Investigating grammar streams

This post investigates the practice of introducing selective grammar streams into comprehensive schools.


  • Reviews recent advocacy for this practice.
  • Distinguishes grammar streams from other, related approaches to within-school selection.
  • Urges revision of the official distinction between ability and aptitude, based on the erroneous position taken by the School Adjudicator.
  • Places grammar streams in the wider context of the evidence base for and against streaming.
  • Adopts as a case study two grammar streams recently introduced by the United Learning Trust (ULT).
  • Concludes with a summary of key points and assessment of the potential value, or otherwise, of more grammar streaming as one dimension of government policy in the wake of the selection green paper.


Background – previous posts and the selection green paper

Publication of the green paper ‘Schools that work for everyone’ (September 2016) has generated much debate about the advantages and (mostly) disadvantages of increasing between-school selection.

I have made my own small contribution, including an initial analysis of the green paper proposals – Be careful what you wish for (September 2016) – and an alternative proposal – Rounding-up: Killer stats and 10-point plan (February 2017) – which sets out a coherent support strategy for high attainers, especially disadvantaged high attainers, that does not depend on increasing the number of selective school places.

The green paper proposes three between-school selection strategies:

  • Supporting the expansion of the existing 163 secondary selective schools.
  • Permitting the introduction of new selective free schools. These may be either wholly or partially selective; the former selecting their entire intake ‘on the basis of ability’; the latter selecting ‘a proportion of their places by ability or aptitude and a proportion without reference to aptitude or ability’.
  • Allowing existing non-selective schools to become selective. The text implies that full selection is the only option, since partial selection is not mentioned:

‘These schools would become selective in response to local demand and would have flexibility to select 100% of their intake on the basis of ability.’

There is additionally a series of proposals to strengthen the contribution made by existing selective schools to support learners educated elsewhere.

These include a proposal to encourage multi-academy trusts (MATs) to establish a single centre, real or virtual, ‘in which to educate their most able pupils’.

This ‘centres of excellence’ model is essentially a hybrid, combining between-school selection (because more than one school is involved) and within-school selection (because it operates only after learners ‘have been admitted to their individual school through a non-selective admissions process’).

I have written about the conditions that such centres would need to satisfy – and some of the issues they raise – in ‘Making sense of centres of excellence’ (September 2016).

Otherwise the green paper does not explicitly mention within-school selection as a complementary strategy, presumably because the emphasis is predominantly upon increasing the number of selective school places, as one means of increasing the number of ‘good and outstanding’ school places overall.

Nor is there any allusion to parallel work we know is being undertaken to assess:

‘…the best ways to support the most academically able pupils across the full range of state schools, particularly those from the most disadvantaged backgrounds’.

Consequently the green paper fails to consider the needs of (disadvantaged) high attainers in the round.

Readers wishing to understand the development of policy on between-school and within-school selection over the last 20 years might be interested in two older posts of mine:

  • The politics of setting (November 2014, reblogged December 2016) which also discusses the distinction between setting and streaming and the confusion engendered by failing to appreciate that distinction.


Recent advocacy for grammar streams

‘The politics of setting’ explains how, prior to the publication of ‘Raising the bar, closing the gap’ (November 2007), Tory opposition leader David Cameron had initially campaigned for ‘a grammar school stream in every school’.

Then in September 2014 there were briefly rumours that new Secretary of State Morgan was about to propose compulsory setting in schools, and that this might appear in the 2015 Tory Election Manifesto. This was quickly scotched, however.

Recent Tory Party pressure to reintroduce selection has been orchestrated by Conservative Voice, which launched its initial campaign in late 2014. Some pushed for the creation of ‘specialist academic schools’ though scant attention was paid to partial selection or grammar streams.

But grammar streams have received far more attention since the green paper was published and have some powerful advocates.


Toby Young/NSN

In August 2016, Toby Young published a paean to partial selection in the Spectator, shortly before he was appointed director of the New Schools Network.

Young declares himself ‘ambivalent’ about grammar school expansion, while noting the potentially deleterious impact a new grammar school might have on neighbouring institutions.

He argues for partial selection as a ‘third way’. Academies and free schools should be permitted to select a maximum of 25% of their intake ‘on the basis of general ability’. They would also have to ‘admit more than the local percentage of children on free school meals’.

This would spread the benefits of selection while reducing the impact on other schools.

These partially selective schools would be able to operate a grammar stream; they might also teach entirely in mixed ability settings. Where streams are in place, learners might more easily move between them than between schools.

Young argues that the attainment and progress of disadvantaged learners attending partially selective schools is strong compared with national averages, but makes no comparison with success rates at fully selective schools.

Towards the end of the article he suggest that schools which select a proportion of their intake in this way were formerly known as bilateral schools.

In a press release marking its subsequent response to the green paper the NSN declares that it takes no position on increasing selective places.

It does, however, call for increased selection by aptitude in specialisms:

‘At present, schools with certain specialisms are able to set aside 10% of their places for applicants with a particular aptitude for their specialisms. If this 10% cap was lifted, and the range of specialisms schools could select for was increased, we can envisage an assortment of new schools being set up, including replicas of the highly successful BRIT School, schools specialising in particular sports and schools with a focus on a particular performing art, such as dance, music or drama.’

The press release does not mention whether ‘general (academic) ability’ should be one of these specialisms, but one assumes not.


Centre for Social Justice/Tim Leunig

In December 2016 the Centre for Social Justice published a report ‘Selective education and social mobility’ which recommends partial selection, as well as grammar streaming within individual schools and across MATs:

‘Through the course of our interviews one of the more popular ideas was a school with a partially selective intake and a ‘grammar stream’…

…While it contains problems of its own, one interviewee, otherwise completely opposed to grammar schools, felt that this was the only palatable way of introducing selection: ‘A two-tier system is inevitable without streaming within schools.’

Schools could be given the ability to test as widely as they like, taking 30% of their intake based on test results with the other 70% taken from the local catchment area. This system allows the student population to remain united by the ethos of the school and able to mix socially in extra-curricular activities and mixed ability form groups, which meet at the start and end of the day.

The structure also allows for greater movement at different ages in and out of a ‘grammar stream’, so softening the divisive importance of an 11 plus exam and allowing for late developers and those that later struggle with a more academic education.

