Know Your Limits!


This extended post challenges the argument that all learners can be high attainers.

It sets out the various strands of this argument, highlighting the weaker links and illustrating them with the assistance of two case studies, both branded school improvement strategies.

It uses PISA 2015 data to demonstrate that none of the world’s leading jurisdictions is close to a position where more than a minority of learners is reaching high performance thresholds.


Summarising the argument

Not infrequently one encounters the statement that all learners are capable of excelling in their education, of becoming high attainers.

This may be articulated in relation to:

  • The learner, where it is typically associated with the dangers of ‘labelling’ individuals as (potentially) successful or otherwise. Unless we maintain consistent and universally high expectations of what learners are capable of achieving – so the argument runs – too many will underachieve, and so undershoot their virtually limitless potential.
  • The school, where the argument is often encountered in the context of differentiated teaching, especially the pros and cons of between-school and within-school selection. Some commercial school improvement strategies are also framed to maximise the incidence of high attainment within an institution.
  • The system, where it is argued that those countries and jurisdictions that excel in international comparisons studies – especially PISA – have much larger proportions of high attainers than does England, and are successful precisely because they start from the assumption that all learners are capable of high attainment (often bolstered by a ‘Confucian’ attitude towards success through effort).

Sometimes this argument alternates between ‘all’ and ‘most’ (meaning the substantial majority) or alternatively ‘a much higher proportion’ than presently qualify as high attainers.

When ‘most’ is used it is rarely, if ever, quantified. These educators invariably seem unable or unwilling to fix the proportion of the learner population they envisage as within scope of their ambition, presumably for fear of being accused of ‘writing off’ the excluded minority.

They are also coy about quantifying a feasible degree of improvement from the current position and, worst of all, about defining the high attainment benchmarks that they believe all learners are capable of achieving.

If a benchmark is set in such a way that it counts in almost 50% of all learners (as does Ofsted’s highly unsatisfactory ‘most able’ definition) many more will be capable of reaching that threshold than, for example, the new KS2 ‘high standard’ benchmark (secured by 5.4% of the KS2 cohort in 2016), or an equivalent Attainment 8 score (say 70+).

Ultimately the lesson from international comparisons studies is that, even in the world’s most successful jurisdictions, high attainers remain a minority and the ‘long tail’ has not been eliminated.

These jurisdictions have successfully moved the tail so that it begins higher up the attainment distribution than ours, but the tail continues to wag nevertheless.

The rate of improvement in leading jurisdictions is not sufficient for this position to change any time soon, if ever (though it must be acknowledged that recent backsliding might be associated with the introduction of computerised testing).

While some exceptional schools will buck the national trend, I am not persuaded that the majority of institutions can ever be successful in these terms.

I am not persuaded that any given school improvement strategy can consistently lift all (or even ‘most’) learners above a suitably challenging high attainment threshold, or eliminate the ‘long tail’ trailing behind. Not without selective admissions, anyhow.

Some progress can be made by teaching other students to emulate the learning behaviours of high attainers but, provided that those high attainers are also supported to become the best that they can be, the distribution of attainment will remain as wide as ever.

And limiting high attainers’ further progress is not an acceptable educational strategy, whether at national or institutional level.

Finally, too many of those wedded to this argument tend to assume that those who do not subscribe to their beliefs are clinging to an outmoded educational philosophy, cannot see beyond Twentieth Century notions of inherent, genetically-predetermined ability.

But, if one takes the trouble to distinguish the concepts of attainment and ability, it is perfectly possible to accept that ability is fluid and yet reject the notion that all learners are capable of high attainment.


The relationship between ability and attainment

Much debate in this field is bedevilled by confusion between ability and attainment. These terms are often used interchangeably, yet they must be kept distinct.

While definitions abound, it seems to me that:

  • Attainment is a judgement about what the learner can achieve now in a particular subject or across a defined group of subjects, for example a GCSE grade 9 or an Attainment 8 score of 70+. It is important to distinguish prior attainment, which is what the learner has already achieved at a previous assessment point, for example a KS2 average scaled score of 115. There are significant gaps between the proportions of advantaged and disadvantaged learners at all attainment thresholds, including excellence gaps between advantaged and disadvantaged high attainers. It is in the interest of disadvantaged high attainers to close these excellence gaps; it is not in their interest to close performance gaps between low and high attainers. Prior attainment is a very strong predictor of future attainment, but not infallible, particularly for those from disadvantaged backgrounds.
  • Ability is a judgement or prediction of what the learner may be capable of achieving in the future, derived from evidence of their existing capabilities. This may include (prior) attainment evidence, typically confined to English and/or maths, but will normally foreground evidence of generic capabilities, such as verbal, non-verbal and visual-spatial reasoning. Eleven-plus selection and CATs tests are prominent examples of ability-based assessments. Selection by ability is supposed to reveal unfulfilled potential identifying, amongst others, those most likely to overcome current under-achievement, whatever its cause. Although there is significant overlap, in every given assessment a proportion of high attainers will not be able to demonstrate high ability and a proportion of middle attainers will replace them. But, without intervention to level the playing field, selection by ability may not be more equitable – it can be even more biased towards those from advantaged backgrounds than selection by attainment.

Sometimes alternative terms are used, including ‘potential’ (for ability) and ‘performance’ (for attainment). Unless these are properly and differently defined – for they can mean something subtly different – they may be assumed to be synonymous.

The vagaries of selection by ability are such that it is often more straightforward to assume that all high attainers are a subset of those with high ability, those that have already successfully combined and converted ability, opportunity and effort into positive educational outcomes.

That is my preferred position, even though high attainment may occasionally be rather more attributable to effort and opportunity than to ability.

It seems to me reasonable that learners should be able to compensate for ability by prolonged and intense effort, provided there is evidence that it can be sustained without the crutch of relentless private tuition; far less so that lack of opportunity may undermine both ability and effort, including lack of opportunity to prepare adequately for a selection process. Hence priority must be given to securing equality of opportunity.

But, at the risk of repeating myself, let me emphasise that by no means all highly able learners are high attainers – there is extensive underachievement, some of it attributable, at least in part, to socio-economic disadvantage.

