Jump to text
On this page
Copy and source
Published source confirmed at last check

Source changed 2026-11-18 21:11:39 UTC · snapshot created 2026-11-18 21:12:35 UTC · last check 2026-11-18 22:30:10 UTC

RMP.4 - Preserve a Criticism-Ready Research Trace

Type: Research Method Practice practitioner method pattern Status: Stable

RMP.4:1 - Problem Frame

Use this pattern to preserve the material for a particular examination of research or to establish how an earlier inquiry produced a reported result. Name the examination and locate the records it needs.

For example, residents measured downstream salinity during a thaw and missed their visit to the reference reach. Their chart preserves the downstream observations. Record which visit they missed, why, what can be established about the reference reach from other sources, and what the remaining observations support.

This pattern connects the course of an inquiry and the material retained for a named examination. Its ResearchTracePackage gives the reader the necessary records or reliable references and explains how plans, conduct and changes bear on the result. The records can reside in several places. Include limitations that affect the intended use.

A package is “criticism-ready” for an examination when the material that examination needs is inspectable under the stated conditions. A partial account can establish where inspection stops and how that limits the conclusion.

During planning or fieldwork, preserve at-risk records while they are available. For inherited work, start with the surviving material. Use RMP.1–RMP.3 if a needed question, design or protocol still has to be developed.

Use an existing notebook, laboratory record or repository when it supports the examination. Apply RMP.4 when a connection between conduct and material, a change or a limit of use needs explaining. Obtain a substantive assessment, computational reproduction, access decision or new research separately when the work calls for it.

RMP.4:2 - Problem

An account written after the result is known can lose an earlier plan, record a missed observation as a blank cell or omit a failed computation. A step called “processing” may hide choices by researchers or AI systems. The reader needs the particular choices and events that bear on the result.

A large collection can omit the decisive exclusion or its timing, be inaccessible to its intended readers, or disclose material outside the authorized use. A public summary can be sufficient for one examination and lack the detail another requires.

What records will let a reader examine the course of an inquiry, and what can be established when material has to be reconstructed, has been lost or has restricted access?

RMP.4:3 - Forces

ForceTension
Capture before lossA later critic needs consequential detail that may disappear before a final claim is formulated.
Proportionate recordsMore detail can enable later questions, while indiscriminate capture increases burden and disclosure risk.
Plans and conductA dated plan can establish a commitment. A later methods account may also provide evidence of conduct. What happened depends on the grounds for that account.
Adaptive inquiryKnowing when an interpretation changed and which outcomes the researcher had seen helps a reader evaluate the change.
AccessA protected original may support a permitted examination that its public rendering cannot support.
ReconstructionSurviving records can establish part of a history and leave the rest open. A reader needs to know which claims the records support.
ReuseDifferent examinations of the same records can require different detail, provenance or access.

RMP.4:4 - Solution

Specify the examination, preserve the material it needs while that is possible, and connect the records to the choices and results they support. Identify what the reader can inspect, what can be reconstructed, and how loss or limits on access affect the answer.

RMP.4:4.1 - Name the examination and choose the entry

State the question the examination should answer. For instance, the investigator may need to explain an exclusion, establish whether a comparison took place, inspect the basis for an interpretation or find the inputs to a calculation. Identify the relevant design and material where they are known.

During preparation, identify what will later show whether the comparison took place and what material the proposed analysis needs. If a result is reported later, connect it to the material used to obtain it. An examination of an earlier plan can finish before any result or ClaimCandidate exists.

Choose the entry that fits the work:

  • Planned or ongoing work. Identify the question, design and operational choices that matter to the examination. Arrange to capture the needed records while their contents can still be established.
  • Existing or inherited work. Start with the question, any claimed result and the surviving material. Recover the design, protocol and conduct to the extent supported by records, testimony or other evidence.

