How to measure student belonging consistently across all your schools

Ask each academy in a trust how belonging is going and the answers tend to arrive in seven different formats. One head reports a happiness question, another quotes an engagement measure, a third offers a summary from the last assembly. None of it can be set side by side, so the board cannot see where support is needed and a head cannot see how their pupils compare with those down the road.

Consistency is what turns belonging from a set of impressions into a planning tool. The seven ideas below follow the order in which most trusts build it, from agreeing the definition through to acting on the results and measuring again.

1. Agree what belonging means before choosing questions

Imagine two academies in the same trust that both report "belonging". One asks pupils whether they are happy at school. The other uses an engagement measure built around lesson participation. Both numbers are reasonable, but they describe different things, and the board cannot compare them until someone settles what the trust is actually trying to measure.

A workable definition is usually short. Many trusts describe belonging as pupils feeling accepted, valued, included and safe to be themselves at school. That single sentence then breaks into four to six themes, such as relationships with staff, relationships with peers, inclusion in school activities and fairness. The themes become the frame that every question later hangs from.

One approach is to hold a single working session with headteachers and inclusion leads, rather than circulating a draft for comment. Disagreements surface faster in the room, and heads who helped write the definition are more likely to use it. The session also needs to produce phase-specific wording. A Year 2 pupil and a Year 11 pupil do not read "I feel included" in the same way, so primary and secondary versions of each theme need agreeing at the same time, while keeping the underlying theme identical.

The common mistake is allowing each school to adapt the definition to its own priorities. It feels respectful of local context, but it makes trust-level comparison meaningless, because the numbers no longer describe the same thing. Local priorities can be handled later, through a separate local section of the survey.

Two checks show whether this has worked. The first is whether every academy has signed off one definition and one theme list. The second is the percentage of survey items that map to an agreed theme. Any item that fits no theme is a candidate for removal or for the local section.

2. Use the same validated question set in every academy

Suppose five secondaries each run a Year 7 survey. If all five use the same closed-response set, the statement "I feel I belong at this school" can be compared directly, with the same wording, the same agree/disagree scale and the same position in the questionnaire. If each school has written its own version, the five results look similar on a slide and mean very little.

Wording matters more than it seems. Small changes to a statement, a different number of scale points or a different question order can move scores without any change in how pupils feel. Identical instruments remove that noise, so a difference between two academies is far more likely to reflect a real difference in experience. The principles behind better belonging survey questions apply here, especially around clear wording and consistent scales.

Many trusts choose one validated set per phase and lock the core wording. A short, clearly separate local section then gives each school room for its own questions, perhaps about a new house system or a lunchtime change. Keeping the local section at the end protects the order of the core items, and keeping pupil questions to ratings, agree/disagree statements or multiple choice keeps the data clean to analyse.

The usual pitfall is editing questions between terms or schools, often with good intentions after a head feels a statement is clumsy. Each edit breaks comparison across schools and trend tracking within a school. A better route is to log suggested changes and review them once a year, applying any agreed revision to every academy at the same time and flagging the break in the trend line.

To judge progress, look at the share of academies using the unchanged core set, and at whether item wording has stayed stable across terms. Both should sit at or near 100 per cent.

3. Fix the timing and conditions of every survey window

A Year 7 survey in week 3 of the autumn term captures pupils who are still settling in and have had few chances to fall out with anyone. The same survey in week 10 captures pupils after half-term tests, a wet-play spell and the first friendship reshuffles. If one academy runs in week 3 and another in week 10, the gap between them may owe more to the calendar than to culture.

Delivery conditions create the same problem. A survey completed silently in a form room with a teacher walking round differs from one completed in a noisy ICT suite where pupils can see each other's screens. Pupils who suspect their answers are visible tend to give safer ones.

Trusts that manage this well usually set a trust calendar with a two-to-three-week window each term and ask every academy to complete within it. A short briefing script means every pupil hears the same explanation of purpose and anonymity, and brief guidance on supervision tells staff where to stand and what not to say. Schools can still choose the day inside the window, which leaves room for trips, assessments and INSET days.

The common error is letting schools pick their own dates around local events. Every school has a reason, and each reason is sensible, but the result is a spread of measurement points across the term. Fixing the window at the start of the year, before diaries fill, avoids most of these conversations.

Two figures show whether it is holding: the percentage of academies completing within the window, and the variation in completion dates. A narrowing spread over successive terms is a good sign that the process has become routine.

4. Raise and protect response rates

A school reporting a 90 per cent response overall looks well covered. If only 55 per cent of pupils with SEND completed the survey, the school has a blind spot, and probably in the group where belonging matters most. The headline rate hides it, and the results will lean towards the pupils who were easiest to reach.

Representation is the real aim. Pupils who are frequently absent, who find reading hard or who have English as an additional language are the least likely to be heard in a loosely organised survey, and they are often the pupils leaders most want to understand. Good student voice practice depends on hearing from all groups, not only the most accessible ones.

Several practical choices tend to help:

1. Scheduling the survey in form time or a timetabled lesson, rather than leaving it for pupils to complete at home.
2. Briefing staff so they can explain the purpose and answer questions consistently.
3. Telling pupils and families what the survey is for and what will happen with the results.
4. Providing translation and read-aloud support for EAL pupils and for those who find reading difficult.
5. Agreeing a minimum response threshold, below which a result is not compared across academies.