Some learners with ‘spiky profiles’ might be highly proficient in English but not maths (or vice versa). Others’ underachievement might be discernible from the discrepancy between, say, their CATs scores and KS2 scaled scores. Others still will have a variety of special needs.

Some selection processes may be designed to accommodate such learners; others may be designed deliberately to exclude them, to identify all-rounders. The latter are more likely to place the disadvantaged at a further disadvantage.

I would like to be able to quantify the overlap between attainment and ability – to supply the venn diagram with percentages affixed – but this varies according to the thresholds applied.

We do know that many of those admitted to grammar schools are not high prior attainers on the measure used in the primary performance tables which, though still broad, is not quite as generous as Ofsted’s, counting in 31.3% of pupils in the GCSE cohort at state-funded mainstream schools in 2016.

The data reveals that 45% of grammar schools’ 2016 GCSE cohorts contain 90% or fewer high prior attainers. Eight schools have fewer than 70% of high prior attainers in their GCSE cohorts, with one school down as low as 44%. The average percentage of high prior attainers across all 163 grammar schools is 88%.

So roughly one in every eight grammar school pupils in the 2016 GCSE cohort was not in the top third of high prior attainers at the end of KS2.

These finer points aside, ultimately it is part of all schools’ educational task to help all learners to try to overcome underachievement, whatever its causes, so converting high ability into high attainment. To accomplish this task schools must provide educational opportunity and instill the capacity for effort.

One might reasonably argue that it behoves all schools to maximise the proportion of learners that surpass defined high attainment benchmarks.

But it is open to question whether all learners – or even the substantial majority of learners – are capable of surpassing those benchmarks. It might be preferable to frame the ambition in terms of helping all learners to become the best that they can be, recognising that some will always fall short of such thresholds.


What international comparisons studies measure

International comparisons studies are attainment-based assessments, though the most prominent take somewhat different approaches

According to the OECD:

‘Every PISA survey tests reading, mathematical and scientific literacy in terms of general competencies, that is, how well students can apply the knowledge and skills they have learned at school to real-life challenges. PISA does not test how well a student has mastered a school’s specific curriculum.’

England’s PISA 2015 National Report confirms a strong (but nevertheless unquantified) correlation between PISA scores and GCSE grades, adding:

‘Whereas GCSEs examine pupils’ knowledge of specific content and application of specific techniques as defined by national curricula, PISA measures pupils’ ‘functional skills’ – their ability to apply knowledge to solve problems in real world situations. This is also in contrast to other international studies, such as TIMSS, where the assessment framework is aligned to a set of content agreed by the International Association for the Evaluation of Educational Achievement (IEA) who oversee the study.’

England’s TIMSS 2015 National Report describes that assessment as:

‘…more focused on pupils’ knowledge and understanding of curriculum content than PISA, which assesses pupils’ ability to apply their science, maths and reading skills to everyday situations.’

This post relies exclusively on PISA outcomes.

PISA participants are ranked on each assessment according to their mean score. The average mean score across OECD countries is 500.

Mean scores are also divided into proficiency levels for each assessment. In maths there are six levels of proficiency from 1 (the lowest) to 6 (the highest). In reading and science level 1 is divided between 1a. and 1b. Level 6 was introduced for reading only in 2009.

There are descriptors for each level in each assessment. For example, level 6 in mathematics is described thus:

‘Pupils can conceptualise, generalise and utilise information based on their investigations and modelling of complex problem situations, and can use their knowledge in relatively nonstandard contexts. They can link different information sources and representations and flexibly translate among them. Pupils at this level are capable of advanced mathematical thinking and reasoning. These pupils can apply this insight and understanding, along with a mastery of symbolic and formal mathematical operations and relationships, to develop new approaches and strategies for attacking novel situations. pupils at this level can reflect on their actions, and can formulate and precisely communicate their actions and reflections regarding their findings, interpretations, arguments, and the appropriateness of these to the original situation

In broad terms, level 2 is considered a baseline for proficiency, level 3 is the median performance level in most jurisdictions, while levels 5 and 6 are considered ‘top performers’, or high attainers.

In relation to science, the subject focus in PISA 2015, top performers are described thus:

‘Students at these levels are sufficiently skilled in and knowledgeable about science to creatively and autonomously apply their knowledge and skills to a wide variety of situations, including unfamiliar ones.’

PISA also records the percentage of students who achieve Level 5 or 6 in all three domains – science, maths and reading.

Finally PISA identifies the proportion of ‘resilient students’. This group comprises learners who are:

‘…in the bottom quarter of the PISA index of economic, social and cultural status in the country/economy of assessment and perform[s] in the top quarter of students among all countries/economies, after accounting for socio-economic status’.

The performance measure for resilience is confined to the focus subject for the relevant PISA cycle – and those subjects alternate.

The latter part of this post compares England’s performance on PISA 2015 assessments with the scores registered by the world’s best-performing jurisdictions.


The ‘all can attain highly’ argument

Those educators who argue that all learners can be high attainers typically advance some or all of the following arguments:

  • Irrespective of whether we choose protectionism or globalism (and of the outcomes of Brexit) we face an economic imperative to build our national stock of human capital. As part of this effort, it behoves us to maximise the number of high attainers at all stages within our education system, so matching our international competitors and emulating the ‘Asian Tigers’ (amongst others) who are actively pursuing this strategy.
  • We routinely under-estimate the abilities of many (most?) learners, especially those from disadvantaged backgrounds. A much larger proportion than we assume is capable of exceeding defined high attainment benchmarks.
  • Too many teachers, learners and parents also believe, wrongly, that ability (as opposed to high attainment) is vested predominantly in the children of the highly educated, wealthy and successful, when it is actually distributed much more evenly within the learner population, independently of gender, ethnicity and socio-economic background – and of the complex interaction between those three variables. The tendency for more advantaged lower attainers to overtake less advantaged higher attainers during KS3 and KS4 is not set in stone.
  • Too many teachers, learners and parents believe, falsely, that ability is genetically pre-determined when it is highly malleable, environmentally dependent, develops at different rates in different learners and can, to some extent, be taught.
  • We must model the performance of successful high attainers and teach all learners to emulate this. That involves acquiring and sustaining a ‘growth mindset’; learning perseverance through the development and application of ‘grit’ or resilience; and practising the application of a variety of pertinent meta-cognitive skills.
  • Although there is evidence to the contrary, it is defeatist to argue that education can make only a relatively small difference to closing socio-economic attainment gaps, including excellence gaps between advantaged and disadvantaged high attainers. The best schools already ‘beat the odds’ and it is legitimate to expect that, with the right support, all (most?) other schools can emulate them.
  • In differentiating learning by ability (or even by attainment) – and by otherwise distinguishing or ‘labelling’ high, middle and low ability learners – we create a self-fulfilling prophecy. Consequently these practices are to be avoided. 
  • We must instead maintain consistently high expectations of all learners, pushing beyond the imperative to ‘become the best that they can be’ by expecting them to achieve, and preferably exceed, defined high attainment benchmarks. It is feasible to expect all learners can become high attainers or, failing this, that the vast majority can do so.


