Source use and currentness
The source question for this repertoire is: Which cross-sector and design-family sources change research membership, allocation, operations and trace, the claim warranted by analysis, the useful examination of a credibility threat, warranted corpus synthesis, exact knowledge revision/return and situated Method improvement or retention?
| Source | Role in the available patterns | Limit preserved |
|---|---|---|
| OECD, Frascati Manual 2015 | Five-criterion cross-sector R&D comparator for RMP.1. | Primarily a statistical R&D methodology; it neither decides truth nor supplies a universal institutional gate. RMP.1 adapts all five criteria explicitly. |
| UNESCO, Recommendation on Open Science | Holder-neutral citizen/community participation and plural research actors. | Openness is constrained by ethics, safety, rights, confidentiality, security, access, and community authority. |
| MRC/NIHR, framework for complex interventions | Uncertainty-led plural perspectives, non-linear entry/reconsideration, and stopping for RMP.2. | Its intervention phases and health scope are not exported as a universal research lifecycle. |
| O’Brien et al., Standards for Reporting Qualitative Research | Transparency and flexibility cues for qualitative/reflexive work. | A reporting standard is not a design selector or proof of execution. |
| American Historical Association, Statement on Standards of Professional Conduct | Careful source use, methods/assumptions, evidence-interpretation relation, bias awareness, and alternative interpretations. | Professional standards do not supply one historical claim, archive access, or universal holder requirement. |
| Cochrane, Handbook for Systematic Reviews of Interventions | Question-led eligibility, search, appraisal, and synthesis relation. | Intervention-review guidance is not a universal evidence hierarchy or a substitute for a traced qualified corpus. |
| National Academies, Reproducibility and Replicability in Science | Distinguishes computational reproduction from new-data replication. | Neither move is a universal successor, validation result, robustness result, or decision authority. |
| SCORE 2026 analytical robustness and replicability studies | Consequential analytical alternatives and explicit comparison criteria in RMP.5/RMP.6. | Their social/behavioural sampled fields, rates and tolerance conventions are not universal claim thresholds. |
| Flake and Fried, Measurement Schmeasurement, and Goos et al., measurement reliability, validity and reporting, 2026 | Methodological candidate and bounded failure evidence for connecting construct, measure, administration and context in RMP.3, then locating the actual use in RMP.4. | Reporting and reuse do not establish validity in the present context; Many Labs findings are not a universal research failure rate. |
| Lakens, Sample Size Justification | Quantitative information/precision rationale tied to the question and dependence, rather than an unexplained sample count. | A resource-limited design can remain insufficient; numerical planning and unlike qualitative/source-critical adequacy retain their applicable Methods. |
| SPIRIT 2025 and COS TOP 2025 | Consequential protocol detail and distinct timed registration, protocol and analysis intentions. | Trial reporting is not design-quality appraisal; a label does not establish a prospective commitment or require universal freezing. |
| Haven et al., qualitative preregistration Delphi study, and Braun and Clarke, RTARG and review | Informative starting commitments with method-appropriate evolution; a correction to treating generic reporting or coder agreement as universally appropriate. | Neither consensus nor the reflexive-TA critique validates one procedure for every qualitative tradition. |
| Rethlefsen et al., PRISMA-S | Search-version, source/platform/date and retrieval transparency for an actual trace. | Search reporting does not perform search, appraisal or synthesis; the supplied review question and qualified Methods govern those acts. |
| Miske et al., Investigating reproducibility, 2026, and Workflow Run RO-Crate | Failure evidence and an implementation example for the particular input–transformation–run–output connection. | The study concerns papers from 2009–2018 and obtainable datasets. A provenance format is optional and supplies neither scientific truth nor a performed reproduction test. |
| Carroll et al., Operationalizing CARE and FAIR | A correction to openness-only reasoning: actual data authority, purpose, access and the criticism each view permits. | CARE retains its Indigenous data-governance basis; FAIR is not identical to open access, and a policy does not establish inspectability. |
| European Commission, responsible generative AI use in research, third edition, May 2026 | Proportionate records of research-shaping AI contributions and the responsible person’s verification and decisions. | Non-binding guidance distinguishes substantial contribution from routine editing; retained outputs do not guarantee service replay or authorize disclosure. |
| Cochrane chapters 5, 10, 12, 13 and 15 | Reports/studies, quantitative combination and alternatives, missing contributions and bounded interpretation in RMP.7/RMP.8. | A model must fit its quantity and dependence; intervention-review guidance supplies neither a universal hierarchy nor a receiving decision. |
| Cochrane chapter 21, GRADE-CERQual and JBI mixed synthesis §8.3 | Context-preserving qualitative synthesis, finding-level confidence and question-relative integration or separate synthesis. | A confidence judgment is of its stated kind; transformed findings retain their source meaning and losses. These Methods do not turn an interpretation into prevalence. |
| Morris, White and Crowther, simulation-study design; Williams et al., simulation transparency; Pawel et al., failures and non-convergence | Aim, estimand, generating situations, performance, uncertainty and consequential failures for a selected RMP.9 comparison. | The established design anchor and newer reporting work do not select a study, supply its result or establish general Method superiority. |
Currentness was checked for the inherited question/design uses on 4 September 2026, for the operationalization/trace methodological comparison on 6 September 2026, for the analytical and credibility distinctions on 8 September 2026, and for the synthesis and revision contributions on 8 September 2026. The Method-improvement source comparison is qualified on 8 September 2026 at its stated statistical, reflexive and historical reach; its Method Engineering stopping and acquisition contribution was rechecked on 9 September 2026. A later edition number alone reopens nothing. Reopen the affected source claim when a new source changes a criterion, family-specific criticism, participation boundary, or use limit; when a cited current web source changes the relied-on content; or when a serious omitted alternative can change an RMP disposition.