Some of our interviewees felt that this would differ little from ‘setting’ that already happens at many schools, while others felt that this similarity made the idea more palatable to those that oppose grammar schools.

A streaming model would prevent social division, schools ‘creaming off’ the best students and teachers, while giving the most academically able – from whatever background – the best chance of a high quality education…

…Another model for school streaming would be a completely non-selective intake at year 7 with streaming decided at the end of the first year by teacher recommendation. This could remove the importance of a one off academic test, and allow teachers to select based on a thorough knowledge of pupils’ academic potential.’

The paper argues that the introduction of grammar streams might herald the introduction of more widespread tracking ‘with different streams of a school sharing a core academic curriculum, but offering different degrees of vocational or technical education alongside, depending on the aptitude and interests of the child.’

It recommends that DfE should commission analysis of tracking in other countries, including Germany and the Netherlands, whose systems are described briefly in an annex to the report.

The Dutch approach had already featured the previous month in a one-off evidence session on selective education conducted by the Education Select Committee.

When asked about international evidence, Tim Leunig, Chief Scientific Adviser and Chief Analyst at DfE was pointed to extensive selection in Hong Kong and Singapore but opted instead to discuss the Netherlands:

‘They select at the age of 12 through something that would look recognisably like selection here. 20% of people go into the most academic stream and those people in that school are then explicitly destined for conventional research universities. The next 20% of people go into the next group of schools…They are essentially destined for what we would have once called polytechnics, applied vocational universities. The remaining 60% of children are in a less academic stream that is explicitly technical. It is not the sort of Rab Butler tripartite. It is a tripartite of academic, semi-academic and technical. They are then stratified within each stream and in total they have nine different strata, so it is a very selective system.’

The Select Committee’s report (February 2017) argues:

‘The Government’s proposals must take account of the needs of the economy for a broadly skilled workforce, recognising that generally technical specialisation occurs later in a student’s education, and take into account the UK’s competitiveness in a globalised economy. This will involve having regard to international trends and the performance of other countries’ education systems, which do not always point towards earlier specialisation within school systems, and attention should be paid to the Dutch model, which is overall a successful system and one that includes selection. If England is to take this course, it would be important for the Government to demonstrate clearly how this policy will meet the requirements of the Industrial Strategy’.

But this debate about the definition and operation of technical tracks in our education system is a red herring, distracting from the narrower question of the desirability, or otherwise, of operating grammar streams, or full and partial selection.

The fundamental problem is not resolved by increasing the number of selective pathways, or adopting a system-wide tracking reform which would, incidentally, cut right across the government’s market-driven philosophy and espousal of greater school autonomy.


Mary Curnock-Cook

In January 2017, Mary Curnock-Cook, the outgoing Chief Executive of UCAS wrote a paywalled article for the Daily Telegraph arguing the case for grammar streams.

An accompanying commentary says that:

‘…while she is not in favour of grammar schools, she believes that offering selective streams within state schools is the “best of both worlds”’.

There is no further insight into her reasons for believing this.

According to a parallel article in the Mail, her argument is that:

  • Grammar streams are preferable to teaching higher-attaining pupils in mixed ability classes (the Mail says they ‘languish’ in such settings, but it is not clear whether that is Curnock-Cook’s language).
  • ‘There’s no cliff-edge exam at 11-Plus, and pupils are able to move in and out of it [the grammar stream] where progress dictates.’
  • ‘There’s no segregation on the bus routes, no segregation by school blazer, just streaming with an aspirational flavour.’
  • The grammar stream might be supported through independent-state partnership.

Curnock-Cook’s article coincided with a visit to Swindon Academy by Schools Minister Nick Gibb.



She sits on the Academy’s governing body and also chairs the secondary curriculum committee which oversees its grammar stream. This bears all the hallmarks of a concerted publicity campaign to bolster a flagging initiative.


Distinguishing grammar streams from partial selection


Partial selection

All three commentaries above are guilty to some extent of sowing confusion over the distinction between grammar streams and other approaches to within-school selection.

The typology of partial selection within our system is complex and poorly understood.

There are:

  • Bilateral schools

A few surviving examples operate in selective areas. According to the elevenplusexams website there are either seven or eight of them (they include Erith School in Bexley in one list but not another. The School’s own website suggests that is was formerly bilateral but no longer operates as such).

The distinguishing feature of bilateral schools is that they have parallel selective and comprehensive intakes, with selection operating prior to admission. If they do not fill their selective places the vacancies are reallocated to the comprehensive intake.

A pure form of bilateralism would maintain the distinction between these two cohorts by operating a completely separate grammar stream, but I can find no evidence of the pure form existing in practice.

Some appear keen to downplay their bilateral status while others are more upfront.  Most seem to rely on varying combinations of setting and mixed ability teaching, so the distinction between the two streams is limited, even negligible.

One of these schools has been placed in special measures (December 2015), Ofsted noting insufficient challenge for the most able.


  • Schools maintaining historical partial selection arrangements

A somewhat larger group of schools have permission to maintain partial selection arrangements, whether by ability and/or aptitude (for distinction see below), that were in place at the beginning of the 1997/98 school year.

Partially selective schools cannot exceed the lowest selection rate they have adopted since 1997/98. They too must reallocate any unfilled selective places.

The list published on the elevenplusexams site includes 36 schools in 17 different local authorities, but this may not be completely up to date. One of the schools on the list is also bilateral.

There are seven schools in Hertfordshire; four each in Croydon and Wandsworth. Twenty-eight of the schools select by ‘general ability’, seven by one or more aptitudes and ten by a combination of aptitude and ‘general ability’.

One of the schools is said to rely on ‘weighted banding’. (According to the 2017 admission arrangements Archbishop Tenison’s in Lambeth places applicants in three ability bands and normally admits 30% above average applicants, 50% average and 30% below average, so 110% in total!)

The percentage selected by ‘general ability’ varies between 9% and 35% of the intake. Eleven schools select 25% or more by ‘general ability’.

Where schools select by both ability and aptitude, the total percentage of the intake never exceeds 45% (the proportion reached at Queen’s School, Bushey).

The aptitudes represented include technology, music, dance, sport, maths and ‘nautical’ (the latter at London Nautical School, though the 2017 admissions policy makes no reference and depends instead on ability-based banding).The percentage selected by any single aptitude ranges between 1% and 17%.