My take on this

I am more than willing to accept the first five points in this argument, all of which featured to some extent in the national gifted and talented programme back in the day. They are equally relevant to the contemporary ‘most able’ paradigm.

I like to think that I have made some small contribution to the human capital argument, having written The Gifted Phoenix Manifesto for Gifted Education and The Economics of Gifted Education Revisited (both March 2013).


The economic imperative

Perhaps I might be permitted to translate the Manifesto’s Premiss from a ‘gifted education’ paradigm to a ‘most able’ paradigm. The adjustments are marginal and leave the substantive argument unchanged:

  • Effective ‘most able education’ involves maintaining a balance between excellence – raising standards for all – and equity – raising standards relatively faster for those from disadvantaged backgrounds.
  • At national level, excellence might be measured by increasing the proportion of learners achieving the high achievers’ benchmarks in international comparisons studies such as PISA, TIMSS and PIRLS. This can be described as increasing the ‘smart fraction’.
  • Similarly, at national level, equity might be measured through a country’s success in closing the excellence gap between the performance of high achievers from advantaged and disadvantaged backgrounds. This helps to increase the ‘smart fraction’ as more disadvantaged learners reach the high achievers’ benchmarks.
  • Taken together, these two actions contribute significantly to national efforts to increase the supply of highly skilled human capital which has a significant positive impact on economic growth.
  • Efforts to increase the ‘smart fraction’ and narrow the excellence gap must begin during – and be sustained throughout – compulsory schooling, through a coherent system-wide strategy. This should link seamlessly with continuing efforts within the national higher education system, and beyond.
  • The costs of this strategy can be offset against the much greater benefits that will accrue through stronger economic growth, so justifying the initial investment, even during a period of austerity.
  • This national investment will also generate several highly important spillover benefits, not least stronger social mobility as more learners from disadvantaged backgrounds compete on a level playing field with their advantaged peers. There are also cultural, sporting, political and ‘feel-good’ benefits. (These include improving the quality of political leadership which seems increasingly impoverished in many countries at this time, including our own.)
  • It would be wrong to focus investment disproportionately in areas such as STEM and IT, partly because other fields can make a substantive contribution to economic growth, and partly because of the important spillover benefits outlined above.

Back in 2013 I also proposed a ‘Gifted Phoenix Equation’ to accompany the premiss. This was admittedly rather tongue-in-cheek, for I am not into self-aggrandisement.

But if you insist, I will rechristen it the Dracup Equation and restate it here.

By minimising excellence gaps and maximising the smart fraction (both the size of the fraction and the average high attainment level achieved within it) we can maximise the impact on economic growth.

Min ExG + Max SF (size + score) = Max EcG

I challenged readers to debate whether diminishing returns might set in – whether the biggest ‘smart fraction’ might not be the optimal position. If there has been any development of this argument in the intervening period, I am afraid I have missed it.

In short I agree with Dylan Wiliam:

‘In the past, we have treated schools as talent refineries. The job of schools was to identify talent, and let it rise to the top. The demand for skill and talent was sufficiently modest that it did not matter that potentially able individuals were ignored. The demand for talent and skill is now so great, however, that schools have to be talent incubators, and even talent factories. It is not enough to identify talent in our schools anymore; we have to create it.’

At least, I agree with him up to that point. Where I differ is over his assertion that the solution lies exclusively in improving teacher quality. If we are to establish a national talent factory we need a much more comprehensive and targeted strategy.

Here’s one I prepared earlier.


Where I draw the line

My reservations relate to the last three bullet points in the argument set out above, which I have italicised and emboldened.

First, I am wary of the notion that education can by itself close excellence gaps between advantaged and disadvantaged high attainers – and especially that all schools can emulate the best schools and so successfully ‘beat the odds’.

I am sure that schools can make a significant difference but, unless there is a more systematic effort to bear down on the negative effects of poverty, I fear their impact will always be compromised.

As for all schools learning how to ‘beat the odds’, this seems to me a leap of faith, with relatively scant evidence to support it. Such a belief is embedded at the heart of the ‘school-led self-improvement’ movement. I wish it could be true but I doubt that it can be so.

Secondly, I have more profound difficulties with the ‘labelling’ argument.

Like many ideas in education, it is sensible if applied in moderation, yet profoundly stupid if applied too rigorously.

It is easy to see how poor schools and unskilled teachers might go about their business in a manner that undermines the imperative to maintain universally high expectations – and so damage the self-esteem of learners.

But we trust them to handle these matters more sensitively, balancing the need for effective differentiation against the feelings of the learner. We do not expect them to sacrifice overt differentiation to spare those feelings.

To impose conditions in which groups of learners can only be discussed in terms that carry no attainment-related connotations (or even ability-related connotations) is frankly a bridge too far.

If we follow this to its logical conclusion not only do we wave goodbye to a raft of potentially valuable teaching and learning strategies, we also prohibit the use of sensible descriptors such as ‘SEN’, ‘catch up’, ‘extension’ and ‘fast-track’.

Educators are forced to invent euphemisms to replace these terms, confounding communication with parents and with the learners themselves.

