The School That Made the Argument Michael Gove Could Not Settle

Katharine Birbalsingh’s Michaela Community School has combined an uncompromising academic culture with results that are hard to wave away: in 2023–24, its Progress 8 score was 2.56, well above average. That makes Michaela an important test of the claim that disadvantaged pupils need less demanding schooling; it does not prove that Michael Gove’s reforms rescued England, or that the same teaching model will work everywhere. For certain, the teaching model of the ‘rest’ does not work everywhere, and when it does not, bias and other progressive excuses are used to rationalise poor performance.

One London school, two arguments about failure

Michaela Community School is a small institution carrying a large political burden. To its admirers, it is evidence that a school can insist on knowledge, order and effort in an inner-city setting and still produce exceptional academic progress. To its critics, it is an outlier: an intriguing case, perhaps, but not a template for a national system. Katharine Birbalsingh’s account of the school, and of the reforms associated with Michael Gove, presses on a familiar division in education debate. When results are poor, commentators reach for structural explanations. When results are strong, the school is often treated as too unusual to teach anyone else a lesson.

That difference in response deserves scrutiny. It does not follow that structural explanations are false, or that a successful school has identified a universal formula. It does follow that both sides should demand explanation. A claim that poverty, discrimination or institutional history constrains achievement must be tested against evidence. So must a claim that low expectations, weak instruction or disorder are responsible. Neither a school’s postcode nor its pupils’ starting points can tell us in advance what they can learn.

Birbalsingh’s central proposition is simpler than the public argument around her: pupils should be given serious knowledge, taught directly, expected to concentrate and held to routines that protect learning. She treats hard work not as a substitute for good teaching but as one of its conditions. A teacher cannot make a child learn by command, but the teacher can refuse to treat difficulty as proof that the child should be offered less. That is a stronger version of “no excuses” than the phrase usually invites. It directs the demand at adults as well as pupils: teach the material, explain it, revisit it, and do not confuse a child’s disadvantage with a limit on the child’s potential.

The question is not whether that vision sounds appealing. It is whether the results support it, how much they support it, and what sort of conclusion they can bear. Michaela’s record is strong enough to make dismissal lazy. The national evidence is complicated enough to make triumphalism lazy too. Between those two failures lies the more useful task: separating the school-level evidence from the national story Birbalsingh attaches to it.

Michaela’s achievement also disrupts a comfortable assumption that rigorous academic expectations and an inner-city school serving a diverse community are incompatible. That disruption has value even if the explanation for the results remains contested. A model need not be universally replicable to expose assumptions treated as settled. But once the example is used to justify national policy, a second burden begins: to show which practices caused the gains, which conditions made them possible, and whether they can be transferred without copying only the visible rituals.

The results are real; the explanation is harder

The strongest evidence in Michaela’s favour is not its reputation. It is the Department for Education’s progress measure. The school recorded a Progress 8 score of 2.56 for pupils completing key stage 4 in 2023–24, in the “well above average” band. The measure compares GCSE progress with that of pupils nationally who had similar key stage 2 results. Of the 116 pupils at the end of key stage 4, 103 were included in the calculation; the school’s confidence interval ran from 2.29 to 2.83. The official service describes pupils as achieving, on average, over half a grade higher in each qualification than similar pupils across the country.¹

Those qualifications matter. Progress 8 is not simply an awards table comparing raw GCSE grades. It attempts to account for where pupils began. Michaela’s 2.56 is not a modest statistical wrinkle: it is an exceptional result on the measure the state uses to compare progress. The figure keeps returning to the centre of Birbalsingh’s argument for a reason. If pupils with comparable prior attainment make less progress elsewhere, then the school has achieved something worth examining.

Ofsted’s 2023 inspection supplies a different kind of evidence. Inspectors judged Michaela to remain outstanding and reported a culture of high expectations, strong behaviour and a rich curriculum. Its governing message, “work hard and be kind”, joins discipline to a social ideal rather than presenting order as an end in itself.^2. Inspection is not a controlled trial, but it is independent corroboration that the school’s routines and academic ambition are visible beyond its founder’s own account.

