Methodology preview
UK School STEM Workshop Outcomes Benchmark
A prospective analysis protocol for future workshop outcome evidence.
This is a methodology preview, not a findings report. No benchmark result is presented, and this page does not claim that a workshop causes an outcome.
Protocol version: 0.1.0 · Evidence date: 23 July 2026 · Reviewer role: Research protocol owner · Status: prospective protocol
Current evidence boundary
What the current data cannot support
The historic feedback source is useful for designing a future study, but it is not a valid outcomes benchmark.
Post-only eligible funded cohort
Current feedback covers a post-only eligible funded cohort. It has no matched pre-workshop baseline, so it cannot estimate change in an outcome.
First-workshop attribution limitation
Existing responses may be attributed to the first workshop recorded for a booking or participant context. Where exposure could include more than one workshop, the data cannot reliably isolate an individual workshop effect.
Contract-funded distribution
The observed cohort follows the distribution of contract-funded delivery. It is not a random or representative sample of UK schools, pupils or all Hyett Education bookings.
No PII publication
No personally identifiable information, pupil-level record or public DBS identifier will be published. Any future public output will use aggregated data and the approved disclosure-control rules.
Prospective protocol
Analysis plan before any findings
These decisions must be completed and versioned before eligible outcome data are analysed.
1. Primary outcome
The primary outcome will be the change in a pre-specified STEM outcome measure between matched pre-workshop and post-workshop responses. The construct, item wording, scoring direction and aggregation rule must be frozen before the first analysis.
2. Pre/post instrument
The same version-controlled short instrument will be used before and after the workshop. Collection windows, administration instructions and scoring rules will be fixed in the protocol before data collection used for a benchmark begins.
3. Unit of analysis
The intended primary unit is a matched pupil response, nested within a workshop cohort and school. A de-identified matching key may connect pre and post responses, but it remains pseudonymous personal data while it can be linked inside the controlled workflow. School-level sensitivity analyses will be specified before results are calculated.
4. Child data governance
No eligible pupil data collection may begin until a documented privacy and DPIA review has approved the lawful basis, controller and processor roles, school and parent or pupil transparency route, data minimisation, access controls, retention and deletion schedule, and incident handling. The instrument, primary outcome, matching rule, eligibility and exclusions must be frozen before the first eligible collection.
5. Cohort matching
Pre and post responses will be matched only within the same eligible workshop cohort using a de-identified key. Unmatched responses will not enter the primary change estimate and will be reported separately in the response flow.
6. Exclusions
The protocol will pre-specify exclusions for responses outside the eligible cohort or collection windows, duplicate records, invalid match keys, and records without the primary outcome. Each exclusion reason and count will be reported before any outcome estimate.
7. Missing data and attrition
The response flow will show eligible, invited, pre-workshop, post-workshop, matched and analysed records. Missing items and loss between stages will be reported. Any imputation or weighting method must be specified before outcome data are analysed; otherwise the primary analysis will use the pre-specified complete matched cohort and state that limitation.
8. Denominators
Every percentage, mean and subgroup summary will state its pupil, cohort and school denominator. Denominators will not be inferred from a different stage of the response flow.
9. School clustering
Pupils from the same school are not independent. The final statistical plan will account for school clustering and will include a school-level sensitivity analysis. The model choice will be fixed before outcome data are reviewed.
10. Small-cell suppression
Numeric small-cell and complementary-suppression rules are pending a pre-specified governance decision. Cells below the approved threshold will not be published. Findings publication is blocked until the rule is fixed before analysis and all published cells meet it.
11. Minimum sample thresholds
Minimum matched-pupil and school thresholds are pending a pre-specified governance decision. No benchmark finding will be published until both thresholds are fixed before analysis and the eligible dataset meets them.
12. Limitations
Any future report will discuss selection into the programme and survey, self-report measurement, programme and school differences, missing data, attrition, clustering and the absence of random allocation. A pre/post change cannot by itself establish that the workshop caused the change.
Governance
Publication gate and version control
This preview is version 0.1.0. The instrument, outcome definition, exclusions, analysis model, minimum sample thresholds and disclosure-control rules must be approved and recorded in a later protocol version before results are calculated. Findings publication remains blocked until the pending numeric thresholds are fixed in advance and met.
Questions about the protocol can be sent through the Hyett Education contact page.