The linguistic police are ever vigilant. All are trapped in a Kafkaesque nightmare, haunted by misunderstanding, perpetually reinforced by the resounding echo of Chinese whispers.

Thirdly, and most significantly of all, while I accept that many more learners are capable of achieving defined high attainment thresholds, I do not accept that all – or even nearly all – can do so.

I far prefer to aim to support every learner ‘to be the best that they can be’ recognising that a minority will never achieve those defined high attainment thresholds.

The more absolute position has the strong potential to backfire when – as will inevitably happen – a significant minority fail to make the grade. What of learners’ self-esteem then? And what is the point of setting up even the very best schools to fail?

There is a bell curve – and that curve can be shifted – but it cannot be flattened entirely. To suggest otherwise is delusional.

And, as noted above, those that hold this position tend to vagueness when asked to quantify their position. They are inclined to gloss over which high attainment benchmarks they apply and, when pressed, to identify the lowest available.

Or, if they ratchet the expectation down from ‘all learners’ to ‘most learners’, they decline to supply an approximate percentage. There is a world of difference between a 51% assumption and a 99% assumption.

I shall return to this in the analysis of PISA 2015 outcomes below.

But first, in writing this post I had two case studies in mind. Both are branded school improvement strategies with their own books and websites. There are other striking similarities which I have not drawn out in the brief outlines below.

I have consulted the websites but have not read the books – it is a longstanding principle that this blog deals only in free-to-view material.

The first case study illustrates my arguments against the anti-labelling movement; the second exemplifies my concerns about ‘all/most learners’ statements, especially when these are not properly quantified.


Learning Without Limits


Learning Without Limits is an approach pioneered by Alison Peacock, formerly headteacher at the Wroxham School, a primary school in Hertfordshire.

Peacock was appointed to the headship in 2003. She evolved an approach that became closely associated with a Cambridge-based research study, originally set up in 1999, to which she contributed as a teacher researcher.

The project website describes the purposes of the original study as being ‘to re-open debates about the effects of ability labelling and ability focused practices’ and ‘to offer alternative models of practice that do not rely on ability labelling’.

The core argument is that such labelling carries with it ‘the expectation…that these differences will persist and be reflected in comparable differences in academic performance in the future.’

Teachers are encouraged to adopt as an alternative ‘transformability’, which is described in these terms:

‘The future is unknowable and unpredictable because it is being actively created in the present. Everything that happens has an effect, for better or worse, on young people’s learning capacity. By understanding the forces that shape and limit capacity to learn in the present, and through the choices they make, teachers can exploit the possibilities available to them for enabling all young people to become better learners.’

That seems reasonable enough.

This was followed by a second research project (2005-2012), a partnership between Cambridge and Wroxham School which applied ‘transformability’ to whole school development by (with my emphasis):

‘…exploring the wider opportunities for enhancing the learning capacity of every child that become possible when a whole staff group works together to create an environment free from the limiting effects of ability labels and practices. What happens when staff members jettison officially prescribed practices of predicting or pre-judging what any individual children might achieve? When they work, instead, to identify and lift limits on learning? When they replace the fatalism of ability labels with a more hopeful, powerful and empowering view of learners and learning?’

One article on the website outlines the alternative school improvement model emerging from this work, waxing lyrical about Wroxham’s development of:

‘…ways of talking about children’s learning without reference to notions of ability or levels of attainment, to ‘special needs’ or deficits, focusing instead on children’s present powers as learners, and on how these could be further strengthened and extended.’

This seems a more radical departure than envisaged in the initial project.

While that can be interpreted as advocating – albeit in rather inventive and roundabout language – a sensible focus on current attainment instead of predicting future attainment through ability-based measures, it now appears that attainment-based judgements are equally reviled.

Indeed this whole article adopts a subversive tone, suggesting that Wroxham’s practice is antithetical to the policy of successive governments, to raise standards through a robust assessment and accountability framework.

Yet Wroxham has been judged positively within that framework. Ofsted assessed Wroxham as an outstanding school in 2006, 2009 and 2013.


Recent developments

The most recent Ofsted report describes pupils’ achievement as outstanding, but there is no explicit reference to the performance of the most able learners, even though this was a national priority for inspectors by this time.

In 2014, Peacock was made a Dame for her services to education. Hardly the recognition meted out to a subversive influence! She served on several government working groups and in March 2015 was appointed to the ‘expert group’ charged with drafting an ITT core content framework.

The Schools White Paper (March 2016) said that this would:

‘…include a specific focus on stretching the most academically able pupils and cutting edge evidence on how these pupils can be challenged and stimulated to achieve the very highest standards’ (para. 6.59).

But in the event, this cutting edge evidence comprised eighteen words:

Providers should equip trainees to be able to inspire and provide extra challenge for the most able pupils.

Truly brevity is the soul of wit.

Given Peacock’s position on ‘labelling’ (and elsewhere the framework says explicitly that ‘providers should ensure that trainees…avoid labelling by group…’) this task must have placed her in an awkward quandary.

Does that help to explain why the expert group so obviously failed to satisfy this part of its remit?

Peacock resigned as Wroxham’s Executive Head in January 2017 to become Chief Executive of the Chartered College of Teaching.

Looking forward, will the College provide professional development in support of ‘most able’ education, or other provision that relies on ‘labelling’ by ability and/ attainment?




There is negligible information on the website about the impact of this programme on levels of attainment at Wroxham School, presumably because the project wishes to imply that such judgements are comparatively worthless.

Yet we know from Ofsted’s 2013 judgement that this subversive positioning is nothing but smoke and mirrors, designed to popularise the approach amongst those who hold to this ideology.

I’m sure that Peacock at least wanted to demonstrate that a ‘label-free’ approach could actually deliver better outcomes for her learners.

So I was curious to discover what the 2016 performance tables revealed about the efficacy of this approach, particularly with high attainers –  for they seem most likely to be vulnerable in an attainment-blind (as well as ability-blind) environment.

Wroxham has 239 pupils aged 4-11 on roll, 31 of which were within scope of KS2 assessment in 2016.