Use a compatible question, design or operational decision that is already available. RMP.1–RMP.3 can help develop a missing contribution. If an earlier design is uncertain, give the basis of any reconstruction and the date it was made.

Identify the part of the study, sources, encounters, specimens or computation in question and the relevant period or version. For a different examination, check whether the material still suffices.

RMP.4:4.2 - Preserve the smallest material whose loss would matter

Ask what the critic would be unable to distinguish if one record disappeared. Start with those distinctions. Examples include which specimen was measured, which document was inspected, why a source was excluded, or whether an outcome was already visible when a choice was made.

For ongoing work, capture the event or choice while it can be observed. If a suitable laboratory or data practice already preserves it, cite the record with its relevant qualifications.

Keep the versions and dates of the plans relevant to the examination, including amendments. A registration, protocol and intended analysis can record different commitments at different times. To establish a prospective commitment, check its contents and timing against the work.

Distinguish how each part of the account is supported.

Basis for the accountHow to report it
A contemporaneous record supports the event or choice.Identify its scope and the particular claim it supports. Check the basis of recorded statements on which the examination depends.
A person reports the event later.Identify it as a later report and retain what supports or limits that account.
Surviving sources support a reconstruction.Mark the reconstructed part, its sources, reasoning and uncertainty.
The relevant conduct, reason or timing is unknown.Say what is unknown and which examination depends on it.

A single notebook can contain all four kinds of account. Date any reconstruction and identify the evidence that places an intention before or after access to the data. If its timing cannot be established, leave it open.

Apply the retention, access and destruction requirements that govern the material. Where permitted, preserve evidence of loss and explain its effect on the examination.

RMP.4:4.3 - Connect actual conduct, transformations and uses

Follow a consequential reported result back to the observations, sources, encounters or computations used to obtain it. For a planned comparison, identify the material it would require and what has been collected so far. Show how selection or transformation connects the material to its intended or actual use.

For observations, retain the identity of the specimen, case, participant, site or sample at the permitted level of detail and the conditions needed to interpret it. Distinguish instrument indications, qualified measurements and derived values. Use FPF C.16 when an examination depends on units, uncertainty or qualification that the recorded values alone leave unclear.

For documentary sources, specify whether the investigator used a catalogue description or inspected the document. Preserve the edition, locator, dating, provenance, reasons for selection and any relevant transcription or translation. Identify contrary evidence and unexamined collections where they affect the answer.

For transformations, show which operation produced the relevant output from which input. Record filtering, recoding, deduplication, aggregation, translation, corrections or computational steps that affect the examination, with versions and parameters where they matter. Use FPF A.10 when a claim needs its evidence and provenance made explicit.

For a reported computation, link the output to the particular run, its inputs, code and relevant environment, parameters and random seeds. Include failures and manual interventions that affect interpretation. A folder that contains several input versions may leave the version used in the run unknown.

For qualitative work, retain the material and interpretive choices on which the findings depend. Explain a change in the researcher’s position or questioning when it matters. Apply a coding-agreement criterion when the selected approach and examination warrant it.

For evidence synthesis, preserve the searches performed, including platforms or sources, dates, query versions and retrieved material. Trace screening, duplicate handling, exclusions and changes in extraction. These records can explain how the corpus was obtained. To establish appraisal or synthesis, locate evidence of the work performed.

In a mixed inquiry, preserve how each component contributed and how the investigators combined or interpreted their results. Identify shared data and assumptions. If two findings rely on the same participants or retrospective source, account for that dependence when assessing corroboration.

RMP.4:4.4 - Explain consequential changes and contributions

When a change from an earlier intention matters, identify the earlier plan, what changed, who made the choice, when it was made, and what material it affected. Give the supported reason and the data or results available at the time. Identify any part of that history that remains uncertain.

A missed observation, an excluded case, a failed run, an inaccessible source and a revised interpretation have different consequences. Identify the event. For example, an intact record may have been excluded from an analysis, whereas a missed observation has no recorded value to exclude.

