Who am I?

  • Professor of Urban Analytics at the Centre for Advanced Spatial Analysis, UCL
    • data, urban policy, quantitative methods, maps
  • A former state secondary school teacher
  • A Brighton resident, and a parent with a child at a city state school
  • Everything tonight is built from published data, and all of it is open: the analysis, the code and a policy simulator you can try yourself

Tonight in five parts

  1. The situation. Ten schools, a falling cohort, money running out — and a city whose schools do better than it thinks, held back by one thing above all: absence.
  2. How families choose. They read the published exam score, which mostly describes the intake. And they want a school near home — the evidence on that is strong.
  3. Why build a model. Because in 2024 the city argued about options nobody could see the consequences of.
  4. What the model says. It reproduces the city closely. Then it says some uncomfortable things about the levers on offer.
  5. Options. Five objectives, and what each one costs. Almost nothing here is free — except one thing.

1. The situation

Ten schools, and where the children live

The shape of the system is the shape of a closure

CoMArt closed in 2005. It stood in East Brighton, and the catchment system that replaced distance-based allocation in 2008 was drawn around its absence.

That is why the east of the city has two secondary schools — Longhill at Ovingdean and Brighton Aldridge on the Lewes Road — both well short of their admission numbers, and neither of them where most eastern children actually live.

Modelled with a school back on CoMArt’s old site, the city’s mean journey falls and East Brighton’s children stop travelling past each other.

The decisions a city takes about school places outlive the people who take them. Twenty years on, Brighton & Hove is still arranging itself around a building that is not there.

That is the argument for deciding the next one carefully — and for being able to see what it would do before it is taken.

A system built for more children than it has

  • 2,515 Year 7 places across ten secondary schools
  • 2,267 children in 2026, falling to 1,835 by 2035, and those children are already in primary school
  • About 720 empty Year 7 places a year by 2035
  • Nine of the ten schools have been spending their reserves

Longhill takes about 90 children in 2026 on 210 places, and sits at the far eastern edge of the city.

But the money is a city problem, not only a Longhill problem.

The cohort is already falling

Nine of ten schools are spending their reserves

What being in deficit means

  • 7 of the ten schools are maintained by the council, and all 3 that are already overdrawn are among them
  • A maintained school over 5% in deficit must agree a recovery plan with the council, which cannot write the debt off
  • If a struggling school becomes a sponsored academy, its deficit stays with the council, paid from its own budget
  • Under pressure, schools cut teaching assistants, teachers and curriculum breadth, and pupils needing extra support lose most (Ofsted, Making the cut)

A recovery plan assumes the pressure is temporary. A falling cohort is not.

Taking places out only helps if the academies and faith schools share it: cutting only the council’s schools makes their budgets worse.

Poor access and child poverty in the same places

A city whose schools do well

A story of success, not failure

Rank among 151 English authorities, after accounting for intake

Pupils Rank
Disadvantaged 15th
Not disadvantaged 4th
All pupils 5th

The 2024 admissions consultation started from a story of failure. The evidence says this is one of the highest-performing authorities in England once you account for who its schools teach.

The better question is not “what is the city doing wrong?” but “what is it doing right, and what is holding it back?”

The problem is absence, not the mix of pupils

Disadvantaged pupils (the lever the council chose)

  • City: 28.8% · England: 28.3%
  • National percentile: 52th
  • Unremarkable

Absence (the lever left alone)

  • City: 10.8% · England: 8.2%
  • 2nd highest of 152 authorities
  • Among the worst in England

Leave absence out of the model and the city ranks 69th for disadvantaged pupils rather than 15th. That difference, about 1.4 GCSE points per pupil, is roughly what absence costs the city’s disadvantaged children.

What causes differences in Attainment 8 between schools?

How much can a school change?

Moving a child to another building changes neither their attendance nor their circumstances. That is why redistribution is a weak instrument for raising results.

Concentrated disadvantage: the counter-intuitive finding

Across England, after allowing for absence and prior attainment:

  • Not disadvantaged pupils do a little worse in schools with more disadvantaged pupils
  • Disadvantaged pupils do slightly better, not worse
  • Consistent across years and different measures, and in line with the DfE’s own 2015 research
  • The effect is small and runs mostly through schools organised to serve those pupils, not through concentration itself

This is not an argument for concentrating disadvantaged pupils.