This is a comparatively advantaged school. Only 1.7% of pupils have a SEN statement or EHC plan, while 7.1% have SEN support; 11.5% do not have English as a first language (compared with 20% nationally). The overall FSM percentage is a mere 3.6% and the ‘ever 6 FSM’ rate is 5.7% (compared with 25.4% nationally.

Amongst the KS2 assessment cohort none were EAL, none had an SEN statement or EHC plan and 19% were disadvantaged. There was only one learner with low prior attainment (3%), plus 20 middle attainers (62%) and 10 higher attainers (35%).

Given these characteristics, the 2016 outcomes are unspectacular and somewhat inconsistent, particularly amongst high attainers:

  • Overall progress scores are +0.8 (reading), +1.2 (writing) and +0.7 (maths). All three are average in national terms.
  • Some 77% of the cohort met the expected standard across reading, writing and maths combined, much higher than the national rate of 53% and 18 percentage points ahead of the Hertfordshire average.
  • But only 6% achieved the higher standard across all three assessments, two percentage points lower than the Hertfordshire average and not much higher than the national rate of 5%.
  • 23% of the cohort achieved a higher standard in reading (compared with 19% nationally), 16% did so in writing (against 15% nationally) and 23% did so in maths (17% nationally).
  • Progress scores amongst disadvantaged learners were +0.9 (reading), 0 (writing) and +1.0 (maths). All are deemed average in national terms. Although 33% met the expected standard across all three assessments, none at all achieved the higher standard, compared with a 2% national success rate for disadvantaged learners.
  • Progress scores were noticeably different for middle and high prior attainers across the different assessments. In reading the middle attainers were at +1.6 and the high attainers at -0.3; in writing the middle attainers were at +1.2 and the higher attainers at +1.3; and in maths the middle attainers were at +0.3, while the high attainers reached +1.3.
  • While 100% of high attainers met the expected standard across reading, writing and maths, only 20% met the higher standard.

Judged on this evidence Learning Without Limits is hardly a resounding success.


High Performance Learning


High Performance Learning is a school improvement vehicle launched by Deborah Eyre, formerly Nord Anglia’s Education Director and, earlier in her career, Director of the National Academy for Gifted and Talented Youth (NAGTY).

The vision statement on the website says:

‘Everyone wants a high performing education system, but it continues to be assumed that only a minority of advantaged children can reach high levels of educational performance.

HPL believes that we should expect many more students to reach the academic levels once seen as the preserve of the very few. Indeed we should structure the education system with this in mind. Rather than creating a system which focuses on early detection of signs of failure, with pupils guided immediately into less demanding work, the system should be structured to provide the conditions that generate high performance and pupils steered towards this objective.’

The broader argument was set out in ‘Room at the Top’, Eyre’s 2011 publication for Policy Exchange which I reviewed and critiqued on the Gifted Phoenix blog – parts one and two – in May of that year.

My overall assessment:

‘I support entirely the broad thrust of the central argument. Many of the core messages are consistent with the views expressed on this blog over the past year.

I also have some fairly significant reservations about certain aspects of the treatment.’

Re-reading my arguments six years later, virtually all of my analysis stands the test of time. I will not summarise those arguments here, but instead refer you back to my original post for the detailed explanation.

High Performance Learning is a refinement of the argument in ‘Room at the Top’, translating it from a system-wide treatment into a commercial school improvement package.

The model is based on a set of principles, presented most succinctly in material supporting a High Performance Learning Award for schools.

This follows broadly the same approach as the institutional quality standard developed outwith NAGTY as part of the national gifted and talented programme.

It also owes something to NACE’s Challenge Award, which has a similar design and – to the extent that it is adopted in the UK – will become a direct competitor to the NACE award.

The need for a repurposed, overarching national standard – point three of my 10-point plan (February 2017) – has never been greater.


No quantification

The first principle set out in the Award documentation is:

‘High academic performance is an attainable target for everyone

The standard that underpins the award (which also bears some resemblance to the original quality standard developed within the national gifted and talented programme) requires a ‘mindset shift’ whereby:

‘All leaders, teachers, students and stakeholders assume that high performance for most is a possible outcome.’

This standard also includes defined student outcomes, the first of which is:

A high proportion of students achieve highly, regardless of their performance on entry – this is increasing year on year.

In a related document a ‘summary of the award standards’ begins:

‘The highest academic standards being achieved by almost everyone in the school, regardless of background or starting point.’

There are two related problems here.

The first is the inherent inconsistency between:

  • A principle that high performance is universally attainable
  • The statement that the award demonstrates that the highest standards are being achieved ‘by almost everyone in the school’
  • The statement in the standards that ‘high performance for most’ is possible
  • The need for evidence that ‘a high proportion’ achieve highly and that this proportion is increasing year on year.

The second, related problem is the absence of quantification.

Leaving aside the complexities of comparing performance in different national qualification systems, what in an English setting should count as high performance?

How is it to be defined in terms of specific attainment benchmarks?

Maybe a high standard across KS2 reading, writing and maths, achieved by 5.4% of the cohort in 2016?

How about a GCSE Attainment 8 score of 70+ in 2017, the expected outcome in 2017 for learners with a mean KS2 fine grade score of 5.6 or above?

It seems wildly optimistic to expect those levels of performance from most learners, let alone every learner.

Perhaps something rather less ambitious is intended but, if so, where is the threshold pitched?

And, moving on to the proportion that must be successful, what single percentage can we apply that embraces ‘almost all’, ‘most’ and ‘a high proportion’?

It might conceivably lie anywhere between 51% and 99%.

Then there is the requirement for year-on-year improvement, something we know is hard for schools to maintain, especially given volatility between year groups.

For how many years must this be demonstrated? Is the award forfeited if the school slips back for a single year? None of this is clear.

The lack of quantification badly undermines a standards-based award of this kind, because the most critical element of all is a moveable feast.


Negotiable standards?

Perhaps the plan is for these details to be negotiated on a school-by-school basis, but that rather defeats the purpose of a standards-based award.

According to its website, HPL is currently working with just eight ‘pioneer schools’ to achieve the standard necessary to secure the award.

Three are British international schools abroad. Of the five in England, two are selective – Sir William Borlase’s Grammar School and the Royal Latin School, both in Buckinghamshire.

