Decide what will be preserved, protected, documented and shared before the first research file is created
A research data management and sharing plan turns broad commitments into operational decisions. It identifies what data will exist, how files will be documented and protected, what can be shared, where, when and under whose oversight. English matters because words such as data, metadata, de-identified, restricted and retained have technical, ethical and sometimes legal consequences.
Choose the route closest to your task, or read straight through for the complete system.
People search for data management plan examples, research data sharing plans and DMP templates because funders increasingly ask teams to plan before data collection begins. A useful plan is not a promise to make everything public. It is a controlled explanation of what will be created, documented, secured, preserved, shared or restricted.
The writing must keep several authorities visible at once: the live funder policy, research protocol, ethics approval, consent language, data-protection obligations, agreements with partners and repository rules. A sentence that satisfies one cannot silently override another.
Good plans are specific enough to guide real work. They name file types, metadata, storage environments, access roles, repositories, timing, licences or access controls, cost ownership and review points. They also say what is unknown and when it will be resolved.
This article is educational guidance checked on 11 October 2026. It does not replace legal, ethics, information-security, records-management, indigenous-data-governance or institutional advice.
Did You Know? The US National Institutes of Health guidance on writing a Data Management and Sharing Plan says its 2026 DMS Plan format is now required for competing and non-competing awards. The current format asks about maximum appropriate sharing, timing, retention, justified limitations, protection of participants, expected data types and repositories, plus genomic-data questions when applicable. That is NIH-specific guidance; another funder, institution, ethics board or repository may require different fields.
Find the section you need
Define the data · 2 chapters
Document the record · 2 chapters
Protect people · 2 chapters
Plan access · 2 chapters
Assign oversight · 3 chapters
A generic DMP template may omit a funder’s required elements.
Record policy version, required format, word or page limits, questions, submission stage, update duties and assessment context.
The NIH 2026 question-and-table format is not assumed to fit a different national funder.
What to check
- Check policy
- Check template
- Check limit
- Check update rule
Create a control note for “Read the controlling plan requirement”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: A sample plan is evidence of form, not permission to copy commitments.
Learning transfer: Instruction matching supports applications.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
Teams may use data to mean raw files, analysed tables, code, images or everything in the project folder.
List data types, formats, estimated scale, collection or generation method and which outputs underpin findings.
Interview audio, transcripts and coded extracts are separated because their risks and sharing routes differ.
What to check
- Name data type
- Name format
- Estimate scale
- Map finding
Create a control note for “Define the scientific data in scope”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Use the policy’s applicable definition rather than a private shortcut.
Learning transfer: Classification strengthens science.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
3. Describe tools, software and code dependencies
Back to contentsShared data can become unusable when the reader lacks the tool or environment.
Name specialised software, code, workflows, schemas and access route, including open alternatives or documentation where feasible.
A proprietary instrument file is accompanied by an export format and conversion note.
What to check
- List dependency
- Check licence
- Plan export
- Document version
Create a control note for “Describe tools, software and code dependencies”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Do not promise redistribution of software the team does not own.
Learning transfer: Dependency thinking supports computing.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
A filename rarely explains variables, units, provenance or missing values.
Specify readme files, data dictionaries, codebooks, metadata standards, naming conventions and provenance records.
A survey dataset includes question wording, value labels, units, transformations and suppression rules.
What to check
- Name standard
- Define variables
- Record provenance
- Explain missingness
Create a control note for “Choose documentation and metadata standards”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: A standard name alone does not prove complete metadata.
Learning transfer: Precise definitions improve vocabulary.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
Preservation at the end cannot repair insecure or lost working data.
Describe approved storage, encryption where required, backup, disaster recovery, role-based access, transfer and device restrictions.
Identifiable files remain in an approved restricted environment while de-identified extracts are used for analysis.
What to check
- Name environment
- Name access roles
- Check backup
- Control transfer
Create a control note for “Control storage, backup and working access”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Consumer sync tools may not meet institutional requirements.
Learning transfer: Security language supports digital life.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
De-identification does not automatically remove every disclosure or contractual risk.