It is an argument against assuming that deconcentration helps them, especially when it means longer journeys, and longer journeys mean more absence.

BACA: the local example

How it was seen

  • ‘Requires Improvement’ from Ofsted
  • Among the city’s lowest headline results
  • A catchment campaign for the right not to attend it

What the data shows

  • About half its pupils are disadvantaged, with high absence and low prior attainment
  • The city’s strongest results for disadvantaged pupils, once intake is accounted for

Disadvantaged pupils scored about +5.7 GCSE points above comparable pupils in 2024–25, and +3.1 on the four-year average.

April 2025: Ofsted rated it Good in every area. No surprise on the contextualised evidence.

A single year is noisy, so lean on the four-year figure. It is still far more than redistributing pupils could deliver.

What the council has been working on

The 2024–25 admissions review. Catchments redrawn, open admissions proposed at 20% of places and settled at 5%, a free school meal priority introduced, and admission numbers cut. Its stated aim was to narrow the gap between disadvantaged pupils and the rest.

September 2026 was the first intake under the new map.

The Educational Equity Framework, at People Overview & Scrutiny this week, sets out what comes next: a new framework for 2027, built on disadvantage, attendance, place, and better use of data.

The council’s own evaluation of its 2022–26 disadvantage strategy found “limited evidence of a positive impact on attainment, attendance and exclusions.”

Its papers now say what this work has been saying: attendance is a critical driver, place matters, and the city needs to use its data better.

That is common ground. The disagreement is about which lever to pull — and that is a question a model can help with.

2. How do families choose?

Families choose on the wrong number

The council’s secondary admissions guide tells parents to look at each school’s “National Curriculum test results, public exam results” and its Ofsted reports: figures that mostly describe who a school admits.

A loop the council’s own advice sustains

flowchart LR
  A["Council guidance:<br/>look at exam results"] --> B["Families prefer schools<br/>with higher headline scores"]
  B --> C["Schools with more disadvantaged<br/>intakes are chosen less"]
  C --> D["Their intakes shrink and become<br/>more disadvantaged"]
  D --> E["Their headline scores fall"]
  E --> B
  C --> F["Smaller rolls, less funding,<br/>Longhill's viability at risk"]
  classDef key fill:#EEDEFF,stroke:#3A1857,color:#3A1857;
  classDef hit fill:#fdecd9,stroke:#d95f02,color:#3d1c00;
  class A key;
  class F hit;

BACA is last of ten on the raw Attainment 8 but 4th on value added. It ranks 1st for disadvantaged student value-added.

Objectively a great school - but parents choosing on raw attainment and it is losing out.

Changing the guidance, and publishing a contextualised measure beside the headline, would cost nothing.

What families ask for, catchment by catchment

Cardinal Newman and King’s sit inside the Hove Park / Blatchington catchment without being its catchment schools. Counting only catchment schools would put that catchment at 44% — counting every school its families can walk to puts it at 95%, the highest in the city.

Two things are true at once

Families want a school near home. In the model, living in a catchment multiplies the odds of choosing a school there by between 2.7 and 13 times, depending on the catchment. It is the strongest force in the system after distance itself.

Every consultation the council has run has heard the same thing from parents, and the preference records agree: most families name the school they can walk to.

But that pull is not unconditional. In Hove, 95% of families name a school inside their own catchment. In Stringer/Varndean, 81%. In Longhill’s catchment and Brighton Aldridge’s, two in five.

The difference is not how much those families value a local school. It is what the published score says about the local school.

Proximity brings families to the door. The headline number decides whether they walk through it.

What the consultations actually settled

Admission numbers: what the council proposed for 2026 entry, and what was determined.
School 2025 places Council proposed Determined Outcome
Blatchington Mill 330 300 330 Reduction not upheld
Dorothy Stringer 330 300 330 Reduction not upheld
Longhill 270 210 210 Unopposed

The council proposed cutting Blatchington Mill and Dorothy Stringer to 300 places. Six objections reached the Schools Adjudicator: Blatchington Mill’s governing body, Queen’s Park Primary’s governing body, and four from parents — one submitted on behalf of 500. Dorothy Stringer did not object. Its parents did. The Adjudicator restored both schools to 330, finding the cuts would frustrate parental preference. Longhill’s cut from 270 to 210 was unopposed and stood. Everything else was upheld: the new boundaries, the free school meal priority, the 5% open admissions.