None of this tells us which part of the Michaela package produced which part of the outcome. Progress 8 cannot isolate curriculum from leadership, teaching, staff training, peer culture, family support, admissions, attendance or the school’s ability to keep a coherent staff team. Prior attainment is useful context, not a complete description of circumstances. It does not make pupils identical at the starting line, and it cannot remove all differences that influence later examination performance.

Nor does a school average describe every pupil’s experience. The measure includes pupils with usable key stage 2 results, not every student who attended the school. Its confidence interval acknowledges uncertainty; it does not dissolve the difference between a cohort-level result and an individual child’s story. A school with strong averages can still have pupils who need more help, or practices that should be changed.

The fair conclusion is substantial but bounded. Michaela shows that a demanding curriculum, systematic teaching and tightly maintained behaviour can coexist with very high measured progress in one school. It undermines the suggestion that high expectations are inherently incompatible with serving children who begin behind. It does not establish that all pupils would make the same gains under the same model, or that every successful school owes its results to the same ingredients.

While on both sides of the Atlantic we are faced with poorly performing inner-city schools with pupils from marginalised populations who massively underperform, politicians refuse to look at solutions for ideological reasons and opposition from the unions. Black educators like Roland Fryer have tried to apply logic and data to the issue, and we have seen a Black-led increase in ‘no-excuse’ networks of schools with mostly African American pupils show vast improvements in outcomes. This shows unions that it is not the raw materials going in (pupils), but the performance and attitude of professional educators that make the difference.

Knowledge is not the enemy of thought

Birbalsingh’s educational argument turns on a distinction that is often blurred in political debate: knowledge and thinking are not rivals. A pupil needs something to think with. In history, that means chronology, institutions, people and events; in literature, vocabulary and knowledge of the text; in science, concepts and evidence; in mathematics, fluent recall and procedures that make more complex reasoning possible. An invitation to discuss a subject cannot compensate for not knowing enough about it to discuss it well.

That is why she favours explanation, practice, memorisation, retrieval and testing. These can sound like a return to an older schoolroom, but the useful question is not whether they look traditional. It is whether they help pupils remember and use what has been taught. A teacher’s clear explanation does not rule out a pupil’s curiosity. A factual foundation does not prevent original judgement; it gives judgement material to work on. “Child-centred” is no more a guarantee of learning than “teacher-led” is a guarantee of dullness.

The case for explicit instruction is especially clear where a foundation skill is at stake. In England, the phonics screening check was introduced to test whether Year 1 pupils could decode words using sound-letter correspondences. Ofsted’s research review records a 24-percentage-point rise in the proportion meeting the check’s standard since 2011, while warning that about three in ten disadvantaged pupils still do not meet the minimum decoding standard.³ The figures do not prove that every element of a knowledge-rich school improves outcomes. They do show why systematic instruction and checking whether pupils have learnt a basic skill can matter. They also show that a policy can improve an average and leave a serious failure unresolved.

Evidence for one method should not be inflated into a verdict on every other method. A systematic review by Kayleigh Skene and colleagues examined 39 studies of guided play with children whose mean age ranged from one to eight; 17 studies entered the meta-analysis. Guided play performed better than direct instruction on selected outcomes, including early mathematics and shape knowledge, and better than free play on spatial vocabulary. For many other outcomes, the review found no difference between the approaches. The authors also noted substantial variation in how guided play was defined and delivered.⁴

That review concerns early childhood, not a secondary school such as Michaela. It cannot be used as a shortcut to condemn explicit teaching in older pupils, any more than Michaela’s results prove that teacher explanation should dominate every early-years lesson. The evidence asks for distinctions by age, subject and objective. Children learning spatial language in play are not the same as teenagers learning a complex literary text or the causes of a war.