In 2016:

  • Sir William Borlase’s returned an average Attainment 8 score of 71.4 and an above average Progress 8 score of +0.44. High prior attainers recorded 71.5 and +0.39 respectively
  • Royal Latin recorded an average Attainment 8 score of 70.8 and an above average Progress 8 score of +0.37. High prior attainers registered 71.1 and +0.36 respectively.

In both schools results for disadvantaged pupils were suppressed since there were only three in each cohort. The same applies to those deemed middle prior attainers at KS2 since there were only five of these in each cohort.

On the basis of this evidence, both schools are already satisfying the ‘universal high performance’ criterion precisely because they are selective.

Either there is zero challenge for them in this element of the standard, or their attainment thresholds will have to be pitched extremely high – say an Attainment 8 score of 80+ – so they do not have an unfair advantage over comprehensive schools.

But perhaps they are pursuing HPL in anticipation of radically increasing their intakes from disadvantaged backgrounds in the wake of the white paper?

At the very least, the award should require schools to satisfy a transparency condition under which they must publish the year-on-year performance thresholds they plan to exceed – and the percentage of the cohort they expect to achieve this outcome.


The PISA evidence


The basic argument

‘Room at the Top’ advocates system-wide adoption of high performance learning arguing that:

‘The Asian ‘Tiger’ and Middle East countries are already powering ahead with this kind of approach and when their education systems start to match ours in terms of sophistication they are set to outperform us at every level.’

I’m not sure that educationalists in Hong Kong or Singapore would regard their systems as any less sophisticated than England’s, but let that pass.

A post on the HPL website takes a similar line:

‘Structures come and go but the heart of high performing systems is the focus on the student and their achievement. The lesson from the Asian Tiger countries, who are holding most of the top positions in the OCED [sic] league tables, is that you do not create different pathways for different kinds of students. Everyone follows broadly the same demanding track and everyone is expected to achieve. Does this place too heavy a demand on students? No.‘

This argument was again advanced in a recent discussion, during which it was also suggested that England might learn from some jurisdictions that are already successfully demonstrating system-wide high performance.

I expressed polite scepticism on this point, given what I know of PISA performance data.

It might conceivably be argued that a handful of jurisdictions are working towards a position where a majority of their learners are high performers, but none is there yet – and the trend data calls into question whether this can be achieved with consistency any time soon.

In ‘Doing Differentiation Differently’ (February 2016), I compared the performance of Shanghai and England in the PISA 2012 maths assessment at each proficiency level.

Here is the data


L6 L5 L4 L3 L2 L1 <L1
Shanghai 30.8% 24.6% 20.2% 13.1% 7.5% 2.9% 0.8%
England 3.1% 9.3% 18.7% 24.5% 22.8% 13.7% 8.0%


And here is my commentary upon it:

‘From this it is evident that Shanghai has:

  • a much shorter absolute ‘tail of underachievement’ (3.7% at level 1 or below compared with 21.7% in England) and
  • a much larger absolute ‘head of overachievement’ (to coin a phrase) (55.4% at levels 5 and 6 compared with 12.4% in England).

So Shanghai has successfully lifted the distribution towards the higher levels. There are ten times as many level 6 performers as in England and a sixth as many at the bottom end.

But the span of attainment remains unchanged. In fact it might have increased because we know nothing about attainment beyond level 6.

I’m willing to bet that, if the PISA maths assessment had a level 7, at least some of the Shanghai population already at level 6 would achieve it. (A possible reform for PISA 2018?)

And Shanghai still has a relative ‘tail of underachievement’ (24.3% at level 3 and below), which is broadly comparable with the percentage in England at level 1 and below.

Judged in this manner, Shanghai’s approach is most definitely not managing to ‘leave no pupil behind’.’

The remainder of this post sets out the latest evidence, from PISA 2015. The first three sub-sections below deal with maths, science and reading respectively.

Each begins with a graph showing the distribution of performance by PISA proficiency level in the five jurisdictions with the largest proportions of level 5 and 6 performers, comparing these with England and the OECD average.

Singapore heads the ‘top 5’ in each case, but the supporting cast is slightly different in each assessment.

Each graph is followed by a brief commentary drawing out the key points

The final sub-section looks briefly at all-round performance and high-level performance by disadvantaged learners (so-called ‘resilient students’).




Chart 1: PISA 2015 maths: Distribution of performance by proficiency level in the five highest-rated jurisdictions by percentage of top performers, in England and the OECD average

The percentage of top performers in the PISA maths assessment is invariably higher than the equivalent percentages in the other two assessments. Chart 1 reveals that:

  • 34.8% of Singapore participants are top performers (levels 5 and 6 combined), so just over one-third of the population. Only three other jurisdictions have recorded over 25%. England lags some fifteen percentage points further behind, only just exceeding the OECD average. 
  • The leading jurisdictions also have a noticeably ‘shorter tail’ than England, with 7.5% of learners below the level 2 baseline in Singapore compared with 22.1% in England. Notice that Macao outperforms Singapore on this measure. 
  • But the span of attainment – the gap between the highest and lowest performers – remains equally large – no jurisdiction has completely eliminated low performance (and, as noted above, it is far more likely in Singapore than in England that some of the Level 6 performers would be capable of reaching level 7 if it existed).
  • And, even in Singapore and Macao, there is still a relative tail of underachievement. In Singapore 19.9% of students are at level 2 and below, as are 21.7% in Macao, broadly equivalent to the 22.1% at level 1 and below in England. Roughly 1 in 5 students is part of this ‘long tail’ but the tail begins higher up the distribution.

The proportion of level 5 and 6 performers in Singapore in PISA 2015 is very much lower than the 55.4% returned by Shanghai in the PISA 2012 maths assessment. In PISA 2015, Shanghai’s results are combined with three other Chinese provinces.

There are reasons for questioning Shanghai’s 2012 outcomes which are beyond the scope of this post. But other leading jurisdictions have also registered a decline in top performance in PISA 2015 compared with PISA 2012.

