Written and maintained by CASRAI Editorial Board
Last updated
The Research Excellence Framework (REF) does not produce a single research-quality score for a UK university. It produces three separate scores per subject-based Unit of Assessment (UOA), each built from evidence of a genuinely different kind, then combined into an overall quality profile. Understanding outputs, impact and environment as three distinct evidence-gathering exercises — not three sections of one narrative — is the single most useful mental model for anyone new to preparing, managing or interpreting a REF submission. See REF (Research Excellence Framework, UK) for the umbrella definition of the exercise itself; this guide goes one level deeper into how each of the three elements actually works in practice, across both the completed REF 2021 cycle and the current REF 2029 cycle.
The three elements, and why their names keep changing
Every REF cycle to date has scored each UOA submission on the same three elements on a five-point scale (4*, 3*, 2*, 1*, unclassified): the quality of research outputs, the wider real-world impact of that research beyond academia, and the environment that supports research activity. What changes between cycles is each element’s name, weighting and evidence format — a source describing "REF weighting" without naming a specific cycle is describing a snapshot, not a fixed structure.
| Element | REF 2021 (completed) | REF 2029 (current cycle) |
|---|---|---|
| Outputs | Outputs — 60% | Contribution to Knowledge and Understanding (CKU) — 50% |
| Impact | Impact — 25% | Engagement and Impact — 30% |
| Environment | Environment (REF5a/REF5b narrative statements) — 15% | Strategy, People and Research Environment (SPRE) — 20% |
The direction of travel across every cycle so far — REF 2014, REF 2021, and now REF 2029 — has been to reduce the weight on outputs alone and increase the weight on impact and environment, a deliberate signal that publication count and journal placement are not, on their own, what the exercise is built to reward. For the full historical detail behind each cycle, see REF 2021 and REF 2029.
Outputs: what gets submitted, and who counts as eligible
The outputs element assesses the quality of the research itself — journal articles, monographs, datasets, practice-based outputs and other formats, read and scored by discipline-based expert panels through peer review, informed but not driven by citation data. Two mechanics determine what an institution can actually submit:
- Staff eligibility. REF 2029 introduces a structural change here: outputs are decoupled from individual staff and submitted at institution level rather than as staff-output pairs. In place of the old REF 2021 portability rule, REF 2029 uses a "substantive link" test — an eligible employment relationship (minimum 0.2 FTE, at least 12 months’ continuous employment, with a role descriptor carrying an explicit research-activity expectation) between the submitting institution and an author who made a significant contribution. See REF 2029 Output Portability Rules for the full decoupling mechanics.
- Output eligibility period. Outputs must fall within a defined lookback window — two years prior to publication for standard outputs, extended to five years for monographs and other long-form, extended-process outputs (with a 100-word justification required and panel assessment, not automatic acceptance). See REF 2029 Output Eligibility Period for the confirmed detail.
Open-access compliance is a separate but closely linked gate: outputs published from 1 January 2026 onward must be deposited within three months of the date of first publication (rather than three months from acceptance, the REF 2021-era rule for outputs published through 2025) to remain REF-eligible. See UKRI Open Access Policy for Journal Articles and REF 2029 Requirements for the deposit mechanics.
Impact: evidencing effect beyond academia
The impact element is judged almost entirely through impact case studies — four-to-five-page documents each covering a summary of the impact claimed, the underpinning research, references to that research, and evidence (including named corroborators) supporting the claim. The underpinning research must have been carried out at the submitting institution and must clear a minimum quality bar, conventionally at least 2* on the REF’s own scale. Case studies are scored by discipline sub-panels on two dimensions — reach (how widely the impact was felt) and significance (how much difference it made) — using the same star-rated scale as outputs and environment.
Impact’s share of the overall quality profile has grown across every cycle to date: 20% in REF 2014, 25% in REF 2021, and 30% under REF 2029’s "Engagement and Impact" framing. The REF 2021 impact case study database (publicly searchable at results.ref.ac.uk) has become one of the largest public collections of evidence-based research-impact narratives in existence, and is now used as a research corpus in its own right for studying impact typologies, not only as a compliance archive. See REF impact case study for the full evidentiary structure.