That matters for what follows. The places in the popular schools are still there — and the model says those places are most of what protects families in the city’s tightest catchment from being displaced.

3. Why build a model?

Brighton Secondary School System Simulator

Every option in this talk is a run of this — in a browser, no login, and you can try your own.

Why build a simulator?

You can try a policy before you try it on children.

In 2024 the city argued for months about options whose effects nobody — council, schools or parents — could see in advance. Decisions were taken on assertion, and trust was spent that has not come back.

A model does not settle what the city should want. It answers a narrower question: if you did this, what would follow? Journeys, intakes, the social mix, budgets, who gets displaced.

And it makes the trade-offs visible. Almost nothing here is free: every option that helps one thing costs another, and the argument is about which price is worth paying.

It is accurate enough to be worth using.

Modelled against the places the city’s schools actually offered for September 2026: R² = 0.95.

Against the council’s own count of pupils by home ward, from a source the model never saw: correlation 0.96.

It is fitted to what families ask for, not to the offers it is then tested against — so reproducing the offers is a check, not a circularity.

Every input is published data. Every line of code is open. If you think it is wrong, you can show where.

4. What the model says

A model of the whole city

flowchart LR
  d1["<b>Children</b><br/>census, postcodes,<br/>council projection"]
  d2["<b>Journeys</b><br/>walk and bus,<br/>routed"]
  d3["<b>What families ask for</b><br/>factsheets, catchment<br/>preference table"]
  d4["<b>Places and rules</b><br/>admission numbers,<br/>council priorities"]
  m["<b>Demand</b><br/>children shared by distance,<br/>attractiveness, catchment,<br/>competition"]
  c{"<b>Full?</b>"}
  p["<b>Priorities</b> decide<br/>who is admitted"]
  r["<b>Refused</b> re-offered:<br/>partner school,<br/>then 2nd preferences"]
  f["<b>Intakes</b> →<br/>journeys<br/>disadvantaged mix<br/>budgets"]
  s["<b>29 options<br/>× 3 years</b><br/>and the<br/>simulator"]
  d1 & d2 & d3 --> m --> c
  d4 --> c
  c -- yes --> p --> r --> c
  c -- no --> f --> s
  classDef data fill:#f3f5f8,stroke:#9aa7b4,color:#1f3b57;
  classDef run fill:#EEDEFF,stroke:#3A1857,color:#3A1857;
  classDef out fill:#edf6ef,stroke:#2b7a4b,color:#102a43;
  class d1,d2,d3,d4 data;
  class m,c,p,r run;
  class f,s out;

Every option tonight is a run of this model, built from published data and open to anyone to check or rerun.

It reproduces the 2026 offers

Pull one lever, the effect shows up elsewhere

The 2024 plan, run through the model

Cutting places displaces about six more children. Opening 20% of places displaces about twelve more. Doing both displaces twenty-eight more — the two levers multiply rather than add. And the segregation the whole exercise was meant to reduce goes up.

Why segregation goes up

Longhill — already the most disadvantaged school in the city — gets more disadvantaged. Dorothy Stringer, already below the city average, gets less. Both move away from the middle, which is exactly what the segregation index measures. The 60 places taken out of the two popular schools do not disappear: the children who would have filled them fall back on whatever has room, and what has room is Longhill.

Who actually gets displaced

No child in the city is displaced because their catchment ran out of room. All of it is families in the Stringer/Varndean pair who named only one of the two schools — and Dorothy Stringer has empty places.

What if families read the other number?

Brighton Aldridge goes from 91 children to 175 on 180 places — half empty to nearly full — without a boundary moving or a place being cut.

The number that matters most

Today, 93% of the city’s empty Year 7 places sit in just three schools.

If families read value added, that falls to 75%.

A surplus spread across ten schools is a budgeting problem.

A surplus piled into three is a viability problem — and that is the one Brighton & Hove actually has.