For example, Singapore returned a 40% total in 2012 (19.0% at level 6 and 21% at level 5). In this and most other leading jurisdictions, level 5 performance seems to have changed relatively little, compared with significant declines at level 6.

It has been suggested that the switch to computerised testing might be to blame but, regardless of causation, this suggests that Shanghai’s one off achievement of a majority of high performers in maths in PISA 2012 is not about to be repeated any time soon, so reinforcing my argument above.




Chart 2: PISA 2015 science: Distribution of performance by proficiency level in the five highest-rated jurisdictions by percentage of top performers, in England and the OECD average


Chart 2 reveals that:

  • 24.2% of Singapore students – so just under one-quarter of the population – are top performers (level 5 and 6 combined) in science. This is far fewer than are top performers in maths, but still almost ten percentage points ahead of its nearest competitors in science. England is much closer to the leading jurisdictions, only one percentage point behind New Zealand, the fifth strongest performer, and four percentage points higher than the OECD average.
  • The correlation between a high proportion of top performers and a ‘short tail’ is similar to that in maths. Science (and indeed reading) adopt a slightly different approach to proficiency levels than in maths, splitting level 1 performance into two sub-levels. Singapore has 9.7% of its population performing below the level 2 benchmark in science, somewhat higher than in maths. Japan does very slightly better, while Finland and Taiwan are both a couple of percentage points behind. England has a slightly smaller tail than New Zealand and is well below the OECD average of 21.2%.
  • As in maths, the span of achievement is unchanged and one can see the same pattern whereby the tail begins further up the attainment distribution, though it is not quite so pronounced.

Singapore’s top-end performance in the 2015 assessment is slightly better than in 2012, when 22.7% of the population achieved levels 5 and 6 but, whereas level 5 performance has increased by something approaching two percentage points, level 6 performance is slightly down.

However, Singapore remains a long way from having a majority of its population reach these high performance benchmarks. Even Shanghai recorded only 27.2% at levels 5 and 6 combined in the 2012 science assessment. Singapore has lost three percentage points in 2015 compared with that 2012 high water mark.




Chart 3: PISA 2015 reading: Distribution of performance by proficiency level in the five highest-rated jurisdictions by percentage of top performers, in England and the OECD average


Chart 3, above, indicates that:

  • 18.3% of Singaporean students achieved the high performance benchmarks in reading in PISA 2015, significantly fewer than did so in both maths and science. Its four leading competitors come in at between 12% and 14%, while England is at 9.9%, a little ahead of the OECD average. Other than in Singapore, these percentages are broadly on a par with the achievement of the leading jurisdictions in science.
  • The ‘long head, short tail’ correlation persists, with Singapore at 11.1% below level 2. Both Canada and Finland perform even better than Singapore on this measure, with South Korea a couple of percentage points behind. England, at 17.8%, almost matches New Zealand at 17.3%, both a little way ahead of the OECD average.
  • The span of achievement is unchanged and the tail in the leading jurisdictions starts further up the attainment distribution.

Back in 2012, Singapore reached 21.2% at levels 5 and 6, with 5.0% at level 6 and 16.2% at level 5, so both are down in 2015, the level 6 proportion quite significantly so.

In PISA 2012, Shanghai was ahead of Singapore, recording 25.1% at level 5 and 6 combined, so a smaller percentage than in both maths and science. So, even then, only one quarter of its population exceeded the high performance benchmarks and, for whatever reason, the best performing jurisdiction in 2015 is coming in some seven percentage points lower.


All-round high performers

So, whereas Shanghai briefly reached the position in 2012 where a small majority of its students were achieving the high performance benchmarks in maths, the leading jurisdictions are undershooting this considerably on all three assessments in PISA 2015.

Even in maths, Singapore – the best-performing jurisdiction – is below 35%, while in science it is below 25% and in reading below 20%. This hardly supports the argument that the world’s leading jurisdictions have reached a position where most of their learners are high achievers.

Moreover, the trend is such that they are unlikely to reach this position in the foreseeable future.

And these are single, subject-specific assessments. The proportion of students achieving level 5 and above in all three assessments is significantly lower.

In PISA 2015 the strongest performer was, not surprisingly, Singapore, at 13.7%. They were far ahead the next strongest jurisdictions, which were at 7.6% (B-S-J-G China), 6.5% (Japan), 6.4% (Canada), 6.1% (Netherlands and Estonia), 6.0% (Finland), 5.6% (Taiwan) and 4.8% (Hong Kong).

The 2015 National report gives a figure of 4.8% for England only, matching Hong Kong’s performance. The OECD average was slightly lower at 3.7%.

Charts 4 and 5 below compare the proportions of students from Singapore and England achieving level 5 and above on two or more PISA 2015 assessments.



These show that larger proportions of English students were successful only in science or only in reading, whereas a much larger proportion of Singaporean students were successful only in maths.

The science/maths combination was particularly strong in Singapore relative to England. Rather more English students achieved the combination of science and reading. Conversely, Singapore was ahead in the combination of maths and reading.

In 2012, Shanghai reached 19.6% achieving high performance on all three assessments – and Singapore has fallen back significantly compared with the 16.4% it registered in 2012.

As things stand then, the world’s best jurisdiction can boast rather fewer than one in seven of its population securing high performance in maths, science and reading combined.  

It is certainly true that the most successful jurisdictions at proficiency levels 5 and above have comparatively large proportions of ‘resilient’ students – relatively high performers from disadvantaged backgrounds – although this group is defined according to a lower performance threshold, so the read-across is approximate.

In PISA 2015, 29.2% of participants were deemed ‘resilient’ across the OECD as a whole in science, but the proportion reached as high as 75.5% in Vietnam.

The next strongest jurisdictions on this measure were Hong Kong (61.8%), Singapore (48.8%) and South Korea (40.4%). Since only 24.2% of all Singapore students achieved levels 5 or 6 in science, the proportion that did so despite a disadvantaged background will be significantly lower.



Having summarised my argument at the outset – and having now summoned extensive evidence to support that argument – there is little point in recapitulating at length at the close of this already extensive post.