Birbalsingh is right to insist that visible activity is not the same thing as learning. A lively room can leave pupils with little to recall later; a quiet explanation can lead to real understanding. But activity, discussion and pupil choice are not inherently empty either. The better test is durable learning: what knowledge did pupils acquire, can they remember it, and can they apply it beyond the exercise that introduced it? That question cuts through the branding on both sides.

A quiet classroom is an educational resource

Strict behaviour is the most immediately visible feature of Michaela’s model and the one most likely to be caricatured. Birbalsingh’s argument starts from the classroom, not from a political theory: a teacher must be able to teach, and pupils must be able to listen. If interruptions, intimidation, or constant negotiation set the terms of a lesson, the pupils least able to command attention often lose the most. Order is therefore not a decorative preference. It is a resource everyone in the room shares.

That point is easy to miss when discussion focuses only on the pupil receiving a sanction. Rules also protect the student trying to concentrate, the child who is reluctant to speak over a louder classmate, and the teacher expected to explain difficult material. Predictable routines can reduce the time spent bargaining over basic instructions. A school that asks children to practise hard work while allowing a disruptive minority to make concentration optional is not treating the rest of the class generously.

Michaela’s inspection record matters here because inspectors reported that pupils behaved well and that the school maintained high expectations. It is evidence that the school’s discipline forms part of an observed institutional culture, not merely a slogan. But no school earns a blank cheque by calling a policy strict. A rule can be clear and still be disproportionate. A sanction can protect others’ learning, or it can humiliate the pupil who receives it. Consistency is important, but consistency alone does not establish fairness.

The standard should be demanding in both directions. Adults need enough authority to stop conduct that obstructs learning; pupils need rules they can understand, consequences that fit the conduct and a way to be heard. The purpose of discipline should be to restore attention, safety and the possibility of teaching, not to demonstrate that adults can compel compliance. Respect for teachers should protect their work without making every decision beyond question.

A further distinction exists between a pupil who behaves only to avoid punishment and a pupil who has learned to manage attention and disagreement. The first is compliance; the second is self-regulation. But it is a mistake to assume that removing structure automatically produces autonomy. Children do not become independent simply because adults stop setting boundaries. Clear routines, explained expectations, and repeated practice may help them acquire control over their own conduct.

The real debate, then, is not discipline or kindness. It is what kind of authority makes a school more humane and better able to teach. A quiet classroom can protect pupils whose learning is otherwise easiest to interrupt. It can also become an excuse for inflexible treatment if school leaders stop asking whether the rules serve a purpose. Michaela makes the case for order; it cannot remove the obligation to judge how that order is enforced.

What Gove changed—and what autonomy cannot prove

Birbalsingh’s description of Michael Gove as a hero is not principally a claim about personality. It is a claim about permission. As education secretary, Gove promoted reforms to curriculum, qualifications, assessment and school organisation that gave leaders room to pursue a defined educational approach.

The Department for Education’s record of the coalition years describes academies as having greater control over curriculum, budget and staffing, and free schools as institutions that teachers, parents and other groups could propose in response to local needs. The policy case was that school leaders and local groups should have greater freedom to organise provision in response to pupils’ needs.⁵ Whether one agrees with that diagnosis or not, the institutional change is clear. School leaders had greater scope to organise provision differently; autonomy was an opportunity, not a guarantee of improvement.

For Birbalsingh, this shift matters because a school cannot build a coherent alternative if it must negotiate every decision through layers of oversight that favour the existing pattern. Michaela’s curriculum, routines and staff development can reinforce one another because its leadership can shape them together. Freedom is valuable in that sense: it permits a school to make choices whose effects can then be evaluated.

But permission is not proof. More autonomy can enable innovation; it can also enable failure, fragmentation or decisions with weak oversight. A school leaving local authority control does not stop owing its community an account of how it uses public money and how it treats pupils. Critics of academisation were not necessarily critics of high standards. They could reasonably ask who would scrutinise a school after power shifted from elected local bodies to academy trusts and central government.