If the outcome was known when a choice was made, retain that timing and the grounds for the choice under the method used. An advance commitment needs earlier evidence. Date any later sensitivity analysis or independent criticism separately.

Identify the people, teams, tools or services whose contributions affected the inquiry. For AI-assisted work, preserve the following details when they matter to the examination.

  • The model or service, its known version or configuration, and the time it was used.
  • The relevant input context and the output used, to the extent permitted by the applicable protection and access conditions.
  • What was checked against which source, who or what performed the check, what was accepted, rejected or revised, and how that affected the work. Where the applicable research guidance assigns responsibility to a human researcher, identify that researcher’s verification and decisions.

Preserve the wording of a prompt when it can change how a reader understands the research choice. Routine spelling assistance usually contributes nothing to that question. If a service withholds its internal state, describe what can be examined from the inputs and outputs that survive.

If a service suggests a nonexistent source, preserve the suggestion, the check and the resulting decision when they affect the examination. These records can establish the service’s contribution and the response to it. Any scientific claim that relied on that citation still needs a source that exists and supports the claim.

RMP.4:4.5 - Determine actual access and the consequences of loss

Establish which material survives and which parts the intended reader can inspect under the applicable permissions. A repository locator identifies a location, a policy sets conditions, and a custodian can report what is available. Compare the detail in the accessible version with the detail the examination needs.

For the proposed examination, determine what each available source or view supports.

Material availableWhat the material supportsUse or remaining need
The intended reader can inspect the relevant originals in context.The reader can examine the named connection when its necessary material is present.Use or supply the available material. Undertake scientific analysis or criticism when its result is requested.
Originals survive under restricted access. A public version contains less detail.State what the reader can examine in the accessible version.Use that version if it suffices. If further detail matters, use authorized access or a qualified inspection result where available. Otherwise state the limit on the conclusion.
Surviving sources support a reconstruction.The reconstruction can answer a historical question within the scope of its evidence.Give the sources, reasoning and uncertainty. If an essential distinction is still open, use stronger evidence when available or withhold the dependent conclusion.
A needed record was never made, was lost or can no longer be retained.Identify the affected examination, what can still be established and any narrower use of the remaining material.Report the gap when that answers the question. If further work depends on missing evidence, identify what it would need. New observations or a new run have their own design and trace.
Loss of evidence or access defeats a design premise or changes the research question.Identify the affected relation and what the remaining material supports.Reconsider the design under RMP.2 or the question under RMP.1 when further work depends on it. Preserve what was established about the original inquiry.

A custodian’s qualified inspection can answer a question within its stated scope. To establish that another reader has access, check that reader’s permission and ability to inspect the material.

RMP.4:4.6 - Return the trace with a truthful receiving scope

Give the reader enough information to locate the material and understand its connections and limits. For the examination at issue, identify the following.

  • The examination, research question and study portion, with the design as far as it is known.
  • The surviving plans and the grounds for claims about what took place.
  • The material used and the selections or transformations that affected it.
  • The timing of consequential changes and the contributions of people, tools or services.
  • What can be inspected or reconstructed and which conclusions access restrictions or losses limit.

One record can support several parts of the account. Link to suitable records and specify their identities or versions where later changes could make a reference ambiguous.

Trace a connection that matters to the examination. For a reported value or interpretive excerpt, locate the source and the choices that shaped the result. For a planned comparison, identify what material it would need and what is already available. If a needed record or access is missing, explain which part of the examination remains possible.

A limited package can answer a question about the surviving record. For example, “The downstream measurements and handling decisions can be inspected. The reference reach was inaccessible, and we have found no comparable reference measurement for the same thaw.”

RMP.4:5 - Archetypal Grounding

The following examples specify hypothetical events and records for examining research.