Of those schools with the largest incidence of selection by ‘general ability’, only one seems to have anything similar to a grammar stream – Graveney School says it groups learners into three broad ability bands in KS3 and teaches the more able in extension groups.

Others appear to rely on setting to a limited extent. Some of those with recent Ofsted reports are commended for their work with the most able.


  • Specialist schools selecting on the basis of aptitude

All schools with an appropriate specialism may select up to 10% of their intake on the basis of aptitude, in PE or one or more sports, one or more performing arts, one or more visual arts, one or more modern foreign languages, design and technology or information technology.

However, the technology options only apply if the school selected in this way in 2007/08 and in every subsequent school year.

The 10% restriction is an overall limit, applying even if the school selects for two or more different aptitudes.

According to Comprehensive Future, no data is collected on how many schools are selecting on the basis of aptitude.

All schools may also adopt banding – officially an approved form of selection – to generate an intake that is representative of the full range of ‘ability’ amongst applicants, or in the local area, or nationally.

Schools cannot prioritise applicants within bands according to their performance on the selection test. Schools may combine banding and up to 10% selection by aptitude. It is open to schools to use banding to inform subsequent within-school selection, but this seems comparatively rare.


Grammar streams

Partial selection and grammar streaming are distinct concepts because:

  • As is evident from the preceding section, few schools that operate bilateralism or partial selection on the basis of ‘general ability’ maintain a rigid distinction between their selective cohort and their comprehensive cohort, preferring some combination of mixed ability classes and setting that permits greater movement between the two. This is logical since the non-selective cohort is comprehensive rather than ‘modern’: it includes learners who might have taken and passed the selection test but, for whatever reason, have not done so. The comprehensive cohort will also contain some ‘late developers’ who reach the necessary standard at a later stage of their education.
  • Selective streaming can operate independently of the admissions process, provided that the selection takes place post-admission. Selective streaming can operate in grammar schools. (Given wider issues with streaming, such practice is likely to be far less prevalent today, but I can personally attest to its existence in the 1970s.) Selective streaming may also take place in comprehensive schools or in modern schools. It may also operate across a group of two or more schools, as in the case of the MAT ‘centres of excellence’ model floated in the green paper.

Schools that deploy setting as a form of within-school selection are not operating a grammar stream, unless they set and stream in combination. Universal setting, in every subject, is not synonymous with streaming.

For setting involves selecting the learners in a discrete class, normally on the basis of recent attainment in the specific subject that will be taught in that class.

Whereas streaming typically involves selection – more often generic and ability-based, though sometimes also taking into account prior attainment in maths and/or English – that applies across several or even all subjects.

It follows that those selected into a higher stream may not be higher attainers in every one of those subjects.

Universal setting would almost certainly generate more variability between the top sets in different subjects, because some high attainers are strong ‘all-rounders’ whereas others have ‘spiky profiles’ with particular strengths in some areas relative to others.

The EEF Toolkit makes a slightly different distinction:

‘Pupils with similar levels of current attainment are grouped together either for specific lessons on a regular basis (setting or regrouping), or as a whole class (streaming or tracking).’

This does not allow for the possibility that streaming may be undertaken primarily on the basis of generic ability.

The advocates of grammar streams are really advocating partial selection, since all schools are completely free to adopt post-admission streaming, whether or not they are partially selective.


The distinctions between attainment, ability and aptitude


Attainment and ability

There is widespread confusion between attainment and ability – and a tendency to use the terms synonymously, even though they are clearly distinct concepts.

  • Attainment is what the learner can achieve now in a particular subject or group of subjects, for example a GCSE grade 9 or an Attainment 8 score of 75+. 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.
  • Ability is a judgement or prediction of what the learner may be capable of achieving in the future, derived from evidence about their existing capabilities. This may include information about their prior attainment but will typically foreground the assessment of generic capabilities, for example verbal and non-verbal reasoning and/or visual-spatial skills. CAT tests fall into this category, as do the tests most grammar schools use to select their intakes.

High attainers may be regarded as a subset of the highly able, who successfully combine ability and effort to achieve strongly. But not all highly able learners are high attainers – there is extensive underachievement, some of it partly attributable to socio-economic disadvantage.

The nature of ability has become a battlefield in recent years, with some denying its existence and others arguing over its relative significance and causation. Measurement has always been problematic – and remains so.

This debate plays into wider concerns about elitism, equity and social mobility which detract from efforts to meet the learning needs of high attainers and potential high attainers alike.



The green paper includes an unfortunate footnote:

‘The Office of the Schools Adjudicator defines ‘aptitude’ as the potential to attain and ‘ability’ as actual attainment.’

The most recent iteration of the School Admissions Code (December 2014) distinguishes selection by aptitude from selection ‘on the basis of high academic ability’, but neither term is included in its glossary.

That said, aptitude is clearly associated with designated specialist subjects. Admissions authorities must:

‘ensure that tests for aptitude in a particular subject are designed to test only for aptitude in the subject concerned, and not for ability’.

The OSA’s false distinction is further glossed in the free schools admissions guidance.

This December 2014 edition (version 3) says in a footnote (on page 7):

‘Aptitude measures potential. Ability measures attainment or what has already been achieved. The two should not be confused.’

This document about common admissions issues, also published in December 2014 (version 4) adds (on page 12):

‘In testing for aptitude, you must look to determine a pupil’s natural talent for a particular subject, and their potential to develop it further. You must not test a pupil’s prior learning or ability: whether a pupil has agility and balance tests aptitude for sport; whether a pupil has represented the county at football tests ability.’

So, according to this school of thought:

  • Aptitude is ‘natural talent’ or ‘potential’ in a particular subject – and cannot be identified through attainment or prior learning.
  • Ability, otherwise known as ‘high academic ability’ when it refers to selective admissions, is a measure of ‘attainment’, which is a synonym for prior achievement

This is clearly nonsensical, not least since few grammar schools select on the basis of prior attainment.

How did the Adjudicator get himself in such a tangle?

It can be traced back to 2003 when the incumbent opined on the distinction, in this note for the Education and Skills Select Committee:

‘If a neighbouring school, or local education authority, objects to a school’s admission arrangements an adjudicator is called in. Usually the objection is that, while purporting to select by aptitude, the school is actually selecting by ability. So it is crucial we know what the two words mean and the practical difference between them.