The research record urges restraint. In a 2015 analysis commissioned by the Local Government Association, the National Foundation for Educational Research compared academies with similar maintained schools. It found no significant overall performance difference for converter academies on any measure and generally small, mostly statistically insignificant differences for sponsored academies open for two to four years. The report said it was too early to judge the full impact of converter status, while noting tentative evidence of an upward trend among sponsored academies the longer they had been open.⁶

That study does not evaluate Michaela, nor every policy in Gove’s programme. It does challenge the leap from “autonomy allowed this school to exist” to “academy status raised attainment across England”. Gove can be credited with opening institutional space for new schools and different approaches. Whether that freedom improved outcomes depends on what schools did with it, and whether their practices worked. A policy-maker can be a catalyst without being the proven cause of every success that follows.

England did not rise; it fell more slowly in some comparisons

The international case in Birbalsingh’s argument is a claim about relative resilience, not an uncomplicated story of rising scores. England’s PISA results in mathematics fell from 504 in 2018 to 492 in 2022; reading fell from 505 to 496. Science moved from 507 to 503, a decline that was not statistically significant. England remained above the OECD average in all three subjects, and its 2022 results were broadly similar to those recorded between 2006 and 2015.⁷

The OECD trend average also fell sharply between 2018 and 2022: mathematics declined from 490 to 475 and reading from 488 to 477. England’s position can therefore look better in comparison even as its own scores decline. “Held the line” is defensible if it means England remained above the OECD average and did not fall as far as some other systems. It is misleading if it is taken to mean that pupils improved or that the country escaped deterioration.

There is another qualification. England did not meet two of PISA’s 82 technical standards in 2022. The national report found that participating pupils had higher academic attainment, on average, and that fewer had been eligible for free school meals than in the population as a whole. The OECD estimated that the shortfall in representation could have pushed some reported results upwards by roughly seven or eight points, though that does not translate mechanically into a corrected national score. The report therefore warns readers to interpret the figures with care. The result still places England above the OECD average, but it is not an immaculate league-table verdict.

PISA measures 15-year-olds’ performance in mathematics, reading and science. It does not identify which teaching method caused the outcome. The assessment does not ask whether a country’s schools use a knowledge-rich curriculum, stricter rules, more formal testing, more digital devices, or a more child-centred approach and then isolate their causal effect. The scores describe a result; they do not tell us why it happened.

TIMSS provides a different comparison, but not a causal answer. Its 2019 study assessed mathematics and science at fourth and eighth grades, across 64 countries and eight benchmarking systems. England’s reported scores were 556 in fourth-grade mathematics, 537 in fourth-grade science, 515 in eighth-grade mathematics and 517 in eighth-grade science—all above the study’s scale midpoint of 500.⁸ The scores show that England had strengths before the pandemic, at different ages and in different subjects from PISA’s test of 15-year-olds. They cannot establish that Gove’s policies created those strengths.

The disciplined reading is neither that England’s system has collapsed nor that Gove rescued it. International data show decline in important areas alongside comparative resilience. That is a real distinction, and a meaningful one. It is still not a causal explanation. A slower fall is not success; it is a clue that demands more investigation.

Correlation is not a cause, but neither is it a dismissal

Birbalsingh’s explanation for weakness in other systems links several trends: less secure knowledge, reduced emphasis on testing, lower expectations, permissive behaviour policies and teaching that places activity ahead of instruction. The sequence is plausible. If pupils are taught less of a subject, they have less to draw on. If tests fail to reveal gaps, adults may not address them. If disruption consumes lesson time, practice becomes harder. If teachers assume difficult content is unsuitable for some children, the content may never reach them.

Plausibility is a starting point, not a finding. PISA scores do not show that a country abandoned knowledge or standards before its results fell. They cannot separate curriculum change from school closures, attendance, staffing shortages, demographic shifts, inequality or the effects of the pandemic. Several of those forces moved together. A national score moving down after a policy change does not prove the policy caused the decline; a score holding up after a reform does not prove the reform prevented a larger fall.

