How the Admissions Check is tested

For each school a family lists, the Admissions Check says how often families in their position were offered a place in past years: families who lived as far away and put the school first. It says so in words, never as a percentage, and it says “we can’t tell” when the records cannot support an answer. This page shows how those words did when we tested them against years the model had not seen.

How we test it

We rebuilt the model once for each year from 2019 to 2026, each time from the years before it only, and compared what it would have said with what actually happened that year. Each check is one school, one year and one family distance: 84,111 checks for Year 7 and 315,491 for Reception, drawn from 19,464 school-years with a published cut-off.

What it said, and what happened

“What it said” is the Check’s own figure, averaged over the checks in each label. “What happened” is how often families in that position were offered a place on national offer day.

The widest gap was "Rarely" for Year 7: it said 10.0%, and families were offered a place 16.5% of the time, more often than it said. On the standard measure of forecast accuracy (the Brier score, where lower is better) it scores 0.158 for Year 7 and 0.171 for Reception, against 0.197 and 0.196 for distance alone.

When a council publishes no cut-off

Where no distance is published, the Check looks at how often everyone who put the school first got a place. It gives a label from this record only where it is about 7 in 10 or more; below that it says it can’t tell.

Where it is less sure

• Single years vary. Taking each year and each figure on its own, the furthest the Check over-promised was 10.3 points for Year 7 and 11.8 for Reception, and the furthest it under-promised was 10.7 and 11.3.
• Where a council’s past results fell short of our figures, or beat them, the Check adjusts its words for that council and says so beside the label. It never hides a school because of it.
• Within 0.05 miles of the line between two labels, it shows the lower one, because a postcode can sit about 100 metres from a home.

What it doesn’t know

• It describes a family with no priority other than distance. Siblings, faith, catchment areas, looked-after status and an EHCP change the picture.
• It measures a straight line from your postcode. Where a school’s own admission arrangements measure along walking or road routes, or they name no way of measuring and the council’s arrangements for its own schools do, it says it can’t tell. Where we hold neither, it says so beside the label.
• It does not answer for a school that has closed, for selective schools, where a test decides, or for a school whose own admission arrangements give some of its places by a test of ability or aptitude. Nor does it answer for schools that rank faith before distance.
• It gives no figure from distance where some or all of a school’s places go by random draw, where Year 7 places are offered within ability bands (each band has its own furthest distance), or for a school with a religious character whose admission arrangements we don’t hold.
• Our figures were tested up to 2.5 times a school’s cut-off. If you live further away, it gives no figure, except “Rarely” where that was already true at 2.5 times, because the chance does not rise with distance.
• It describes national offer day. Places offered later, from waiting lists, are not in these figures.
• It is history, not a promise. A school’s popularity can change, and so can its admission number.

Where the figures come from

First-choice applications and offers come from the Department for Education’s annual statistics. Cut-off distances come from the councils’ own published offer-day figures, checked against their documents, and every correction we make is recorded. See our data sources and how we write.