It creates no places and removes none. The city still has 299 empty Year 7 places either way. What changes is which schools are empty.

Journeys also shorten — and most for the children who travel furthest. The extra distance deprived children travel falls from 3.4 to 2.5 minutes each way.

Cost: nothing. Consultation needed: none. The council writes the admissions guide.

One assumption holds this up. It takes families reading a contextualised measure as closely as they now read the headline score — a claim about publicity and persuasion, not a fitted number. The model shows what that would be worth if it worked, not that it would. And value added for a single school is noisy: Patcham’s fall rests on one four-year figure.

Longhill: the arithmetic

Longhill is not failing to fill because of the catchment map. It is failing to fill because of where it is, how it is read, and how many children there are. Every option the city has costs something. Doing nothing costs something too, and the bill arrives on a timetable.

5. Options for the city

Access or mixing? No option does both

Open admissions: what it bought

Raising open admissions from 5% to 20% displaces about 18 children rather than 6 — and leaves the city slightly more segregated, not less.

A smaller Longhill at the top of Elm Grove

Option Longhill intake, 2026 → 2030 Deprived children’s extra journey Whitehawk’s disadvantaged children travel Longhill’s annual position
Today 90 → 71 3.9 min 56 min −£0.8m
Elm Grove, 150 places, map redrawn 106 → 86 0.4 min 23 min −£0.7m
Elm Grove package 117 → 91 0.5 min 22 min −£0.7m

What it gains

  • The largest cut in deprived children’s journeys of any option
  • Most of Whitehawk back in Longhill’s catchment, as part of a redesign
  • More children for Longhill

What it costs

  • A site, not yet costed; Longhill still does not break even
  • Longer journeys for Rottingdean, Saltdean and Woodingdean
  • It depends on families following a new map as closely as today’s

Five objectives, and what each one costs

If the city wants… The lever that works What it costs
Schools children can reach A school nearer East Brighton — Elm Grove, or CoMArt’s site A site, and longer journeys for Rottingdean and Saltdean
Longhill viable Whitehawk back in its catchment, or a smaller school nearer its children Concentrates disadvantage — which the evidence says is not the harm it is assumed to be
No child displaced Leave admission numbers alone in the popular schools Nothing — this is what the Adjudicator already protected
Less segregation Open admissions, redrawn catchments Displacement and journeys, for almost no movement in the measure
Fuller schools, shorter journeys Change the number families read Nothing at all

What the city could do

Priority one: attendance

The city has among the worst absence in England but near-average disadvantage. That is not inevitable, and it matters most for disadvantaged pupils.

Absence needs taking apart, because very different situations sit inside one statistic:

  • a few days’ family holiday by an otherwise regular attender
  • a child with special educational needs who struggles with the school environment
  • a young carer with a long bus journey
  • post-pandemic disengagement and mental health

Each needs a different response, and longer journeys push absence the wrong way.

For admissions and places

  1. Change the admissions guidance and publish a contextualised measure beside the headline score. It costs nothing, needs no consultation, and on the model it moves intakes further than any boundary change — and takes the share of empty places concentrated in three schools from 93% to 75%.
  2. Treat the bus network as an admissions instrument. It widens real choice without taking anything from anyone.
  3. Plan places across the whole city. The fall in pupils has to be shared, including by the academies and faith schools.
  4. Decide about Longhill on the financial timetable, and test a smaller school at the top of Elm Grove properly, with costs stated.
  5. Don’t treat a more concentrated intake as a cost in itself. Ask what serves disadvantaged pupils: attendance, journeys and well-organised schools.

Better information for everyone

The council already holds data that would settle most open questions, for example:

  • where applicants live, and their full preferences
  • the criterion each place was given under, at every school
  • children placed at a school they did not name
  • offers to children living outside the city
  • the catchment preference table, every year

Release it in aggregate, as the council already does with other small-area data.

With it, the ranges in this work become estimates, and the next decision can be tested before it is made rather than after.

The final word

The caution: policy anchored to a single story, made in haste, risks doing more harm than good, especially to a system that is working well.

The encouragement: the raw materials for better policy already exist in published data, and so do tools to test options before they are chosen.

The challenge: getting that understanding to decision makers before decisions are made.

CASA logo
AI4CI logo