The mistake to avoid is treating a good overall response rate as enough. The check that matters is response by cohort, set against the agreed threshold. A trust might decide, for example, that any cohort below a stated response level is reported with a caution rather than ranked.

Measure response rate by academy and by cohort each term. Where a cohort falls short, a follow-up catch-up session for absent pupils is usually more productive than extending the window.

5. Break results down by cohort, not just by school

Consider Year 7 belonging scores across five academies. The averages sit close together. Filter to Pupil Premium pupils only and the picture changes: one academy shows a wide gap between disadvantaged pupils and their peers, while another shows a narrow one. The second school has something worth learning from, and the first has something worth looking at, and neither finding was visible in the average.

That is why whole-school averages can mislead. A high average does not mean everyone belongs, since a large majority can mask a smaller group with a very different experience. Comparing like-for-like groups, such as Year 7 against Year 7 or SEND pupils against SEND pupils, is what exposes those gaps. This is also central to inclusive education, where the experience of specific groups matters as much as the overall picture.

Trusts usually build cohorts from MIS data rather than asking pupils to self-identify, which keeps the groups consistent across schools. Common cuts are year group, SEND, Pupil Premium, EAL, FSM and children in care, plus combinations where numbers allow. Setting a minimum cohort size protects both accuracy and pupils, since a result based on six children is unreliable and may make individuals identifiable. Building a standing item into each trust meeting, where the largest gaps are reviewed, stops the analysis becoming a one-off exercise.

Reporting only whole-school averages is the first pitfall. The second is drawing firm conclusions from very small groups, where one or two pupils can swing a percentage. Pooling across two terms or flagging small groups as indicative are both reasonable responses.

The figure to track is the gap between each cohort and the whole-cohort score, termly, for each academy. A narrowing gap is a more telling sign of progress than a rising average.

6. Set up one trust-wide view with benchmarks and trends

A board pack that shows each academy's belonging trend beside its attendance and suspension data tells a different story from any of the three measures alone. Suppose one academy's belonging scores have slipped for two terms while attendance still looks steady. That pattern prompts a targeted support conversation well before the attendance figures move, and it is the kind of evidence that feeds an attendance improvement plan.

A single format matters because trustees rarely have time to learn a new layout for each school. When every academy appears in the same structure, with the same scales, the eye goes straight to change. Term-on-term movement and the trust average give context, and a result that looks low may turn out to be typical for the trust or for the phase.

Many trusts consolidate results into one dashboard or report template and pair it with attendance and behaviour data already held centrally. The belonging figures then sit inside existing conversations rather than arriving as a separate pastoral report that is easy to set aside. Our guide to measuring belonging in a multi-academy trust covers this kind of reporting structure in more depth.

The most damaging mistake is using the data as a league table. When heads expect to be ranked, they have reason to manage the survey rather than learn from it, and candour drops. Presenting the view as a way to direct support, with questions such as "what would help this academy?" rather than "why are you bottom?", keeps reporting honest. Showing each school's own trend as the primary view, with the trust average as a reference line, helps.

Two measures show whether the view is working: whether the board looks at the report every term, and how many support decisions can be traced to belonging data, such as a school improvement partner being deployed or a transition programme being reviewed.

7. Turn the data into named actions and re-measure each term

Take a Year 8 cohort whose belonging scores are low on peer connection. The useful response names a lead, agrees an action such as a mentoring or peer-connection programme, sets a review date, and plans to re-measure the same cohort next term. Without those pieces the result becomes a slide that everyone has seen and no one owns.

Follow-up is also what keeps the measure running. Pupils and staff notice whether anything happens after a survey. When nothing visibly changes, participation falls and answers become less considered, which damages the consistency the earlier strategies worked to build.

In practice, schools that do this well tend to assign an owner and a date to each low-scoring cohort, and to make the action specific enough that it can be checked. A short "you said, we did" message to staff and pupils closes the loop, and can be as simple as a form-time slide saying that pupils reported fewer chances to meet people outside their class, and that a new lunchtime club has started. The headteacher then reports actions to the board each term, alongside the scores, which also answers the kind of trustee questions on inclusion that boards increasingly ask.

The recurring failure is collecting data without visible follow-up. A related risk is spreading effort across every low score at once. Picking the few cohorts with the widest gaps, and doing those properly, usually produces more change than a long action list.

Two checks apply here. One is the share of flagged cohorts that have a named owner and action. The other, and the more important, is the change in those cohorts' scores at the next measurement, which shows whether the action worked and feeds the next round of decisions.

How Satchel Pulse can help, and a first step for this term

A sensible first step is to agree one definition, one question set and one survey window with all headteachers, pilot them, and report the first comparable results to the board. That pilot will show where response rates are thin, which cohorts are too small to compare, and how much the board wants to see each term.

Perceptions, from Satchel Pulse, supports that pilot with ready-made belonging surveys, distribution through MIS integration, automatic reporting, comparison with previous surveys, and cohort breakdowns by year group, Pupil Premium, SEND and other groups. Results from every academy can be viewed in the same format, which gives trustees one place to see trends. Satchel Pulse's own analysis suggests belonging data can point to attendance and behaviour concerns well before they appear in those figures, though that is our finding rather than independent research, and it is best tested against your own trust's data.

Learn more about our services