RMP.4:5.1 - ResidentStream-S17: preserve a useful trace after losing the comparison

The selected design would compare salinity-related observations at a downstream reach and a reference reach during an informative thaw event. Assume the measurement method, calibration basis, safe access and intended site-period relation were qualified when the design was chosen.

During the event, downstream observations are made, but the reference reach becomes inaccessible. The team retains the following linked account:

Retained materialWhat the account says and permits
The design and the plan that existed before the eventThe team planned to compare measurements from both reaches during the same thaw. The downstream record also permits a description of that reach.
Downstream sample and measurement recordsSample IDs connect time, reach, hydrological context, instrument indications and the qualified measurement results, including the relevant uncertainty.
The reference-visit recordThe team missed the planned visit because it could not access the site. The record gives the basis for that explanation.
Preparation of a later chartEach downstream value can be traced to its measurement record. The chart identifies the missing reference observation and the resulting limit on comparison.
Effect on the inquiryWith no comparable reference observation available, the planned comparison lacks one member. The downstream observations remain available for a narrower question. If the inquiry needs a different design, use RMP.2.

The reader can inspect the downstream measurement and handling history. In this case the team’s visit was missed and no other comparable observation of the reference reach during the same thaw is available. If a suitably qualified measurement from that reach and event were found, the reader could reassess the original comparison.

A reference sample taken later in dry weather documents a different hydrological condition. To use it for a comparison, establish the particular relation that makes it comparable and state which question that comparison answers.

If volunteers started sampling before recording a protocol, their labels, instrument logs and later accounts may establish parts of the sampling procedure. Date the reconstruction, identify its grounds, and distinguish a reconstructed action from evidence that the volunteers had committed to it in advance.

Apply the existing permissions to private land-access details. The public account can identify what role those details played and which questions the published material supports.

RMP.4:5.2 - Pump-P4: distinguish absence, exclusion and interpretation

On the valid-marker branch of the seal study, specimen P4-07 is damaged during handling before one planned observation. The trace retains its allocation and material identity, the handling record, the time and basis of the reported damage, the absent observation, the applicable prior exclusion intention and the decision actually taken.

The comparison dataset links each remaining observation to its specimen and explains why P4-07 contributes no value at that point. Three readings from a second seal provide repeated measurements on that single specimen.

A critic can now assess the effect of the missing observation or exclusion on the comparison. Changing the pump’s maintenance interval would require grounds for that change and the applicable authority.

RMP.4:5.3 - A computation with a research-shaping AI suggestion

A researcher uses an AI service to suggest a preprocessing change for a model comparison. In this constructed case, the retained account identifies the supplied design, original data and code version, the relevant proposal and input context, and the researcher’s decision to revise and use the change. The actual run connects the changed code, environment and parameters to its output.

The service also suggests a nonexistent article. The researcher checks the reference, rejects it and preserves the source actually used. This evidence can establish how the suggestion was handled. The scientific claim depends on what the source used supports.

When the original service version becomes unavailable, the retained suggestion, the researcher’s decision and the implemented change can still be inspected. A new run can test the retained implementation. Reproducing the original service response would need its own conditions and evidence.

RMP.4:5.4 - Historical and interpretive records support different criticisms

When asking whether a directive preceded a change in factory practice, preserve the identifiers, editions, dates and provenance of documents inspected, the relevant catalogue entries, contrary evidence and transcription choices. In this case the unavailable earlier series leaves the rival chronology open. An earlier decision could have occurred without leaving evidence in the surviving series.

To study how former workers understood the closure, retain the encounters, permitted excerpts, the researcher’s position and the grounds for changes to prompts or interpretation. A new prompt may still address the same question. Estimating population prevalence would require a different design. Judge the interpretation by the selected approach and its grounding, including a coding-agreement criterion only where it applies.

RMP.4:6 - Bias-Annotation

Identify missing cases, sources, encounters or failed runs where their absence could affect the question. For example, a record restricted to successful runs cannot establish how often runs failed unless it can be linked to an adequate record of all attempts.

