling (see the writing of educationalist Doug Lemov) is an effective vehicle for this - so every student is involved in even the most challenging of cogitations.

If students know there is a reasonable chance of being required to answer any question, they will know they are part of an inclusive environment. Equally, teachers need to be able to respond to the emerging needs of their classes, so utilising techniques that require all students to respond is paramount.

2. Differentiating feedback

Similarly popular is writing differing feedback for students to respond to - a stalwart of many a marking policy. So, while Martin might be told to work on his spellings and his use of metaphor, Maryam might be instructed to watch her punctuation and look again at her essay structure.

Unsurprisingly, more than 90 per cent of the English teachers surveyed used this method, with the percentage dropping across all other subjects down to 77 per cent in science and only 64 per cent in maths.

Pros: This approach can certainly help learners improve the piece of work that has been marked and fed back on. They can see where they made their mistakes and correct them. They get feedback that is personalised to them - something students do value - and time to respond.

Cons: The main one here is workload. The Education Endowment Foundation’s 2016 review of the evidence on written marking found “a striking disparity between the enormous amount of effort invested in marking books, and the very small number of robust studies that have been completed to date”.

The review from the EEF also referred to the government’s 2014 Workload Challenge survey, which “identified the frequency and extent of marking requirements as a key driver of large teacher workloads. The reform of marking policies was the highest workload-related priority for 53 per cent of respondents.”

Essentially, there is no evidence that individual written feedback makes much difference, and there is a lot of evidence that it causes huge workload issues. And if individual students have responded to individual comments, and corrected and improved their work, then it’s only right and proper to check these … individually.

3. Tailoring to prior attainment

Next up is the idea of using prior attainment data or target grades to determine starting points on tasks, or using them to give different tasks, or allowing students to choose the tasks they complete. Around 60 per cent of respondents used these approaches.

Here, I found a notable disparity between primary and secondary respondents, with those working with younger pupils being much more likely to utilise these particular techniques; different starting points, for example, had a sizeable difference in likelihood, with primary teachers at 70 per cent and secondary only at 40 per cent.

Presumably, this stems from the wider range of pupil needs found in a typical primary classroom compared with those found in secondary, where streaming is more common.

Pros: Starting pupils at “their level” means that they can learn at their own pace and are less likely to struggle to progress. The higher-attaining students can skip through the easier fundamentals and attempt more challenging work whereas those learners who progress more slowly aren’t rushed through content.

Cons: This can lead to lowered expectations of pupils perceived as less able and could mean that opportunities for those faster- progressing pupils to overlearn, improving longer-term retention of knowledge, are missed. As with marking policies, this type of approach to differentiated instruction has been identified as increasing workload, too.

Finally, the research behind the new Ofsted framework suggests that this is low-impact differentiation: “In-class differentiation, through providing differentiated teaching, activities or resources, has generally not been shown to have much impact on pupils’ attainment. In Scheerens and Bosker’s (1997) meta-analysis of school effectiveness research, for example, this factor showed no or a very weak relationship with pupils’ outcomes. Hattie (2009) likewise found the effect of differentiation to be among the weakest in his influential work on ‘visible learning’.”

4. Scaffolding

About 90 per cent of teachers reported that they utilised the approach of having high expectations of all and then giving different levels of support or scaffolding to individuals within a class.

This can manifest in a plethora of ways, usually beginning with a worked example - or the explanation of a concept - being modelled by the teacher. This provides the initial cognitive support.

Students can then be provided with frameworks that support the completion of similar problems, often with slight variations to add some increased difficulty as they move towards independence.

Pros: This is reassuring, given that in Barak Rosenshine’s Principles of Instruction a desirable trait of an effective teacher would be to “provide students with scaffolds, or instructional supports, to help them learn difficult tasks”, with the warning to ensure that this support is withdrawn as competence increases to avoid students becoming reliant on such structures.

This is not a proactive approach but must be reactive depending on performance within a particular domain of knowledge. Aspects of cognitive load theory, such as the worked example effect (Clark, Nguyen and Sweller, 2006) followed up by completion problems and fading structures also complement this method of delivery.

Cons: Rosenshine suggested learners should have a success rate of around 80 per cent of a task, but this is very tricky to get right in a classroom where learners have varied needs and abilities.