Sources are selected by answer-changing role, not citation count. Pattern-body SoTA sections state the exact mutation and serious alternative. Domain sources still govern the actual failure mechanism, salinity measure, historical source, instrument, safety condition, or other claim at issue.
What the sources contribute to the shared architecture
The question/design foundation connects two source questions. Frascati supplies an explicit cross-sector comparison for deciding whether Work is research; UNESCO’s participation account prevents that comparison from becoming an affiliation gate. RMP.1 adapts the criteria to a bounded, criticism-ready contribution, including historical and citizen work that need not promise universal generalization or a repeatable event. The criteria and their changes are recoverable in RMP.1’s Frascati comparison. Neither source makes an institutional classification a finding about truth or a study authorization.
For the second question, the MRC/NIHR framework contributes uncertainty-led selection, reconsideration, and stopping. RMP.2 adapts those moves while leaving its health-intervention phases within their source scope. SRQR and the AHA account keep situated interpretation and historical source criticism visible; the Cochrane account makes a synthesis question govern the corpus relation; the National Academies distinction keeps reproduction and replication from becoming interchangeable credibility claims. Together these contributions support comparing what a design can expose instead of ranking method labels. The detailed contribution and serious alternative are in RMP.2’s SoTA account.
This is a methodological synthesis for the pre-study use. It is not evidence that one source supplied the question/design architecture or empirically validated its results. In particular, the PostSelectionNeed counterfactual is the repertoire’s explicit rule for completing selection: it makes the consequence of an unknown value inspectable. The worked holds and selections explain that rule; the cited reporting standards do not certify it.
For operationalization and trace, the selected best-known line is question-relative specification with an inspectable account of actual change and use. Flake/Fried and Lakens make consequential measurement and information choices explicit; SPIRIT supplies bounded specification distinctions and TOP separates timed commitments. Haven and the RTARG critique preserve an approach’s justified evolution while challenging generic checklist use. RMP.3 adapts those contributions to operational choices and precise returns, not to one mandatory protocol.
RMP.4 adapts source-critical, search and run-provenance distinctions to actual material. CARE/FAIR and responsible-AI guidance make data authority, permitted access and substantial human or tool contributions consequential to that account. The applicable owners’ rules determine what can be retained and disclosed; the trace states which examination remains possible. The alternative of preserving and publishing everything can add burden and harm while still omitting a decisive choice. The alternative of keeping only a final methods narrative loses timing and selection. The chosen arrangement accepts selective recording work to preserve those particular distinctions; a competent existing protocol and trace remain the cheaper sufficient route.
For analysis and credibility, SCORE’s distinct examinations make analytical choices, obtainable computation and the comparison criterion consequential. RTARG and AHA preserve unlike interpretive and source-critical warrant. RMP.5/RMP.6 use those contributions to connect a bounded claim, its remaining threat and the test’s possible result; existing adequate analysis and a qualified no-new-test answer can complete the present use.
For corpus synthesis, Cochrane makes dependence, a justified quantitative model, explicit alternatives and missing contributions consequential. Its qualitative guidance and JBI preserve a choice between same-question integration and separately synthesized aspects; CERQual supplies a bounded finding-level confidence approach. AHA source criticism keeps historical dependence and chronology distinct from narrative count. RMP.7 adapts those contributions to the actual research question. RMP.8 combines bounded interpretation with FPF episteme and affected-use relations so that the result changes the exact account and reaches its actual receiver.
For Method improvement, the established statistical simulation-design line and newer transparency and failure-handling work make the selected comparison’s target, performance and limits inspectable. RTARG and AHA preserve interpretation and source-use criteria that a generic benchmark would miss. RMP.9 connects those contributions to a research-specific Method decision, with supported retention, bounded correction and a justified acquisition choice kept distinct. Its Method Engineering source account retains ME.14’s distinction between an adequate current choice and further inquiry selected for its useful obtainable contribution.
The resulting cross-family arrangement is a reasoned synthesis, not an empirically validated universal procedure. A stronger comparison, a changed research use or an observed failure can reopen the affected Method; adding sources without changing an answer is not an improvement by itself.
The retained source dates and scopes matter. Frascati 2015 is a statistical-methodology comparator; the National Academies report supplies particular distinctions; reporting guidance supplies transparency obligations within its field. None is presented as a timeless complete research method. The source-currentness statement above remains the qualification for the claims inherited by this version. A stronger current answer or an observed misrouting reopens the affected claim and its consuming comparison, as described there.