Compare sharing language with participant consent, ethics approval, law, community expectations, partner agreements and data-use terms.
A small qualitative dataset uses controlled access because indirect identification remains plausible.
What to check
- Read consent
- Read ethics
- Check law
- Check agreements
Create a control note for “Align consent, privacy, ethics and agreements”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Do not claim anonymity when re-identification risk remains.
Learning transfer: Ethical reading supports citizenship.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
A repository should fit data type, access model, identifiers, metadata, preservation and policy requirements.
Compare discipline, institutional and general repositories; record accepted formats, costs, embargoes and controlled-access capability.
Imaging data goes to a disciplinary repository that supports its standard and access restrictions.
What to check
- Compare repositories
- Check formats
- Check identifiers
- Check costs
Create a control note for “Select preservation files and repository”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: A popular repository is not automatically appropriate for sensitive data.
Learning transfer: Platform comparison supports careers.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
“After publication” is ambiguous when there are several papers or no accepted paper.
State deposit and access triggers, embargo basis, minimum retention, review dates and destruction or transfer authority.
Data supporting each paper are shared at publication while a restricted source dataset follows an approved access process.
What to check
- Name trigger
- Name date basis
- Check retention
- Name authority
Create a control note for “Set timing, retention and access conditions”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Never promise indefinite retention without resources and authority.
Learning transfer: Timeline precision supports planning.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
Responsible sharing can include justified limits rather than open download.
State the reason for limitation, eligible requesters, application evidence, review body, agreement, monitoring and appeal or review route.
Researchers request access through a committee and sign a data-use agreement.
What to check
- Name restriction
- Name decision maker
- Define criteria
- Set review
Create a control note for “Explain restrictions and controlled access”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Avoid vague “available on request” language without a workable process.
Learning transfer: Procedural writing supports fairness.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
Curation, de-identification, documentation, storage and repository deposits take time and money.
Map each activity to staff effort, service, timing, cost basis and funding source; reconcile with the budget justification.
A data steward’s effort is scheduled before deposit rather than added after the award ends.
What to check
- Map task
- Map owner
- Map cost
- Reconcile budget
Create a control note for “Budget and resource the plan”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Check which costs are allowable and when they may be charged.
Learning transfer: Resource planning supports projects.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
A DMS plan may need to change as methods, repositories or risks evolve.
Name the accountable role, review schedule, compliance evidence, deviation process and version log.
A repository closure triggers an approved revision rather than an undocumented substitution.
What to check
- Name steward
- Set review
- Log change
- Notify authority
Create a control note for “Assign oversight and revise responsibly”. Keep the live instruction, verified fact, interpretation, owner and unresolved question in separate rows.
Explain the decision to a reader outside the team. Preserve the source, date, scope, evidence, uncertainty and next action; if compression changes a credential, obligation, milestone or award term, restore the missing condition.
Boundary: Do not treat approval of a plan as permanent approval for every later change.
Learning transfer: Governance language supports teamwork.
Write one sentence that states the evidence, one that states its limit, and one that states the next responsible action. Then test each noun, number, condition and source against the controlled document. Replace vague confidence with a traceable reason.
How to build this kind of English before the job title arrives
Back to contentsStudents do not need professional authority to practise the underlying language. They can learn to define scope, annotate conditions, compare versions, rebuild a timeline, match a number to its unit and explain uncertainty without embarrassment. Parents can ask calm questions: What decision is this document supporting? Which sentence carries the strongest claim? What would make that claim false? Which source would you open next?
Use a three-pass routine. First, map the document: title, owner, audience, date, sections and decision. Second, trace one claim from source to conclusion. Third, explain the limit and next action. This routine strengthens comprehension, science, mathematics, humanities and workplace readiness because it turns English into an evidence tool.
A student should never imitate professional authority. The useful goal is to recognise when a text needs a qualified reviewer, current rule or official decision. Knowing when to pause is part of literacy.
Data
Audio, transcripts, coded extracts and supporting documentation.
Protection
Approved storage, restricted roles, consent and disclosure review.