Later knowledge can influence a witness’s account of an earlier reason. Preserve the grounds for the reconstruction, including what contemporaneous records and later reports establish about timing.

Data holders, communities and participants may have authority over access and use. Apply the relevant conditions and report what the authorized view allows its reader to examine, including any material limit on the conclusion.

RMP.4:7 - Conformance Checklist

Use this checklist for the package’s declared receiving examination.

  • The account identifies the inquiry, the portion covered and the examination. Claims about completed work have grounds in the surviving material.
  • Plans, actual conduct, later reports, reconstructions and unknowns remain distinguishable.
  • Consequential material can be followed through selection and transformations to its receiving use, at the granularity that examination requires.
  • For a consequential change, the account identifies the earlier intention, timing, reason, effect and information available at the time to the extent supported.
  • The account distinguishes the consequences of missing observations, exclusions, failed runs, inaccessible sources and revised interpretations where they occur.
  • The account identifies significant contributions from people, teams, tools or services, the checks performed and the applicable responsibility, at the detail needed and permitted for the examination.
  • The account distinguishes surviving material, actual permitted access, public views, reconstruction and loss.
  • The material and permitted access support each examination described as criticism-ready. When an examination is prevented, the account identifies the limiting condition, any narrower supported use and a further source or change of design if one is needed and available.
  • A rerun or a later observation is identified by its own date, inputs and conditions.

When relying on the trace, follow a consequential use back to its source or use a qualified inspection that establishes the necessary connection. Compare a prior commitment with conduct when that difference matters. If the examination requires the intended reader to inspect the material, confirm that reader’s access.

RMP.4:8 - Common Anti-Patterns and How to Avoid Them

FailureWhy it failsRepair
The final methods account replaces every earlier plan.The critic loses the timing and content of commitments.Retain surviving versions and distinguish reconstruction from prospective intention.
Missing observations, exclusions and failed work are all recorded as “cleaning.”Their causes and effects on the result can differ.Identify the event, its supported reason and its effect on the examination.
A repository link is treated as criticism-ready access.The material may not survive, be accessible or contain the needed detail.Establish the actual material and access for the named examination.
A public summary is used for a claim that depends on redacted detail.Its reader lacks the detail needed to examine that claim.Identify what each view supports. Use an authorized source or a qualified inspection result if available, or report the limit.
The number of files is offered as support for a scientific claim.The decisive record might be absent from a large collection.Identify the connection the claim needs and examine the evidence that bears on it.
An AI suggestion or citation is used as evidence without establishing what supports it.A saved output can show what was suggested. The source and the inference need their own grounds.Identify the source, checks, decisions and responsible contributors. Where the applicable guidance assigns responsibility to a human researcher, record how the researcher exercised it.
A run with substituted inputs is reported as a replay of the original computation.It tests a different set of inputs.Identify the inputs, decisions and service state that can be recovered for the original run, then state what the new run tested.

RMP.4:9 - Consequences

A reader can see what was planned, what evidence supports the history of the work, how material reached a result and where examination stops. Capture during fieldwork can prevent loss. A later reconstruction can recover part of the history even when the original plan is missing.

Recording the relevant connections and maintaining suitable access take work. When decisive evidence is lost or inaccessible, state the conclusions that the remaining records support and the questions they leave open.

RMP.4:10 - Architectural Rationale

A protocol states intended conduct. A trace supplies grounds for claims about what happened, including departures from the plan and work for which no prospective plan survives. One notebook or collection can serve both uses.

Use the named examination to judge when capture is sufficient. The same material can support one criticism and lack the detail for another. Additional capture and disclosure require justification for the use they would serve.

Use existing domain records or provenance information if they supply the connections the question needs. RMP.4 helps recover how choices, transformations and access affect examination of the particular inquiry.

RMP.4:11 - SoTA-Echoing

Practice question. What does a reader need to inspect the conduct of this study, and which further claims need a replay or a scientific assessment?

