CASRAI Dictionary

Tag: data management plan research

  • Data Management Plan Templates by Discipline

    A data management plan template sets out what data a project will produce, how it will be documented, stored, protected and shared, and who is responsible for each step. There is no single universal template: a physical-science plan built around instrument calibration and file formats looks very different from a life-science plan built around clinical consent and genomic repositories, or a social-science plan built around participant anonymisation and survey metadata. A data management plan (DMP) is a formal document, usually required at grant application or award stage, that describes how research data will be handled across the full research lifecycle, from collection through to long-term preservation or disposal.

    What every data management plan template must include

    Regardless of discipline, funders expect the same core sections. The Science Europe RDM Guide structures its template around 15 questions covering six core requirements: data description, documentation and metadata, storage and backup, legal and ethical requirements, data sharing, and responsibilities and resources. The UK’s Digital Curation Centre (DCC) publishes a parallel checklist used by most UK universities as the basis of their local DMP guidance.

    • Data description — types, formats, volumes and provenance of data to be generated or reused
    • Documentation and metadata standards that will make the data intelligible to others
    • Storage, backup and security arrangements during the active project
    • Ethical, legal and consent considerations, including any restrictions on sharing
    • Preservation, repository choice and long-term access arrangements
    • Roles, responsibilities and resourcing for data management tasks

    These core sections are then adapted to the realities of the data itself. That adaptation — not the boilerplate headings — is where discipline-specific templates diverge, and where a generic one-size-fits-all template becomes a liability rather than a help.

    Physical sciences: instrument and sensor data

    Physical-science DMPs — physics, chemistry, astronomy, earth and environmental sciences — are dominated by high-volume instrument, sensor and simulation output rather than by human-subjects concerns. The template needs to say more about formats, calibration and reduction, and comparatively little about consent.

    • Community-accepted file formats (for example FITS in astronomy, NetCDF in climate and earth science) to guarantee interoperability
    • Instrument and calibration metadata, so raw readings remain interpretable and reproducible years later
    • Data volume and velocity planning — strategies for reduction, transfer and storage of large or continuous streams from sensors, telescopes or particle detectors
    • Software and code versioning, since simulation and analysis code is often as essential to reproducibility as the raw data itself
    • Discipline repositories such as PANGAEA (earth and environmental data) or domain-specific archives maintained by observatories and facilities

    The European Research Council’s DMP template for Horizon Europe-funded projects follows the same emphasis: it requires beneficiaries to describe data volumes, formats and FAIR compliance rather than consent procedures, reflecting the instrument-heavy profile of much ERC-funded physical-science work.

    Life sciences: biological and clinical data

    Life-science DMPs — biology, medicine, genomics, clinical research — carry a heavier ethical and regulatory load. A strong template treats consent, de-identification and repository choice as first-class sections, not footnotes.

    • Ethical and legal compliance: how human-subject or animal data will be de-identified, and how the plan aligns with UK GDPR and relevant research ethics committee approvals
    • Community ontologies and minimum-information standards such as MIAME for microarray data or Darwin Core for biodiversity records, which allow datasets to be compared across studies
    • Persistent identifiers for samples, datasets and participants, supporting findability without compromising anonymity
    • Genomic and clinical data deposition requirements — for example, NIH policy requires eligible genomic datasets to be deposited in controlled-access repositories such as dbGaP or GEO
    • The UK’s Medical Research Council (MRC) requires a personalised DMP, using a published UKRI template, for MRC-funded studentships and non-doctoral training grants

    Because clinical and genomic data are rarely fully open, life-science templates typically distinguish between what will be shared openly, what will be shared under a data access agreement, and what cannot be shared at all — a three-tier distinction that is largely absent from physical-science templates.

    Social sciences: qualitative and survey data

    Social-science DMPs — sociology, psychology, education, economics — centre on informed consent, anonymisation and the management of qualitative material such as interview transcripts and survey responses.

    • Informed consent procedures and how participants are told what will happen to their data, including any future-use or secondary-use provisions
    • Anonymisation and de-identification plans for both qualitative data (transcripts, recordings, field notes) and quantitative survey data
    • Metadata standards for survey and social data, such as the Data Documentation Initiative (DDI), used by archives to describe questionnaires and variables
    • Data access agreements for restricted or sensitive datasets, specifying who can apply for access and under what conditions
    • Long-term archiving through a recognised social-science data service — in the UK, the UK Data Service (based at the UK Data Archive, University of Essex) is the standard repository for shareable social and economic data

    Qualitative data management also needs its own sub-plan: transcription protocols, version control across iterative coding, and a named point at which raw recordings are destroyed or securely archived under the original consent terms.

    Dimension Physical sciences Life sciences Social sciences
    Dominant data type Instrument, sensor, simulation output Genomic, clinical, biological samples Interview, survey, observational records
    Key standards FITS, NetCDF MIAME, Darwin Core DDI (Data Documentation Initiative)
    Primary risk to manage Volume, format interoperability Participant privacy, consent, regulation Anonymisation, re-identification risk
    Typical repository PANGAEA, facility archives dbGaP, GEO, ENA UK Data Service / UK Data Archive

    Frequently asked questions

    How do you write a data management plan?

    Start from your funder’s own template where one exists — UKRI, the ERC and most UK universities publish one — then work through data description, storage, ethics, sharing and preservation in turn. Discipline-specific detail, such as file formats or consent procedures, should be added within each section, not bolted on afterwards.

    What is included in a data management plan?

    A complete DMP describes the types of data that will be produced, the metadata and documentation standards used, storage and backup arrangements, ethical and legal requirements, sharing and access conditions, and preservation plans, with named roles for each task.