The same discipline applies to assertions about grade inflation and classroom technology. If qualifications no longer represent what pupils know, that is a serious defect. If screens displace focused practice without improving understanding, that is a poor use of time. But neither proposition is established by the international results cited here. A laptop can be a distraction or a useful tool. A test can diagnose a misconception or narrow teaching to a test. These are decisions about implementation and purpose, not effects that a headline score alone can infer.

This is where Birbalsingh’s case risks becoming too neat. “Progressive education” covers a vast range of practice, from guided play in early years to collaborative problem-solving in secondary science. “Traditional teaching” also covers a range, from skilful explanation and deliberate practice to rote routines that leave misunderstanding untouched. To describe the first as the cause of decline and the second as the answer is to replace a complicated empirical question with a binary. Some activities labelled progressive may be poorly designed; some structured teaching may be excellent. The label settles nothing.

The same applies to national comparisons. A school system does not choose one method and apply it consistently to every classroom. Teaching varies within countries, regions, schools and subjects. The average score conceals that variation. Before blaming a fashionable philosophy, researchers would need to establish what was taught, how consistently, to which pupils, over what period, and whether a credible comparison group experienced something different.

Birbalsingh’s argument is strongest as a warning against mistaking novelty for learning and compassion for lowered expectations. It weakens when those warnings are offered as a complete explanation for why other countries’ scores fell. A good test of her claim would compare actual curriculum, instruction, assessment and behaviour changes over time, not attach ideological labels after the fact. The data supplied here identify a question. They do not close it.

The double standard in explanations should be examined

A sharper point lies beneath the argument about outliers. In public debate, a struggling school serving black pupils in the United States may be described through a history of segregation, exclusion and unequal resources. Those histories are not inventions: the US Supreme Court’s 1954 Brown v. Board of Education ruling declared state-sanctioned school segregation unconstitutional, after decades of legally enforced separation.⁹ Discrimination cannot be ruled out because a school also has weak instruction or poor behaviour. But the reverse caution should be just as strong: citing systemic barriers should not make questions about teaching, curriculum or leadership illegitimate.

The standards of explanation should not change when the outcome changes. If a school is failing, structural conditions deserve attention, but researchers should still ask whether pupils receive strong instruction, whether classrooms are safe and whether leaders are addressing learning gaps. If a school is succeeding, its students’ backgrounds should not be treated as a reason to discount the achievement before examining it. Structural disadvantage may make a result more impressive; it does not make the result causally self-explanatory.

That is the legitimate challenge in the way Michaela is sometimes discussed. The school’s success can be praised as an inspiring exception, then cordoned off from policy on the ground that it is an exception. But exceptionality is exactly why researchers should ask what is transferable. A school that exceeds the outcomes predicted by its intake may reveal practices worth studying, or advantages not adequately captured by the comparison measure, or both. You cannot know the answer by deciding in advance that either the method or the context must explain everything.

Still, the analogy should not be overstated. Michaela is not a black American school, and Birbalsingh’s biography is not a substitute for data about the pupils she serves or the particular barriers they face. The source material provides no comparative evidence that her school faces a greater evidentiary burden than schools credited with systemic explanations. That claim may describe a pattern in public argument, but it needs examples and systematic analysis before it can be reported as an established fact. But Birbalsingh does have counterparts in the USA in the form of Black economist Harvard professor Roland Fryer. He to has faced opposition from progressives!

There is also no necessary conflict between personal responsibility and structural analysis. A child can be expected to work hard while a government addresses poverty, disability support, housing instability or unequal access to tutoring. “No excuses” should not mean that adults ignore obstacles; it should mean that obstacles are not used to justify withholding worthwhile teaching. The distinction matters because slogans can absolve either side: one can blame the child’s choices and ignore institutional failures, or name structural forces and stop asking what schools can change.

The strongest fairness test is therefore symmetrical: scrutinise claims of systemic causation with the same seriousness as claims of school-level causation, and take successful practice seriously without pretending it settles every structural question. If a policy intervention has a plausible mechanism, measure it. If a school’s approach appears to work, test its components(Roland Fryer tried this). If you cannot measure an explanation, do not treat it as a magic word that ends the debate.