Finding a difference between the meaning of two such words is the sort of exercise lexicographers get up to when they haven’t enough to do. Most dictionaries tend to use the two words alongside each other in the definition of both. In an attempt to draw a clear distinction, legislators have also got in on the act. The School admissions code of practice gives the following definition: “a pupil with aptitude is one identified as able to benefit from teaching in a specific subject, or who demonstrates a particular capacity to succeed in that subject” —not the most helpful guide to anyone who wants to know what it is.

We can find a way through this by using the word ‘ability’ in the same way as we use ‘achievement’. Ability is assessed using the normal tests used in schools and elsewhere—GCSEs, musical instrument grades, swimming proficiency certificates and so on. These are tests to find out what people can do.

The word ‘aptitude’ then means a gift or a talent. It denotes a potential or propensity to develop an ability given appropriate teaching or preparation. In other words aptitude + preparation = future ability.’

In arriving at this distinction, Philip Hunter (for it was he) credits messrs Wiliam, Oates and Whetton for their advice, which informed his note.

The problem lies with the paragraph italicised and emboldened above, which wrongly argues that ability and achievement (which includes GCSE attainment) are synonymous. This is completely wrong.

If there is a clear distinction between ability and aptitude – and that is highly debateable – it can only rest on the subject-specific (or field-specific) nature of aptitude versus the generic nature of ability.

Neither is necessarily dependent on prior attainment (or performance in the case of arts and sports), nor do they require a certain level of prior attainment/performance to have been demonstrated.

In testing for aptitude, admissions authorities should avoid tests that require specific levels of attainment or performance; equally they should avoid tests of generic ability that are not directly related to the subject or field in question.

This confusion arose at a time when the government of the day was desperate to avoid allegations that they were reintroducing ability-based selection ‘by the back door’. That is no longer the case so this fault should be rectified when the Admissions Code is revised to reflect other green paper reforms.


The evidence on streaming


Setting and streaming are treated as interchangeable

In ‘The politics of setting’ (November 2014) I took issue with the way in which the EEF Toolkit combines setting and streaming in a single entry and even refers to a wider set of interventions for gifted and talented learners.

Since setting and streaming are completely different practices it ought to be possible to recognise them as such, rather than treating them as identical. Unfortunately much of the research evidence appears to fall into this trap, probably because the advantages and disadvantages are similar in nature, if not in scale.

The perceived wisdom is that the negative effects of streaming are likely to be more pronounced – because streaming is potentially more inflexible and more inequitable.

Moving streams is ‘a bigger deal’ and so harder for a school to manage, while the higher stream will include some comparatively lower-attaining learners in various subjects – there will be those in lower streams who might be in top sets for certain subjects.

It would be helpful to know how this difference between setting and streaming is reflected in the EEF effect sizes (months of progress). Otherwise the relatively poorer reputation of streaming will continue to distort our view of setting.

There have been periods when the government of the day has advocated one over the other. For example, the Brown Government said in The Children’s Plan (December 2007):

‘Using setting and groups to teach children of similar abilities and interests can bring real educational benefits. But where it is poorly implemented, for example through ‘streaming’ (where pupils are grouped across a range of subjects based on general rather than subject-specific assessment) it can be socially divisive and detrimental to all but the highest achieving pupils’ (p. 68).


Other problems with the evidence base

The evidence on both setting and streaming is often misrepresented.

Much of the downside to setting (and potentially to streaming as well) is associated with poor practice, as opposed to exemplary practice. This downside includes:

  • Learners in lower sets experiencing a comparatively impoverished curriculum.
  • Learners in higher sets experiencing better quality teaching from higher qualified and more experienced teaching staff.
  • Learners moving between sets too infrequently and downward movement being associated with loss of esteem which impacts on attainment.
  • Learners in lower sets having more negative attitudes towards learning and teachers and parents tending to have lower expectations of them.

The EEF-funded study of setting should be exploring to what extent these problems can be ameliorated through the implementation of effective practice. Unfortunately there is no parallel study of streaming.

Additionally, it is commonly understood that there are distinct benefits for high attainers, which may include those from disadvantaged backgrounds.

Lower attainers (and arguably middle attainers) tend to suffer negative effects and, while those categories include most disadvantaged learners, they do not include all of them. The benefits for disadvantaged high attainers are too easily discounted because they are in the minority.


Incidence of streaming

The distinction between setting and streaming may be underplayed in more recent research because it is perceived to be comparatively rare.

This used to be true. In 2000 a study by Ireson of ‘Innovative grouping practices in secondary schools’ reported that:

‘The majority of schools use some form of ability grouping in at least some subjects, although only a very small proportion use streaming.’ (p.1)

A 2013 paper by Hallam and Parsons mentions that, in the primary sector:

‘By the 1990s the incidence of streaming had declined to less than 3 per cent (Lee and Croll, 1995).’

But they go on to reveal a more recent resurgence, using Millennium Cohort Study data to estimate that some 16% of Year 2 learners born in 2000/2001 were in streamed classes in 2008.

The definition of streaming they use is ‘schools group children in the same year by general ability and they are taught in these groups for most or all lessons’.

They also found a comparatively strong relationship between the incidence of streaming and setting:

‘…64.3% of children who were streamed were also set for literacy, 69.5% were also set for numeracy. Among the majority of children not streamed, 23.4% were set for literacy, 29.3% were set for numeracy.’

They comment:

‘This research raises issues as to why those schools adopting streaming were doing so given that this and other research has shown that streaming, of itself, does not raise attainment (e.g. Barker Lunn, 1970; Ferri, 1971) and that many were also adopting setting which has been demonstrated to be more effective in this respect (Slavin, 1987).’

In the secondary sector one would be forgiven for believing that streaming had all but disappeared, but interest in grammar streams appears to be growing rapidly.

We have no national data on the incidence of grammar streams and there is very little contemporary research on within-school selection in England, let alone the adoption of grammar streams.

A handful of recent studies languish behind paywalls, but the published abstracts suggest that paying for their release would not be good value for money.

The best study I know of is still this from 1998 by Sukhnandan and Lee. This 2016 paper by Francis et al. helpfully summarises the findings of more recent studies (as well as citing my own work), but neither offers detailed coverage of streaming, or comparison between the effects of setting and streaming.