Environment: from REF5a/REF5b to SPRE
The third element assesses the strategy, people, income, infrastructure and collaborative activity that make a UOA’s research possible — the institutional conditions surrounding outputs and impact, rather than either directly. REF 2021 evidenced this through REF5a (institution-level) and REF5b (UOA-level) narrative statements. REF 2029 renames and restructures the element as Strategy, People and Research Environment (SPRE), weighted 20% of the overall quality profile and scored through two components:
- Institution-Level Statement (ILS) — 60% of the SPRE score.
- Unit-Level Statement (ULS) — 40% of the SPRE score, with two additions that have no REF 2021 equivalent: a Statement of Representation and a Research Community Statement.
Both statements are organised around the same four sections: context/mission/strategy, people, income/infrastructure/facilities, and collaboration/engagement/impact. SPRE’s working name during its pilot phase was "People, Culture and Environment" (PCE); readers should treat any source still citing a "People, Culture and Environment (25%)" figure as describing that earlier pilot framing rather than the confirmed SPRE structure above. See SPRE (Strategy, People and Research Environment) for the full breakdown.
REF 2029’s timeline
Per REF’s own published project timetable, the current plan sets an output submission deadline of Autumn 2028 and results publication planned for December 2029. REF 2029’s own guidance site is explicit that every guidance module — including SPRE and the outputs decoupling rules above — is still being finalised through 2026 and "may be subject to small revisions," so treat any specific date or figure here as the current project plan rather than a locked-in commitment, and re-check against 2029.ref.ac.uk before relying on it for institutional planning.
Why this matters for research administration
Preparing a REF submission is a multi-year institutional exercise, and each element creates a distinct administrative workload: research offices coordinate output selection and verify eligibility (including the substantive-link test above) for outputs; commission, draft and evidence impact case studies well ahead of the census point; and compile institution- and unit-level environment/SPRE statements that require input from well beyond the research office itself — HR data on people and culture, finance data on income and infrastructure, and strategy input from senior leadership. Because REF results also determine the allocation of roughly £2 billion a year in England’s block ‘quality-related’ (QR) research funding, institutional decisions about which outputs to submit, which case studies to resource, and how environment narratives are framed are not academic exercises — they are recurring, high-stakes funding decisions with a multi-year runway. See REF Rankings by Subject and Institution for how these scores are actually consumed and compared once published.
REF and the wider research-assessment reform movement
REF’s own trajectory — away from a crude output count and toward peer-reviewed, multidimensional evidence across outputs, impact and environment — sits alongside a broader international reform movement CASRAI documents elsewhere. The San Francisco Declaration on Research Assessment (DORA) and the Coalition for Advancing Research Assessment (CoARA) both push institutions away from journal-level metrics as a proxy for individual research quality, and CoARA’s UK national chapter explicitly engages with REF as one of the assessment systems its signatories are working to reform. See DORA and CoARA: Comparing Research Assessment Reform Commitments and DORA vs CoARA for how these frameworks compare, and the Hong Kong Principles for a third, complementary responsible-assessment framework frequently discussed alongside REF, DORA and CoARA.
Frequently asked questions
What are the three elements of the REF?
Every REF cycle scores each Unit of Assessment submission on three elements: the quality of research outputs, the impact of that research beyond academia, and the environment (people, infrastructure and strategy) supporting it. The names and weightings of these elements have changed across cycles — see the comparison table above for REF 2021 versus REF 2029.
How is REF impact actually assessed?
Almost entirely through written impact case studies: four-to-five-page documents describing a real-world effect of research, the underpinning research behind it, and evidence (including named corroborators) supporting the claim. Panels score each case study on reach and significance using the REF’s star-rated scale.
What replaced REF 2021’s Environment element?
REF 2029 renames Environment as Strategy, People and Research Environment (SPRE), weighted 20% and scored through an Institution-Level Statement (60% of the SPRE score) and a Unit-Level Statement (40%), the latter adding a Statement of Representation and a Research Community Statement with no REF 2021 equivalent.
When is REF 2029 due?
Per REF’s own current project timetable, the output submission deadline is planned for Autumn 2028, with results publication planned for December 2029 — both explicitly described by REF’s own guidance as the current plan, subject to update as the exercise progresses.
Who actually submits to the REF?
UK higher education institutions make submissions per subject-based Unit of Assessment. REF itself is run jointly by four UK funding bodies — Research England, the Scottish Funding Council, the Higher Education Funding Council for Wales, and the Department for the Economy Northern Ireland — whose expert subject panels score each submission.