The chances are that, on the independent tasks, a few students will be getting 100 per cent and some might only ever make it to 50 per cent in the time you have. The average might be 80 per cent: is that OK? Or does that mean you’ve effectively met the learning needs for only those few who conveniently get two of the 10 questions wrong?

5. Differentiation by outcome

Mercifully, less than 30 per cent of respondents reported using learning outcomes organised hierarchically with the terms “must, should, could” or “all, most, some” to differentiate where students may end up at the end of the lesson.

The premise of using differentiated outcomes is based on the fact that students will learn and progress at different rates. We therefore need some way to let learners know our minimum expectations over the next 50 minutes or so, given that they will acquire differing amounts of knowledge.

Pros: By the time the bell rings, all learners will be able to leave the room knowing they achieved something (theoretically).

Cons: The tone of this approach is all wrong, and the potential repercussions are incredibly damaging. Many a teacher has heard a student say, “I’ll just be one of the ‘shoulds’ today” in lessons “differentiated” in this way, and this is indicative of a culture of low aspiration for, well, “all, most or some” of the learners.

Luke may have built a reputation of working steadily, despite his low prior attainment, and is getting by comfortably on a diet rich in “musts”.

But Luke could thrive on a varied intake of levelled outcomes if they were delivered in the right way; like blitzing the mushrooms in the food processor before adding them to the bolognese sauce so the children don’t notice them but they still benefit from the B vitamins.

Beyond these five headlines is so much variation it would make you wince. And I don’t want to make you do that, so take my word for it.

So, what now?

It’s true: we are all doing myriad things in the name of differentiation and a lot of it is dubious in effectiveness and catastrophic in terms of workload. And this huge variation may well be down to the fact that, in my hours of research, I couldn’t find any overarching, works-everywhere, easy-to- implement and economically viable approach that all teachers could utilise. There are no easy answers here: differentiation looks different in every classroom because no one really knows what it is or what works.

What struck me, too, though, was that our view of differentiation is so narrow. No one seemed to be researching or talking about the wider differentiation that happens in schools: the phone call from the attendance officer that gets a student into school; the meeting with the member of the pastoral team that resolves an issue and gets a student back to class; putting students on report to support them in improving their behaviour; having a breakfast club that enables students to start the day with enough energy to learn; the learning mentor with hours of training on working one-to-one with autistic students. The list of possible examples of differentiation taking place outside the classroom or done by non-teaching staff is almost endless.

The way forward, I fear, is a controversial one: I would scrap the word “differentiation” from our educational lexicon except when pertaining to special educational needs and disability (SEND).

SEND students have pronounced barriers to learning and the differentiation strategies suggested on educational, health and care plans (EHCPs) truly enable access to the curriculum and much more. Autistic students may benefit from being able to leave lessons two minutes early to avoid the overstimulation of a busy corridor, making them calmer and better able to learn in the next lesson.

Hearing-impaired students may have a signer or need to sit near the front to better enable lip reading. Most EHCP strategies have the ultimate goal of developing the student over a long period of time to improve their chances of thriving once they leave education and move into the world of work.

These aspects of differentiation can literally change the lives of young people. Go and visit a special school where inclusion is a reality, not just a nice thing to put in the school prospectus. It does not have to mean 30 different lessons but a sensible, adaptive way of teaching that gives every child the best opportunity to succeed. Effective SEND teaching has much to teach us.

And for everyone else? We shouldn’t talk about differentiated questions or differentiated assessments or differentiated feedback any more. We should ban the word “differentiation” because, ultimately, it’s not enabling us - it is distracting us. It has become almost an excuse not to know our pedagogy well, not to delve deeply into how our questioning, or behaviour management or scaffolding would impact each individual, not to see those non-academic adaptations as being part of what we are supposed to do. It’s a catch-all that catches no one.

Instead, let’s concentrate properly on what a workload-acceptable, research-based, effective approach to teaching 30 children of varying levels might be. Students don’t need every aspect of their learning slightly differentiated; they just need us to know, recognise and respond to the slightly different ways in which they learn.

Ian Taylor is lead teacher at Trinity Academy Sowerby Bridge, West Yorkshire, and tweets as @MrTSci409

This article originally appeared in the 8 November 2019 issue under the headline “We need to talk about differentiation”