Given the poor reputation of streaming and the negligible research evidence, why are schools developing this practice and what reassurance is there that the practice they are adopting is avoiding the negative consequences outlined above?

The obvious answer to the first question is that most grammar streams operate in comprehensive schools located in selective areas – and they are using them to compete with grammar schools for a proportion of high attainers. But that is not universally true.

The answer to the second question is unclear – and I want to come at it through a case study of two different grammar streams operated in United Learning Trust (ULT) academies.

The first is Swindon Academy, where Curnock-Cook is a governor. The second operates in Glenmoor and Winton Academies in Bournemouth.


Swindon Academy launches a grammar stream

The most publicised grammar stream is that located at Swindon Academy, a mixed all-through (ages 3-19) academy with some 1,700 learners on roll. The School opened in September 2007.

There is a strong partnership between the Academy and Marlborough College. The Academy website refers to joint ‘subject residentials’, student exchanges, support with UCAS applications, debates, seminars, trips and enrichment activities.

The College website says that ULT funds a Marlborough teacher to teach two days a week at the Academy. It also mentions one- and two-day residential visits, maths mentoring, Japanese lessons and Latin taster sessions, evening lectures and week-long teacher exchanges.

Marlborough also supports the Academy’s grammar stream. The Academy’s website contains a press release, dated 17 September 2015, which marked its launch.

It says that from September 2016 (so the start of the present academic year) there will be 30 places available, providing ‘…a highly academic, Marlborough College backed education…’

The venture is described as a ‘joint working partnership’ between the two schools. It is envisaged that it will:

  • Continue throughout the intake’s secondary schooling (so is hopefully a firm seven year commitment for this first cohort at least).
  • Be open to children throughout Swindon (and beyond) with admission via ‘an aptitude test’.
  • ‘Provide a highly academic and rigorous curriculum’ different to that of other Academy pupils, though there will be common tutor groups and enrichment activities.
  • Suit those ‘who consistently achieve at Level 5 or higher in Primary School and who are hardworking and committed to the additional homework and prep need to achieve outstanding academic results’.

ULT published a similar press release.

There are three obvious reasons why a grammar stream was introduced at Swindon Academy:

  • In March 2015 it received a critical Ofsted inspection report which rated it as Requiring Improvement. Inspectors commented:

‘The achievement of the most able students is too variable because the activities set for them are not always challenging enough.’

‘By the end of Year 6, the most able pupils do not reach the levels they are capable of, especially in mathematics.’

‘The most able students are not routinely provided with activities that challenge or deepen their thinking. Some students told inspectors that sometimes their work is too easy.’

  • The Academy’s reputation has been depressing its intake, so it needs to attract more pupils. Planned admission numbers for 2017/18 are 120 in Year R and 60 in Year 7, giving a total year group size of 180 in Year 7. According to the Local Authority, in October 2016 there were between 91 and 105 pupils in each of Years 8-11, suggesting that each was little more than half full. The 2016 Performance Tables show that 60.4% of pupils are ‘ever 6 FSM’ and so eligible on those grounds for the pupil premium. Performance is depressed, with only 1% of pupils achieving the higher standard at KS2 and high prior attainers recording an average Attainment 8 score of 63.8. Progress scores are typically below average for KS2 and average for KS4.
  • Although Swindon is itself non-selective, according to the Academy’s headteacher some children attend grammar schools in Gloucester, some 36 miles away.


How Swindon’s grammar stream operates

There is fairly extensive information about the operation of the grammar stream on the Academy’s own website. Key points include:

  • There is a steering group comprising senior leaders from both Marlborough and the Academy. This oversees curriculum plans and monitors student progress. There is also a Grammar Stream Co-ordinator.
  • Taster sessions are made available for prospective Year 6 applicants. They must also sit an entrance exam in April, after their parents have accepted a place at the Academy but before they begin Year 7. This circumvents the embargo on further 11+ selection.
  • The entrance exam is set ‘in conjunction with Marlborough’ and comprises three 45-minute tests in verbal reasoning, quantitative reasoning and non-verbal reasoning respectively. These are ‘are designed to be taken without any revision or preparation so they can assess a child’s potential in his or her ability to reason’. It is also argued that the results of KS2 tests often arrive too late for them to be used for selection purposes.
  • Grammar stream pupils:

‘…have all of their lessons together providing an ethos of high challenge and academic rigor [sic]. Among other subjects, pupils will study separate sciences, computer science and also Latin and the Classics.  There will also be additional expectations for preparation and homework.’

In fact the KS3 curriculum includes triple science, Latin and Spanish. The stream has lessons at Marlborough every Monday afternoon. Sport and PSHE are taught with the rest of the year group. There is two hours of homework per evening. There is no obvious explanation of how the curriculum will differ in KS4 or KS5.

  • Pupils also benefit from ‘an annual calendar of enrichment and extension activities’ provided at Marlborough. They can apply for a College Sixth Form Bursary and there is a commitment that two (per year?) will receive a full boarding bursary.
  • The grammar stream does not have its own separate staff – teachers also work elsewhere in the Academy.

More recently the headteacher has been reported locally as saying that the grammar stream:

‘…would allow other children in the school to join at any age if they reach the required standard and places were available’,

But, at the time of writing, I could not find this substantiated on the School’s own website.


Reaction to Swindon’s grammar stream

It is fair to say that reaction has been mixed. The School is said to have leafleted 98,000 homes ahead of the first application round.

This seems to be the leaflet. It includes a quote from Sally Coates, Director of Secondary Academies, South for ULT, which may be the source of Curnock-Cook’s ‘best of both worlds’ claim:

‘A Grammar Stream within a comprehensive inclusive school is getting the best of both worlds. It is ensuring the more able pupils have a challenging rigorous academic curriculum within an environment where all pupils can aspire to this.’

In October 2015 the Labour Group Leader on Swindon Council opposed the plans which had been welcomed previously by the Conservative Cabinet Member for Education and the local MP.

He foresaw other local schools following suit:

‘However let’s assume this “grammar stream” provision is successful and the Swindon Academy successfully takes in the town’s brightest pupils, rather than those pupils attending the secondary schools closest to where they live. Can we really expect Swindon’s other secondary schools who are under pressure to produce good results from their GCSEs not to consider their own “grammar streams” in order to retain the brightest pupils in their catchment area? Could we see in a decade a “grammar stream” type provision in all local secondary schools? That is the danger the people who support this scheme are running…’

In October 2016, ULT’s Chief Executive was reported to have argued, in response to the green paper, that ‘creating more grammar schools will do little to raise standards’:

‘…he says the key question is how to galvanise change in working-class communities that feel left behind.