Preservation
Selected transcript derivatives, codebook and provenance record.
Access
Controlled repository, application criteria, agreement and oversight.
Policy
Funder, version, template and update rule.
Data
Type, format, scale, source and value.
Documentation
Metadata, codebook, naming and provenance.
Working security
Storage, backup, roles and transfer.
Authority
Consent, ethics, law and agreements.
Preservation
Files, repository, identifier and retention.
Access
Timing, licence, embargo or controlled route.
Oversight
Owner, cost, review and version.
Each case turns a fluent shortcut into an evidence problem. Explain what changed, whose decision is affected and why the repaired version is safer.
Case 1: Read the controlling plan requirement
The NIH 2026 question-and-table format is not assumed to fit a different national funder. The learner identifies the tempting shortcut, checks the controlling evidence, explains the consequence and records the repair in complete sentences.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Case 2: Choose documentation and metadata standards
A survey dataset includes question wording, value labels, units, transformations and suppression rules. The learner identifies the tempting shortcut, checks the controlling evidence, explains the consequence and records the repair in complete sentences.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Case 3: Set timing, retention and access conditions
Data supporting each paper are shared at publication while a restricted source dataset follows an approved access process. The learner identifies the tempting shortcut, checks the controlling evidence, explains the consequence and records the repair in complete sentences.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Case 4: Assign oversight and revise responsibly
A repository closure triggers an approved revision rather than an undocumented substitution. The learner identifies the tempting shortcut, checks the controlling evidence, explains the consequence and records the repair in complete sentences.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
A parent, tutor, supervisor or peer may ask for the source and logic, but the learner should make the final evidence note in their own words.
Use the live rule and controlled project evidence. Preserve source wording, interpretation, decision, authority and version separately.
Define the data
Identify data types, formats, scale and research value. Use Chapters 1–2 to build a dated evidence note and decision.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Document the record
Choose metadata, naming, versions and quality controls. Use Chapters 3–4 to build a dated evidence note and decision.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Protect people
Align security, consent, law and agreements. Use Chapters 5–6 to build a dated evidence note and decision.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Plan access
Choose repository, timing, licence or controlled route. Use Chapters 7–8 to build a dated evidence note and decision.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Assign oversight
Name costs, owners, reviews and change control. Use Chapters 9–10 to build a dated evidence note and decision.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Finish with three sentences: what is supported, what remains uncertain and what evidence or authorised action comes next.
The evidence should remain stable even when emphasis, detail and next action change with the reader.
Researcher lens
Files and documentation must support valid work.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Participant lens
Consent, dignity and disclosure risk must remain central.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Repository lens
Formats, metadata and access conditions must be workable.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Funder lens
Policy elements and oversight need a credible plan.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Future user lens
Access without context is not meaningful reuse.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Adaptation helps only when the short version preserves identity, source, status, evidence, uncertainty and boundary.
These exercises are deliberately slower than real drafting. Rebuild the decision chain, then compress it without losing the condition that controls the outcome.
Define the scientific data in scope
Teams may use data to mean raw files, analysed tables, code, images or everything in the project folder. List data types, formats, estimated scale, collection or generation method and which outputs underpin findings. Example: Interview audio, transcripts and coded extracts are separated because their risks and sharing routes differ. Before accepting the conclusion, test the named source, scope, date, authority and exception against the controlled record.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Describe tools, software and code dependencies
Shared data can become unusable when the reader lacks the tool or environment. Name specialised software, code, workflows, schemas and access route, including open alternatives or documentation where feasible. Example: A proprietary instrument file is accompanied by an export format and conversion note. Before accepting the conclusion, test the named source, scope, date, authority and exception against the controlled record.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Control storage, backup and working access
Preservation at the end cannot repair insecure or lost working data. Describe approved storage, encryption where required, backup, disaster recovery, role-based access, transfer and device restrictions. Example: Identifiable files remain in an approved restricted environment while de-identified extracts are used for analysis. Before accepting the conclusion, test the named source, scope, date, authority and exception against the controlled record.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Select preservation files and repository
A repository should fit data type, access model, identifiers, metadata, preservation and policy requirements. Compare discipline, institutional and general repositories; record accepted formats, costs, embargoes and controlled-access capability. Example: Imaging data goes to a disciplinary repository that supports its standard and access restrictions. Before accepting the conclusion, test the named source, scope, date, authority and exception against the controlled record.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Budget and resource the plan
Curation, de-identification, documentation, storage and repository deposits take time and money. Map each activity to staff effort, service, timing, cost basis and funding source; reconcile with the budget justification. Example: A data steward’s effort is scheduled before deposit rather than added after the award ends. Before accepting the conclusion, test the named source, scope, date, authority and exception against the controlled record.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Repeat the drill after a source, person, cost, permission, version or timetable changes. The revised note should show exactly why the earlier conclusion was retained or replaced.