There is some truth to the ‘all can attain highly’ position, but I have identified three prominent fault lines, associated with:

  • The extent to which schooling can correct socio-economic disadvantage and the proportion of schools capable of ‘bucking the trend’.
  • The ‘labelling argument’, especially when implemented over-zealously and
  • Failure to quantify the appropriate high attainment threshold that all are capable of attaining and, on the frequent occasions when ‘all’ is downgraded to ‘most’, failure to quantify the proportion of the population that this implies.

Having set the first of these to one side, I have exemplified through parallel case studies the difficulties presented by the second and third respectively.

Finally, I have demonstrated – with close reference to evidence drawn from the PISA study – that no country is anywhere close to reaching a position where the majority of its students are deemed high attainers.

There is undoubtedly much we can learn from the jurisdictions that top such international comparisons studies, but we should not over-estimate their success, or our distance behind them.

We must of course maintain consistently high expectations of every learner, every school and of our own collective national performance. We must all strive to support all learners to be the best that they can be, to attain the highest that they can attain, recognising the trade-off between that and other educational objectives, not least their personal well-being.

It is pleasantly anti-elitist to believe that everyone can be a high attainer, but such a belief flies in the face of reality.

It is as well to temper our expectations with a healthy dose of realism, otherwise we are riding for a fall.



April 2017









5 thoughts on “Know Your Limits!

  1. Interesting post. I have not been able to find any studies on the link between ‘ability’ and effort, do you know of any?

    I absolutely hate to use anecdotal evidence, but my daughter is a grammar school sixth former whose main gripe against the selective system is that it takes no account of actual work at all. She tells me her school is full of 11+ passers, who must be high ability pupils, but she tells me often underachieve due to not applying themselves to work. I admit that her view may be coloured by being an 11+ failer who puts many hours into her studies! But she is now top of her classes and is convinced it’s ‘work’ not ‘ability.’ Though I’d say it’s both.

    Does your study of ability over the years take account of this factor? I’m no fan of segregated schools due to my daughter’s experience, but if we have tests for ability such as the 11+ should they consider more than smarts in a one-off test? Or should that be enough and it’s the school’s job to make a lazy smart child achieve their potential? I realise this is a tricky area for debate…!

    So, is a medium-high ability child with a great work ethic as important for tailored support as a high ability child with no motivation?

    I’m putting this to you because I feel our definitions are often too narrow and based on CAT/IQ with no regard for much else.

    Of course we do get into tricky areas when we bring personality into the equation, but if it personality is a significant factor in achievement we should surely consider it.


    1. Thanks Jo. I have been looking for research studies on this issue in relation to 11+ selection in England but, like you, I haven’t discovered anything…at least not yet.

      Anecdotally I guess we’re in opposite camps. Our son passed a highly selective 11+ with ease. He’s blessed with ability – his CATs scores were pretty much at the ceiling – but not quite so blessed with drive and commitment. We baulked at paying for private tuition, despite several of the cognoscenti telling us that was essential, and I kept my distance on principle (so no-one could connect his success or failure with my role as national G&T lead) but my wife supervised weekly practice sessions in the months leading up to the test. Entering the 11+ was his choice – we didn’t press him either way.

      At school he always did just enough, preferred to be in the middle of the pack, but refused to bust a gut – was simply not interested in putting in the extra work for Oxbridge. He’s a well-adjusted individual and work-life balance is important to him. He’s now doing a reasonably demanding degree at a university with a good reputation and he’s very happy. I think he made absolutely the right choices for him.

      Moving to the more general argument, I often wonder whether grammar schools rather like the private tuition industry, because it guarantees them a steady flow of learners committed to hard work with wealthy, supportive parents who push them to succeed. It’s almost as if, for all but the super-bright, there’s an unofficial effort-based selection criterion already in place.

      Problem is, learners from poorer backgrounds can expend all the effort they have and still come up short – because there is no equality of opportunity. Another potential downside is that heavily-tutored middle attainers may start to fall behind once they pass the 11+, or can only keep up through a steady diet of additional tuition which they have to squeeze in alongside the heavy homework load – and are constantly pressed by their parents to succeed. According to our son there were several of that ilk in his year group too. Some rebelled or went off the rails.

      I have a hunch that the most successful students – by which I mean the highest-attaining – are those with a ‘balanced scorecard’ – strongish ability and strongish drive – while those who lean too much in one direction or the other are rather more at risk of coming a cropper. Wouldn’t it be great if there were some decent research studies testing out this hypothesis?



    2. I think you can derive a lot of truth from anecdotes and personal observations. The Archimedes ‘Eureka’ story could be fiction, but I think we can safely assume personal observations led to that truth, as opposed to a five year RCT and some complex statistical analysis.

      We shouldn’t need support for something so reasonable, but OTOH we did get that silly ‘10,000 hours rule’ and some still believe it. So, Plomin’s behavioural genetics gang are a very good source for a few things around education and this is their very relevant paper:

      “The high heritability of educational achievement reflects many genetically influenced traits, not just intelligence”

      I don’t see how it could be sufficiently reliable, but given selection I would like personality traits (e.g. self-motivation, focus, determination) to be part of the selection.

      Circling anecdotes and effort, my Middling Comp attending 13yo daughter has both the very high CAT score and the ‘right stuff’ in terms of personality traits. One thing that has regularly followed us throughout her school-life is the incorrect assumption that the high-attainment must because she spends her home-life working her socks off. In many cases I think this is because they, teachers and parents, would have had to work their socks off to achieve as highly. For instance although I can’t fault the form teacher for the first round, but we’ve now had two earnest parents’ evening chats around high-attaining girls, perfectionism and mental health. I’m expecting round three in the summer term because ‘Doesn’t apply here’ just doesn’t get through.

      Ability clearly explains some of why my daughter doesn’t need to invest so much time, but I included ‘focus’ in my example personality traits because she can turn that on, burn quite brightly so to speak, and that concentration reduces the time she needs to spend on things. This is all subject dependent of course e.g. the maths syllabus contains relatively few concepts up to GCSE and is much more vulnerable to ability, whereas Art just takes time regardless.


  2. Hi Tim,

    I have been folowing your posts as they are particularly relevant to my family circumtances, is this the best way to email you.

    Best Wishes

    Julie Facer



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s