“So people who feel they’re struggling to get by, who want something for their children, we don’t put them in a position where they’re hoping they’re the one in three that get through the test.

“We have to put them in a position where they’re part of the 100% who get something better.”’

It is hard to understand why this doesn’t apply to Swindon’s grammar stream.

That same month local MP Justin Tomlinson visited the grammar stream and was reportedly ‘overwhelmed with positive feedback’ from pupils and parents.

He is quoted:

‘The common theme from parents was that their child had been bored at primary school because they had found that the work wasn’t challenging enough. Therefore they wanted to ensure that their child was given a rigorous academic education which would allow them to reach their full potential, without having to pay fees for a top independent school, which many would not be able to afford.

I firmly believe that the education system in this country should provide opportunities for EVERY child…and the Grammar Stream is providing that opportunity to some of the more academically minded children here in Swindon. What’s more, the stream is integrated into the Academy so every child in the school is given a chance to join the programme at whatever age, so it isn’t completely dependent on a child being able to pass an 11+.’

In January 2017 Minister for Schools Nick Gibb visited the Academy. According to a local media report, he reacted positively:

‘This is a school that serves an area of high disadvantage but schemes like this will help pupils from disadvantaged background to succeed. There is a real focus on rigour. It will help to provide opportunities for those children who might not otherwise have access to certain subjects and such a challenging curriculum.

But a Schools Week piece published just before the ministerial visit gave the results of a Freedom of Information request it had submitted about the 2016 intake:

‘…of 24 pupils who applied to the grammar stream, 23 were offered places with three on “conditional” offers. The one pupil not given a place was “too close to the pass mark” and given a normal place in the academy instead, a trust spokesperson said.’

So there were insufficient applications to fill all the places available and almost every applicant was granted a place. I could find no reference on the Academy website to how the ‘pass mark’ is set and whether it can be adjusted freely to reflect demand. Nor is there any reference to the scope to make ‘conditional offers’.

It is hard to escape the conclusion that the outcome was manipulated to ensure a just-about-viable cohort (though with fewer than 80% of places filled). Some of the successful candidates are clearly borderline, so the level of demand may have to be adjusted downwards to meet their needs.

The School confirms that 30 places will again be available in September 2017, but appears unconvinced about future prospects. The report says it ‘will continue the project for at least another year and may expand the model’.

Local Authority data shows that the Academy’s September 2016 Year 7 cohort comprises 136 pupils, an increase of 45 on the Year 8 cohort, but still 44 pupils short of capacity.

The grammar stream has accounted for just over half of this increase (though some of those pupils might have attended the Academy regardless). We know that there was only one pupil who applied to the School but failed to pass the selection test.

There is no information about the incidence of ‘ever 6 FSM’ learners amongst the successful applicants. If representation in the grammar stream is much below the 60% figure for the whole intake that gives another reason to question the success of the endeavour.

The Chief Executive of ULT claimed that:

‘…the initiative provided “unusual opportunities to a group of young people who come from an area of genuine deprivation” and that the trust “wouldn’t rule out” repeating the model elsewhere.’


Another ULT Academy already operates a grammar stream 

In fact a grammar stream had already been introduced into ULT’s Glenmoor and Winton Academies in September 2015, a year ahead of Swindon’s initial foray.

Glenmoor and Winton are neighbouring single sex 11-16 academies in Bournemouth, Dorset, so a selective area.

Both schools were last inspected in June 2015 and rated Good. The reports say that the most able students generally achieve well but are not always sufficiently challenged.

The grammar stream arrangements are subtly different to those at Swindon.

This description relies on the official brochure,  a presentation by the Executive Principal prior to the first intake, a presentation by the Grammar Stream Co-ordinator (September 2016) and ULT vacancy details for that post.

According to these sources:

  • It is overseen by an Assistant Principal, has its own Co-ordinator and a Tutor in each school.
  • It contains up to 30 girls and 30 boys and covers only KS3 and KS4, though the schools are said to be introducing a sixth form from 2017. Both schools are 6FE, each admitting 180 pupils into Year 7, so the grammar stream accounts for one-sixth of their intake.
  • Currently 20% of the grammar stream is eligible for pupil premium. (The 2016 Performance Tables give an ‘ever6’ FSM rate of 23.7% for Winton and 27.1% for Glenmoor, so this is very respectable, while not perfect.)
  • Pupils are admitted automatically if they have passed the Bournemouth selection test, and these applicants take priority for admission, so in theory the proportion of places available for others could be limited.
  • Other applicants must take a dedicated grammar stream exam in April, after they have accepted a school place, thus circumventing the embargo on increased selection.
  • The exam tests ‘ability in Maths and English’. In fact this sounds like an attainment-based exam. The Executive Principal refers to the 2015 edition involving ‘a creative writing essay focussing on ‘school is and isn’t magic’’ and a maths test sampling work from levels 4-6 of the national curriculum. To be successful applicants must exceed the benchmark in both. Successful candidates are admitted in order of their marks until places are full.
  • The KS3 curriculum includes maths, English language and literature, science, computer science, history, geography, French, Spanish, Ethics Religion and Philosophy, drama, music, design technology, art and PE. (The brochure refers to triple science instead of science/computer science). The minimum homework expectation is 7.5 hours per week. Pupils are ‘taught together for all of their lessons’.
  • At KS4 all pupil pursue the EBacc, but they may also specialise in Art and Technology or Expressive Arts. The Executive Principal says every pupil has ‘an A* target for all GCSEs’. Each will also have ‘a clear career plan to take them through A levels and on to a Russell Group university’.
  • Grammar stream students also benefit from a ‘specialist enrichment offer’. This includes The Brilliant Club, free musical instrument tuition and membership of the Grammar Stream Band, the Bronze Crest Award and university outreach activities. (The Executive Head refers instead to a programme to develop ‘cultural capital’ including annual foreign visit and exchange programme and annual music, theatre and arts trips. It is not clear whether one has replaced the other.)
  • Pupils’ progress is closely monitored. The school subscribes to ‘rank order attainment’. Students are sorted by ‘twice yearly rank order attainment tests’. According to the Executive Principal they are also ‘reseated every six weeks based on ‘highest lowest’ shoulder partners’.
  • There is no movement in and out of the stream mid-year but there most certainly is at the year end. The entry policy says:

‘Students who are in the Grammar Stream will be expected to be placed in the top 30 in the summer rank order assessment examinations. Students who are not placed in the top 30 may lose their place in the Grammar Stream.’