A polished line can conceal a missing rule, identifier, comparison, version, permission or condition. Run these tests before relying on fluency.
Data test
List each data type, format, scale and finding it may support; remove the word data where it hides a difference.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Authority test
Compare sharing with consent, ethics, law, agreements, security rules and repository conditions.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Reuse test
Give the record to a qualified future user and identify the documentation, tool or code they still need.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Restriction test
For each access limit, state the reason, decision maker, criteria, agreement and review route.
Record the governing source, version, people affected, evidence and uncertainty. Then write the smallest conclusion that remains accurate if one rule, number, role, date or permission changes.
Keep the failed version beside the repair. The contrast shows exactly where language outran the record.
Answer each prompt in a complete sentence tied to a current instruction, dated source, observation or authorised clarification. Write “unknown” where the record does not support an answer.
- Check policy? Record the evidence and the conclusion it permits.
- Check template? Record the evidence and the conclusion it permits.
- Check limit? Record the evidence and the conclusion it permits.
- Check update rule? Record the evidence and the conclusion it permits.
- Name data type? Record the evidence and the conclusion it permits.
- Name format? Record the evidence and the conclusion it permits.
- Estimate scale? Record the evidence and the conclusion it permits.
- Map finding? Record the evidence and the conclusion it permits.
- List dependency? Record the evidence and the conclusion it permits.
- Check licence? Record the evidence and the conclusion it permits.
- Plan export? Record the evidence and the conclusion it permits.
- Document version? Record the evidence and the conclusion it permits.
Independent-reader test
Give the working note to a careful reader who was not present. Ask them to identify the source, version, reasoning, authority, uncertainty and next action without extra explanation. Repair a missing link rather than adding decorative formality.
Change test
Imagine that the event brief, data risk, timeline assumption or award term changes tomorrow. Mark every sentence that would need revision. This makes cycle-sensitive claims visible and prevents an old note from controlling a new decision.
Decision log
Keep the instruction consulted, date, choice, responsible person and reason. Add rejected alternatives when they may be revisited. A collaborator or future student can then distinguish a documented requirement from a stylistic preference.
Parent or mentor conversation
Ask three calm questions: Which words carry the strongest claim? What evidence controls those words? What condition would make you revise the conclusion? The goal is accountable curiosity, not supplying a ready-made answer.
What is a data management and sharing plan?
It is a project-specific account of how research data will be documented, protected, preserved, shared or restricted and overseen.
Must all research data be public?
No. The applicable policy and ethical, legal, contractual and security conditions determine appropriate access.
Is de-identified data always safe to share openly?
No. Re-identification and contextual risks still require assessment.
Can I choose a repository later?
If the application permits it, record selection criteria and a decision date; a named repository is better when it can be verified.
Is a DMS plan the same as a data availability statement?
No. A plan guides the research lifecycle, while a data availability statement usually reports access arrangements alongside a publication.
Who owns the plan?
Name accountable project and institutional roles; oversight is a continuing responsibility, not a one-time writing task.
- NIH — Writing a Data Management and Sharing Plan
- NIH — Data Sharing Home
- Why English? | Writing a Data Availability Statement
- Why English? | Writing a Research Budget Justification
Cycle-sensitive facts were checked against current official pages on 11 October 2026. Always reopen the authority’s live page before a real decision.