Class changes can distort school attendance data when a report assigns past sessions to pupils' latest classes. A class can appear to improve while another declines, even though every register mark stays the same. Before comparing classes, establish whether the report groups pupils by today's roster or by membership on the session date. Keep that choice visible beside the reporting window and the possible sessions.
This England guide is current to 9 October 2026. The ledger below is a recommended review tool for attendance, data, school and pastoral leads. It isn't a census return or a statutory form. Registration drift here means a change in how sessions are grouped, which may reflect a truthful move or an inaccurate source record.
Two classes exchange places without a changed mark
Redwood, Cedar and all the counts in this combined example are invented. They describe no real pupil, family, school, customer or national statistic. Over ten school days, each of 25 pupils in each class has 20 possible sessions. There are no joiners, leavers, closures or other changes to possible sessions in this example.
| Class | Pupils | Possible sessions | Absences | Attendance |
|---|---|---|---|---|
| Redwood | 25 | 500 | 35 | 93.0% |
| Cedar | 25 | 500 | 25 | 95.0% |
Two pupils move from Redwood to Cedar after this window. Together, their earlier records contain 40 possible sessions and 12 absences. A report using today's class labels moves those earlier sessions into Cedar. Redwood becomes 23 pupils, 460 possible sessions and 23 absences: (460 − 23) ÷ 460 × 100 = 95.0% attendance. Cedar becomes 27 pupils, 540 possible sessions and 37 absences: (540 − 37) ÷ 540 × 100 = 93.1%, rounded to one decimal place.
The ranking reverses. Across both classes, the totals remain 1,000 possible sessions, 60 absences and 94.0% attendance. No register mark changed. That whole-group check helps locate the regrouping, though unchanged totals alone don't prove every source record is accurate. A class label has carried more weight than the evidence permits.
Name the comparison before reconciling it
A current-roster backcast is valid for the question, “How did the pupils currently in Cedar attend during the earlier window?” It doesn't answer, “How did Cedar as constituted then perform?” That second question needs membership on each session date. If the class changed within the window, even one roster snapshot may be insufficient.
A fixed pupil cohort keeps the same named pupils across windows; their available possible sessions can still differ. Changing class totals follow membership changes. Neither view explains why attendance changed. The attendance cohort guide covers the wider comparison choice; this ledger checks the membership behind it.
DfE's Monitor your school attendance user guide, updated on 3 September 2026, explicitly uses latest pupil characteristics for filters. Its current Year 8 comparison looks back at those same pupils in Year 7. This is a useful same-pupils view. It isn't historical class reconstruction, and prior-school data may be missing for entry years.
A compact reconciliation ledger
Keep this ledger in an authorised working record. Link to protected source evidence rather than copying pupil details into a widely shared report. Assign an owner and review date to unresolved entries; retain the original extract and its retrieval time.
- Comparison and window
- Record the question, session dates, extract time and grouping rule. Label a current-roster backcast explicitly. A later refresh can change both characteristics and register marks.
- Class movement
- Record old class, new class, effective session date and source. A move within the same school changes class allocation without itself creating admission or leaving dates. Check tutor group and registration class aren't being treated as interchangeable.
- Joining or leaving roll
- Check admission and leaving dates separately from class moves. DfE's service uses data from admission to the latest session, including leavers up to their leaving date. Don't grant a new joiner sessions before admission or quietly remove a leaver's earlier records.
- Historical membership status
- Mark membership as verified, disputed or unknown and name the evidence. Today's roster cannot establish a past class. Where history is missing, state that the historical comparison cannot be reconstructed reliably; don't infer it from current labels.
- Reconciled totals and disposition
- Retain pupil count, possible sessions, absences and rounding for each view. Explain transferred sessions, remaining discrepancies, owner and review date. Correct a confirmed source error through the authorised process, with a dated correction trail. Preserve a truthful move.
DfE's common-issues guidance warns that incorrect admission, deletion and attendance data significantly affect reports. It also identifies latest MIS characteristics and Wonde permissions as relevant checks. A missing value may be a source or access gap; it isn't evidence that a pupil belonged to another class. Use the pupil record checks for broader rollover issues and the register-mark accuracy check when the underlying mark is disputed.
The ICO accuracy guidance supports keeping source and status clear and correcting inaccurate data. It is under review following the Data (Use and Access) Act. An accurate record of a past class remains useful history after a move; it shouldn't be overwritten merely to match the latest roster.
Carry the unresolved question into human review
DfE's Week 39 attendance release, published on 8 October and updated on 9 October 2026, covers data to 25 September. Its aggregate, indicative figures reflect reporting lag and later register revisions. The methodology uses possible sessions as the denominator and explains that rapid estimates change. A national headline rate cannot validate a local class comparison.
Student Radar's Wonde MIS integration and Attendance Analytics let authorised staff inspect date windows, period comparisons, class breakdowns and pupil lists. Today's class labels and a same-pupils comparison don't restore historical membership. Validate class history and roll dates in source records. Software cannot reconstruct missing history or decide why attendance changed.
Keep any resulting enquiry support-first, in line with England's Working together to improve school attendance statutory guidance, updated on 9 July 2026. Patterns don't diagnose, prove harm or cause, blame staff or families, or predict future absence. They must not automatically trigger support, referral, sanction, penalty or legal action. Staff review the evidence with the pupil and family where appropriate. Urgent safeguarding concerns go through the DSL, local safeguarding or emergency procedures; data-quality review must not delay them.
Before sharing the comparison, attach its grouping rule and ledger disposition. If historical membership remains unknown, share that limitation alongside the same-pupils view. Give the reviewer a question they can resolve, rather than a ranking whose meaning has quietly changed.
To discuss how this review fits your school's workflow, contact Student Radar about attendance.
Sources and further reading
- Department for Education: Monitor your school attendance user guide, updated 3 September 2026. Latest-characteristic filters, same-pupils comparisons and admission/leaving-date coverage.
- Department for Education: Monitor your school attendance: fix common issues, updated 3 September 2026. MIS data accuracy, characteristics and permissions.
- Department for Education: Working together to improve school attendance, England statutory guidance, updated 9 July 2026. Attendance review and support responsibilities.
- Department for Education: Pupil attendance in schools, Week 39 2026, published 8 October and updated 9 October 2026; data to 25 September. Aggregate indicative statistics and revisions.
- Department for Education: Pupil attendance in schools methodology, reviewed 9 October 2026. Possible sessions and the limits of rapid estimates.
- Information Commissioner's Office: Accuracy principle guidance, reviewed 9 October 2026; under review after the Data (Use and Access) Act. Source, status, correction and historical records.