Fryer’s work points to the experiment still missing

The comparison with Harvard economist Roland Fryer sharpens the issue. Fryer’s education research has included randomised field experiments in urban public schools, including a study of management training for principals in Houston. In that experiment, principals received 300 hours of training in lesson planning, data-driven instruction, observation and coaching. The programme raised achievement significantly in the first year, but not in the second year overall; effects persisted where principals were expected to remain in post and implement the training faithfully.¹⁰

The lesson is not that every successful-school practice can be transplanted, or that bureaucrats and unions alone prevented the spread of better methods. The provided evidence does not establish that sweeping claim. The study shows something more useful: leadership, implementation and continuity can determine whether a policy works. An idea that produces gains in a tightly managed trial may fail when a school lacks time, training or stable leadership. The reverse is also true: a school’s impressive results cannot, by themselves, show which practices caused them.

Fryer’s research matters because it moves from admiration to causal testing. Rather than ask only what high-performing schools have in common, researchers can test a defined intervention against a comparison group and track what happens. Michaela, meanwhile, offers a practical institution to study: its routines, curriculum sequence, staff preparation, assessment and pupil outcomes can be described in enough detail for others to ask what travels. There is no need to choose between the school as a model and the school as an outlier. It can be both a source of hypotheses and a case whose effects are not yet fully isolated.

This kind of inquiry is slow and politically uncomfortable. It requires the patience to run trials long enough to see whether short-term gains endure; to record failures as well as successes; and to accept that a practice may benefit one subject or group but not another. It also requires more nuance than the familiar exchange in which one side attributes failure to society and the other to culture or effort. Judge an intervention by a clear outcome, not by whether its vocabulary belongs to the right political camp.

Claims that teacher unions or public-sector bureaucracy blocked the spread of effective practice may be worth testing, but you cannot insert them as established fact without evidence about specific proposals, decisions, and consequences. Institutional resistance can be real; so can concerns about evidence quality, workload, child welfare and the suitability of a programme for different pupils. “The system blocked it” is not an explanation until you identify the blocked intervention, the actors, and the expected effect.

The public should expect more of both proponents and critics. A school should publish enough detail to make its model intelligible, not just celebrate results. Researchers should test plausible components in other settings. Policymakers should protect room for experimentation while requiring transparent evaluation. That is not a concession to bureaucracy; it is a way to stop every reform being sold on a story and abandoned on the next political cycle.

The useful lesson is a demand for evidence, not imitation

The phrase “outlier school” can end a conversation before it begins. Some schools are genuinely unusual; their admissions, leadership, staffing or local relationships may make replication difficult. But an outlier also marks the limits of what the system currently considers normal. If a school serving children whom conventional accounts describe as disadvantaged achieves extraordinary progress, the first response should be to investigate how, not to insist that the result is irrelevant because it is unusual. A reminder: Michaela is non-selective; the local authority makes the final selection on who enrols at the school!

The same caution applies to the opposite error: turning one school into proof that every school could achieve the same result by adopting a few conspicuous practices. Routines travel more easily than culture; a behaviour policy can be copied without a curriculum that gives lessons intellectual coherence, or staff training that makes instruction effective. A knowledge-rich curriculum is not a booklist. A strict system is not a learning culture unless pupils understand material and progress. A reform should be studied as a whole and also broken into parts, so its mechanisms are testable.

That work should begin with observable outcomes. Do pupils retain more knowledge after a unit? Can they apply it to unfamiliar questions? Are reading gaps identified early and addressed? Does disruption fall without increasing exclusion or damaging pupils’ trust? Do gains appear among pupils with different starting points, including those with special educational needs? Which practices help, which merely produce visible compliance, and which require resources a school may not have? These questions are more informative than asking whether Michaela is “strict” or “progressive”.

