Policy simulator

Pull the right lever, on Value Added v2.1: what the model says would happen to a school’s results, or a whole area’s, if absence or intake were different.

NoteHow to use it
  1. Choose a school, or switch to every school in an area to apply the same change across a district, authority, region or England.
  2. Move a lever. Absence is the lever schools and councils can most affect. The intake levers ask what the model expects if a school’s pupils were different, as an admissions change might make them.
  3. Read the result for every group of pupils at once, the gap between disadvantaged and other pupils, and how much each lever contributes.

The address bar keeps your choices, so a link takes someone else to the same scenario.

How it works

The simulator uses the fitted Value Added v2.1 models, one for each group of pupils. Each model explains the log of a school’s Attainment 8 by its intake and setting, so every lever changes a school’s expected result by a percentage:

\[\text{Attainment 8}_\text{new} = \text{Attainment 8}_\text{latest} \times \exp\Big(\sum_j \beta_j \, \Delta x_j\Big)\]

where \(\Delta x_j\) is the change in each lever (in logs for absence, disadvantage and EAL, which enter the model as logs) and \(\beta_j\) is its coefficient for that group. Everything else about the school stays as it was, including what the model cannot explain: its value added. A school starts from its latest published result (2024/25 for most), not from its prediction.

For all pupils, cutting a school’s absence by a tenth (say from 9% to 8.1%) goes with Attainment 8 2.5% higher, about 1.2 points at the average school. For disadvantaged pupils it is 3.5%, and for non-disadvantaged pupils 1.9%. The same cut in absence helps disadvantaged pupils more, so it narrows the gap.

In area mode, every school in the area gets the same change, and the area’s result is the average of its schools’ results weighted by their number of pupils.

What it can and cannot tell you

  • It is the model’s arithmetic, not a forecast. The coefficients describe how results differ between schools with different absence and intake, all else equal. They are not experiments. A school that cut its absence would not necessarily gain exactly this, and how it cut it would matter.
  • Absence is not a dial. The absence lever says what better attendance would be worth, not how to get it. See How school absence changes for what moves it.
  • The intake levers are not policies. They show how much of a school’s result is its pupils. Moving them is a way of seeing that, or of thinking through an admissions change, not something a school can do.
  • The groups are modelled separately. Each group has its own coefficients, so the same change can do more for one group than another. The all-pupils result is its own model, not an average of the others.

What changed from the original simulator

The original Pull the Right Lever simulator ran on the first model. This version follows the review’s recommendations:

  • No Ofsted grade or workforce levers. They are no longer in the expectation.
  • New levers: SEN support, EHC plans, pupil mobility and neighbourhood education.
  • Girls and boys are modelled as groups alongside disadvantaged and other pupils.
  • Area mode: the same change across every school in an area.
  • It runs in the browser, with no server.