Selected approach. Record the origin of the material, the timing of consequential changes and the access available to the intended reader at the detail required for the examination. When records are lost or restricted, establish which narrower questions can still be answered.

Comparison and source roleAdopt, adapt or reject; effect on the Method and limit
TOP 2025 distinguishes research practices and records of commitments.§§4.2 and 4.4 use contents and dates to establish what was specified in advance. A disclosure badge can identify a reported commitment. To verify it, use the relevant record or a qualified verification that covers the claim.
Miske et al.’s Investigating reproducibility (2026) examined attempts to reproduce results in publications from 2009–2018 with obtainable datasets. Workflow Run RO-Crate illustrates how to record particular workflow runs.§4.3 connects reported outputs to the inputs, transformations and runs that produced them. The study documents difficulties recovering computations. The format is one way to preserve the connections needed for inspection.
PRISMA-S gives reporting guidance for searches.§4.3 retains the search as performed, including its query, platform, date, retrieved material and changes during retrieval. Those records can explain how a bibliography was obtained. Judging its adequacy for synthesis requires the relevant appraisal.
Braun and Clarke’s RTARG argument and review examines how papers report reflexive thematic analysis. The AHA’s professional standards describe historians’ responsibilities for sources and interpretation.§§4.2–4.4 use the approach being examined to decide which choices, contrary evidence and grounds for reconstruction to retain. A requirement for coder agreement needs a basis in that approach.
Carroll et al.’s Operationalizing CARE and FAIR develops data-governance principles in the context of Indigenous peoples’ rights and interests.§4.5 asks whose authority governs a particular use, what the permitted purpose is, and what the reader can inspect. A FAIR resource can have restricted access.
The European Commission’s May 2026 revision and its Living guidelines on the responsible use of generative AI in research, fourth edition, October 2026 address researchers’ responsibility for AI-assisted scientific output and disclosure of substantial uses.§4.4 applies that guidance to records of significant AI contributions and the responsible researcher’s verification and decisions. Routine editing calls for proportionate records. These recommendations are non-binding. Access to protected inputs and the ability to replay a service require separate grounds.

Reconsider what to retain if a different examination needs different material, a source or service change affects a connection, or an existing practice can support the examination at lower cost. The cited guidance, research and implementation examples establish the specific contributions described above.

RMP.4:12 - Relations

An issued instruction can establish what it prescribed. To reconstruct how work was performed, use the sources that bear on that question, such as an occurrence record, a participant’s account or an event log. Use RMP.10 when further reconstruction is needed. Dated sources may establish that a method changed. Use RMP.11 when the question asks why or how that change came about. In each case, account for relevant dependence and gaps in what survived.

  • RMP.3 supplies operational intentions and research-relative record needs. Capture can begin while those intentions develop.

  • A lost comparison can be reported with the remaining findings. Use RMP.2 if a revised design is needed. Use RMP.1 to reconsider a question or intended use that has changed.

  • Use a clear, qualified record directly. FPF A.10 helps when the connection from a claim to its evidence needs to be recovered or made reproducible. FPF C.16 helps establish the measurement basis when a distinction between an indication, a qualified result and a derived value matters.

  • Use the relevant professional practice for recordkeeping, source criticism, statistics or interpretation. Apply the existing decisions and conditions governing retention and access. When a new decision is required, obtain it from the person or body with the applicable authority.

  • Use an available analysis or credibility assessment within its scope. RMP.5 helps analyze and criticize a claim when a substantive judgement is still needed. RMP.6 helps select and interpret a credibility examination when the question warrants one. A report that a reproduction test took place needs evidence of the test performed and its result.

  • Use RMP.7 when the material, its relationships and its limits bear on a question about a corpus. A synthesis requires the appraisal and integration appropriate to that question.

RMP.4:End

Referenced in the corpus

50 literal mentions in other sections. Read their context to establish the relation.