The progress data should also be read alongside the circumstances it cannot capture. A result that compares prior attainment can be powerful evidence of progress while leaving open questions about cohort composition, pupil turnover, attendance and support needs. No single indicator should be asked to carry the weight of an entire theory of education. The Department for Education itself advises users not to draw conclusions from a single school-performance measure alone. That caveat is not an attempt to diminish Michaela. It is the standard that makes the praise credible.

On the national question, the same discipline applies. PISA shows England’s mathematics and reading scores were lower in 2022 than in 2018, while still above the OECD average; TIMSS shows strong pre-pandemic results in mathematics and science at two age levels. Those findings give Birbalsingh grounds to challenge the easy assumption that every previously high-performing foreign system is a model England should copy. They do not demonstrate that Gove’s reform programme caused England’s relative position, or that progressive pedagogy caused declines elsewhere.

Gove may fairly be called a hero by someone who values the freedom his reforms gave a headteacher to build a school around an alternative vision. That is a political judgement grounded in real institutional change. It is not a causal verdict on national scores. Michaela’s results may be evidence that rigorous teaching and disciplined routines can produce exceptional progress in a particular place. That is not a narrow finding. It is a reason to ask harder questions of education systems that have been content to call poor outcomes inevitable.

The important distinction is between an example and a proof. Michaela is an example of what one school has achieved, strong enough to deserve serious study and too unusual to settle a national argument alone. England’s international results show comparative resilience, not that a single policy package worked. The next step is neither to dismiss the outlier nor to canonise it. It is to test its practices, measure what transfers, and apply the same evidentiary standard to fashionable explanations of failure and to confident claims of success. The tragedy is that, rather than using the school as a good example, we see an effort at every turn to tear it down!

1. 1. Department for Education, Michaela Community School: Progress measures for 2023 and 2024. https://www.compare-school-performance.service.gov.uk/school/140862/michaela-community-school/secondary/progress-measures-for-2023-and-2024
   2. Ofsted, Michaela Community School inspection record, inspection dated 10 May 2023. https://reports.ofsted.gov.uk/provider/23/140862
   3. Ofsted, “Research review series: English”, updated 15 July 2022. https://www.gov.uk/government/publications/curriculum-research-review-series-english/curriculum-research-review-series-english
   4. Kayleigh Skene et al., “Can guidance during play enhance children’s learning and development in educational contexts? A systematic review and meta-analysis”, Child Development, 2022. https://ueaeprints.uea.ac.uk/id/eprint/90660/1/Child_Development_2022_Skene_Can_guidance_during_play_enhance_children_s_learning_and_development_in_educational.pdf
   5. Department for Education, “2010 to 2015 government policy: academies and free schools”. https://www.gov.uk/government/publications/2010-to-2015-government-policy-academies-and-free-schools/2010-to-2015-government-policy-academies-and-free-schools
   6. National Foundation for Educational Research, “Analysis of Academy School Performance in GCSEs 2014: Final Report”, 1 July 2015. https://www.nfer.ac.uk/publications/analysis-of-academy-school-performance-in-gcses-2014-final-report
   7. Department for Education, PISA 2022: National Report for England, December 2023. https://assets.publishing.service.gov.uk/media/656dc3321104cf0013fa742f/PISA_2022_England_National_Report.pdf
   8. Mullis, Martin, Foy, Kelly and Fishbein, TIMSS 2019 International Results in Mathematics and Science, TIMSS & PIRLS International Study Center, Boston College. https://timss2019.org/reports/wp-content/themes/timssandpirls/download-center/TIMSS-2019-International-Results-in-Mathematics-and-Science.pdf
   9. National Archives, “Brown v. Board of Education”, describing the 1954 ruling against state-sanctioned segregation in public schools. https://www.archives.gov/milestone-documents/brown-v-board-of-education
   10. Roland G. Fryer, “Management and Student Achievement: Evidence from a Randomized Field Experiment”, Harvard University. https://fryer.scholars.harvard.edu/publications/management-and-student-achievement-evidence-randomized-field-experiment