  • The brochure confirms that seven boys and six girls (so around 20% of the cohort) were replaced at the end of 2015/16 by students from lower sets. More than half of these 13 replacements were eligible for pupil premium. (We are not told how many of those leaving were similarly eligible.)
  • The Principal describes this practice in glowing terms:

‘The unique difference between this and the local grammar school offer is movement into and out of the grammar stream is fluid. Students can move in or out depending on progress and performance making the grammar stream a ‘reality for all’ encouraging real social mobility.’

  • The responsible Vice-Principal adds:

‘Glenmoor and Winton are not selective academies, and indeed, we are vehemently against selection…. We are passionate in our belief, however, that 11 years old is too young to put a child on a path which will be the right path for them for their entire secondary school career. Things may change for children who begin Year 7 in the Grammar Stream which mean it is no longer the best option for them. Similarly, children who at age 11 were not ready for the rigour of the Grammar Stream, or perhaps for reasons beyond their control were not in a position to perform well in the entrance tests, may, later on, become ready for it – often, in our experience after a period of the high-quality education we offer here reveals their ‘true’ ability as well as raises their aspirations. For this reason, children at Glenmoor and Winton sit regular and robust assessments, and if the results of these suggest any should move out of or into the Grammar Stream, then that option is made available to them.’


Issues with emerging ULT practice 

There are notable differences between the Swindon and Glenmoor/Winton grammar streams.

The Bournemouth schools are located in a selective area so need to compete with their neighbouring grammar schools for a share of high attaining learners. The schools are well-established with good reputations and the proportion of their intake that is disadvantaged is not too much higher than the national average.

Swindon Academy is in a non-selective area (though reportedly still experiencing some pull from geographically distant grammar schools). It is struggling to recover from a difficult past, to overcome a poor reputation. It has been managing low demand and a large number of surplus places and it has a highly disadvantaged intake compared with the national average.

The Bournemouth stream is mid-way through its second year of operation, while the Swindon stream is mid-way through its first. Hence both are at a comparatively early stage of development. It is too early to obtain reliable information about the impact on attainment.

Nevertheless, one hopes that rigorous independent evaluation was set in place from the outset, so that it can have a formative influence on this emerging practice and supply the necessary baselines for summative evaluation.

It will be important that evaluation includes comparison with the ‘next best alternative’ which in this context may well be within-school selection via setting.

Demand seems to have been relatively strong from the outset in Bournemouth, whereas the Swindon stream is struggling to gain traction, even with the heavy involvement of a prestigious independent school.

This is a worrying warning sign. The Academy is investing heavily in information and publicity, but that is not sufficient to bolster an unpromising reform. Swindon needs to commit now to full implementation in the longer term – and begin to generate and publish a steady stream of hard evidence demonstrating positive impact.

It would also be helpful to give parents and learners more clarity about determination of the ‘pass mark’and the potential for ‘conditional offers’.

The starkest difference in practice between the two grammar streams lies in their approach to transfer in and out. In Swindon this is merely an afterthought, because the prime motive is to attract in new pupils to fill empty places. Provision for greater fluidity is necessary, however, to avoid one of the principal shortcomings identified in the research literature.

Conversely, the Bournemouth stream is taking fluidity to the extreme. There is a real risk that movement inwards and outwards on this scale will seriously destabilise the stream and so reduce its positive impact on participants’ attainment.

Secure all-round high attainers may be unaffected, but there will be a group of less secure members of the stream, some with ‘spiky’ attainment profiles, who are at far greater risk. This group may include disproportionately large numbers of disadvantaged high attainers.

They are unlikely to benefit greatly from annual movement in and out of the grammar stream, engagement in a repeated ‘relegation battle’. This will be a critical element in the evaluation of the Bournemouth scheme.

Overall, both reforms could work up to a point, but the evidence base to prove that is not yet in place. It would be wrong to dismiss these issues as teething problems – each of these grammar streams is undermined by a serious weakness that requires urgent attention.


Conclusion – summary of key points

The key points from this analysis are:

  • The selection green paper envisages new partially selective free schools selecting by ability and/or aptitude, but seems not to countenance the introduction of partial selection – as opposed to full selection – into existing comprehensive schools, or increasing the proportion selected in schools that are already partially selective. This may be a fault with the drafting rather than a concrete policy decision.
  • The green paper is remiss in not discussing within-school selection options alongside between-school selection – or wider support for (disadvantaged) high attainers. The only foray into within-school selection is the proposal to establish MAT ‘centres of excellence’ for high attainers, which is essentially a hybrid – a within-school model that operates across two or more schools.
  • Since publication of the green paper, increased partial selection has been advocated by three influential stakeholders, each of which has referred directly to the advantages of ‘grammar streams’: Toby Young, subsequently director of the New Schools Network; The Centre for Social Justice, some of whose ideas have been picked up by Tim Leunig, DfE’s Chief Scientific Adviser and Chief Analyst; and Mary Curnock-Cook, the outgoing Chief Executive of UCAS and member of Swindon Academy’s governing body.
  • All three are guilty to some extent of confusing the typology of between-school selection. It is important to distinguish three categories of partially selective schools: bilateral schools, schools with historical arrangements for partial selection by ability and aptitude; and specialist schools able to select up to 10% of their pupils on aptitude.
  • Amongst the group of partially selective schools with substantial cohorts selected on ‘general ability’, few if any operate grammar streams. Most deploy a combination of setting and mixed ability classes. This is logical because the non-selective cohort in partially selective schools is typically comprehensive.
  • Grammar streams can operate completely independently of the school’s admissions process, provided that selection takes place post-admission (or after confirming acceptance of a place, as in the case of the ULT case studies). It follows that grammar, partially selective and comprehensive schools may all operate grammar streams. Legislative change is not required to makes this practice more widespread. This is the nub of the confusion within much of the advocacy outlined above.
  • Setting and streaming are separate and distinct forms of within-school selection which are unhelpfully conflated in too much of the research literature – and in the EEF Toolkit. Setting places learners in a discrete subject-specific class, ostensibly on the basis of attainment in the subject that is being taught in that class. Streaming puts learners in a cohort that operates across several subjects, and potentially their entire school learning experience. Selection for streaming often foregrounds measures of generic ability. Even if there is an attainment-based element it will likely be confined to core subjects. It follows that, if learners are set by prior attainment in all subjects, the selected group will be different to that created through streaming.
  • There is huge confusion between the concepts of ability and attainment. Attainment is what the learner can achieve now, in the present, in a given subject context, or combination of contexts. It is typically measured by scaled scores (KS2) or GCSE outcomes (KS4). Prior attainment is a measure of achievement at a given point in the past. Ability is a judgement or prediction of what the learner may be capable of achieving in the future derived from evidence about their capabilities or potential. This may include evidence of attainment but will most likely foreground cognitive ability tests, for example in verbal, non-verbal or visual-spatial reasoning.
  • This confusion has been compounded by efforts to force a distinction between aptitude and ability. DfE presently follows the lead of the School Adjudicator in arguing that aptitude is the potential to attain, while ability is actual attainment. But ability and attainment are completely different concepts, while aptitude and ability are very closely related and may even be regarded as synonymous.
  • The present distinction is nonsensical because it equates ability with attainment. If there is any distinction between ability and aptitude it must rest on ability being generic – applying independently of specific subjects or fields of performance – and aptitude being confined to a specified subject or field of performance. One might also restrict the concept of aptitude by specifying that it should not be influenced by evidence of prior attainment or performance in that subject or field. This confusion should be rectified when the School Admissions Code is revised to reflect the changes proposed in the green paper.
  • There is a dearth of recent research evidence on the effects of streaming in the UK. I could find very little at all about the impact of grammar streams. Streaming is too often conflated with setting, so we are unable to distinguish the impact of one from the other. The perceived wisdom is that streaming has a more negative effect on lower attainers and a less positive effect on higher attainers because it is more inflexible, less well targeted and so more inequitable.
  • Streaming became highly unpopular in the last part of the 20th century, but appears now to be gaining in popularity, in primary schools as well as through the introduction of secondary grammar streams. That said, there is no reliable data about the incidence of grammar streams. Those schools which do operate a grammar stream tend to be located in existing selective areas and to use the policy to help them compete with their fully selective neighbours. There are exceptions however, as the ULT case studies demonstrate.
  • The evidence about the effects of both setting and streaming is often misleading because any benefits for (disadvantaged) high attainers are typically discounted given the disbenefits for lower (disadvantaged) attainers and arguably (disadvantaged) middle attainers too. Moreover, the evidence usually relates to typical poor practice rather than the most effective practice, designed to circumvent the shortcomings identified.
  • The available evidence suggest that the shortcomings of typical streaming practice include the following points: lower streams may experience a comparatively impoverished curriculum – and curricular differences restrict upward movement into higher streams; higher streams may benefit from higher quality teaching, better and more experienced teaching staff; those in lower streams may develop lower aspirations and poorer attitudes to learning, while teachers (and parents) may have lower expectations of them; movement between streams is too limited and infrequent; and downward movement is associated with loss of esteem which may impact negatively on attainment. The unanswered research question is to what extent these shortcomings can be overcome through effective practice (and the EEF study is only considering setting).
  • ULT is a multi-academy trust (MAT) which operates two grammar streams, one introduced in September 2015 in Glenmoor and Winton schools in Bournemouth; the other in September 2016 in Swindon Academy. The latter has received far heavier media coverage. The contexts in which these two grammar streams operate are markedly different, and the issues they are designed to resolve are also very different.
  • Although information is available about how these grammar streams operate, there is nothing about how they are being evaluated. In the absence of evaluation, we are forced to judge for ourselves whether the policy designs adopted are likely to avoid the pitfalls outlined above. We are also forced to attempt our own approximate judgements on their effectiveness compared with alternatives, such as setting and/or mixed ability classes.
  • This review suggests that the critical issue of movement in and out of the grammar stream has barely registered as yet in Swindon, and so must be addressed with some urgency. Conversely, the Bournemouth stream has embraced fluidity of movement to such an extent that it is ‘devil take the hindmost’. This is almost certainly creating a raft of different problems that will need to be addressed. Developing learners’ ‘resilience’ will help, but only up to a point.
  • The evidence suggests that the Bournemouth stream is relatively popular, in that there is quite strong demand for the places available. But reaction to the Swindon experiment is decidedly mixed and recruitment to the first cohort has been poor, despite heavy investment in information and publicity. We do not know whether the proportion of the cohort from disadvantaged backgrounds is broadly representative of the 60% pupil premium eligibility rate amongst the wider school intake. Several of those accepted into the cohort are borderline candidates who may struggle to fulfil the high expectations placed upon them. The impact on NOR has been moderate, with 45 more school places filled but 44 still vacant. Swindon will recruit again in 2017 but seems less sure about longer term viability, though that is needed to give critical mass to the endeavour.
  • ULT is in a difficult position since it apparently wishes to position itself against increased selection, as envisaged in the green paper. It needs to provide a convincing explanation of why it is anti-selection yet pro-grammar streams. It makes an attempt to defend its position in this response to the green paper, but is not very convincing. The claims it contains about movement in and out of the grammar stream and mixing with other pupils ‘in a range of subjects’ are not fully consistent with existing practice at Swindon as outlined in the documents summarised above.
  • Both ULT grammar streams are bold experiments that should undoubtedly benefit some higher attaining learners. To the (limited?) extent that these are disadvantaged learners, the cause of social mobility will be served. But there is serious cause to question whether – on the basis of the evidence before us – they will be any more effective than setting for those high attainers, and the disbenefits for lower (and possibly middle) attainers may well be proportionately greater.

Whether the follow-up to the green paper will result in more widespread adoption of grammar streams is not yet clear, though the fact that this does not require legislation will be an obvious attraction.

On the basis of this review I would urge caution – and press for properly evaluated pilot studies as a precursor to any wider implementation.

And I commend to all once more this comprehensive 10-point plan to radically improve system-wide support for high attainers, especially those from disadvantaged backgrounds.



February 2017





One thought on “Investigating grammar streams

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