    Do all researchers need a data management plan?

    Most UK and EU research funders — including UKRI, Horizon Europe and members of cOAlition S — now require a DMP as a condition of funding. Even without a mandate, a DMP reduces the risk of data loss and supports compliance with institutional data protection policy.

    What does a good data management plan look like?

    A good DMP is specific rather than generic: it names the exact repository, format and access conditions that apply to the project’s actual data, not boilerplate language borrowed from an unrelated template. It is also a living document, revisited as the project changes.

    What this means for research administrators

    Research offices that hand every applicant the same generic DMP template are setting up avoidable review delays: reviewers increasingly expect discipline-appropriate detail on formats, consent or anonymisation rather than restated boilerplate. Building three lightweight discipline variants — physical, life and social science — from one core checklist, as outlined above, lets an institution keep a single governance structure while giving each researcher a template that actually matches their data.

    The underlying reference point across all three variants is the FAIR data principles — Findable, Accessible, Interoperable, Reusable — first formalised by Wilkinson et al. in Scientific Data (2016) and now embedded in funder policy from Horizon Europe to UKRI. Fair data management is the common thread; the template detail is where disciplines genuinely diverge. Institutions building or revising DMP guidance should treat the research administration function, not just the library, as the natural owner of discipline-specific template maintenance, and consult the CASRAI Dictionary for consistent definitions of the terms used across templates.

  • NSF Data Management Plan: A Directorate Guide

    An NSF data management plan (DMP) is a required proposal component describing how a project will handle, share, and preserve research data — and since 27 April 2026, its exact content is set by a structured Research.gov webform that adapts to the proposal’s lead directorate, meaning BIO, ENG, GEO, MPS, OPP, SBE, and EDU proposals no longer follow one identical template. Treating NSF as a single monolithic funder — the default approach in most DMP guides — now produces plans that miss directorate-specific expectations.

    A data management and sharing plan (DMSP) is the National Science Foundation’s formal proposal document setting out how a funded project will manage, share, and archive the data, samples, and other research products it produces. Under the Proposal & Award Policies and Procedures Guide (PAPPG) section II.D.2(ii) and Policy Notice NSF 26-202, every full proposal must include one — or a documented justification if the project will not generate data.

    What Does an NSF Data Management Plan Require in 2026?

    Every NSF proposal must address six general elements under PAPPG II.D.2(ii), regardless of directorate. These form the baseline that directorate-specific guidance then narrows or extends.

    • The types of data, samples, physical collections, software, and curriculum materials the project will produce
    • The standards used for data and metadata format and content, with a documented workaround where no standard exists
    • Policies for data access and sharing, including privacy, confidentiality, security, and intellectual-property protections
    • Policies for data reuse, redistribution, and the production of derivative products
    • Plans for archiving data and other research products and preserving long-term access to them

    Under Policy Notice NSF 26-202, a proposal that will not generate data does not need a full plan — a short justification statement satisfies the requirement instead.

    How Do NSF DMP Requirements Differ by Directorate?

    NSF publishes supplementary DMP guidance for seven directorates and offices — BIO, ENG, GEO, MPS, OPP, SBE, and EDU — plus at least one program-specific supplement (DMREF, under MPS). Where a directorate has issued no supplement, the general PAPPG rules apply by default. The table below summarises what each adds on top of the baseline.

    Directorate/Office Distinguishing emphasis
    Biological Sciences (BIO) Deposit in community-recognized public repositories (e.g. GenBank-class databases) with persistent identifiers linking data to publications
    Engineering (ENG) Broad coverage of software, models, and physical collections; attention to intellectual property where research has commercial potential
    Geosciences (GEO) Discipline-specific repository requirements — Ocean Sciences awardees, for example, are directed to the Biological and Chemical Oceanography Data Management Office (BCO-DMO)
    Mathematical & Physical Sciences (MPS) General guidance, plus a dedicated program-level supplement for the Designing Materials to Revolutionize and Engineer our Future (DMREF) programme
    Office of Polar Programs (OPP) Governed by a separate Dear Colleague Letter establishing a distinct data and code/sample management policy rather than the standard DMSP framework alone
    Social, Behavioral & Economic Sciences (SBE) Heavy emphasis on human-subjects protections — anonymisation, handling of personally identifiable information, informed consent for data sharing, and deposit in recognised social-science archives
    STEM Education (EDU) Maintains its own directorate-level data-management-plans page addressing education-research data and human-subjects considerations

    This directorate layering is why a plan drafted for a BIO proposal will read very differently from one drafted for an SBE or OPP proposal, even though both start from the same PAPPG baseline.

    What Changed With the April 2026 Research.gov Webform?

    Effective 27 April 2026, NSF replaced the free-standing two-page PDF data management and sharing plan with a structured webform submitted directly through Research.gov. This is the single most consequential change to NSF DMP practice since the policy’s creation, and it makes most “download the template” search results obsolete.

    • The plan is now entered field-by-field in Research.gov rather than uploaded as a standalone PDF attachment
    • The webform adapts its prompts to the proposal’s selected lead directorate, formalising the differences long implied but not enforced by the old PDF format
    • Investigators should verify current guidance for their directorate before assuming a saved PDF template from a prior submission still matches the required fields

    Institutional research offices that maintain locally cached “NSF DMP template” documents should retire the PDF version and point investigators to the live Research.gov webform for the current submission year.

    NSF Data Management Plan Checklist, by Directorate

    Use this sequence to build a directorate-appropriate plan rather than a generic one:

    • Confirm the proposal’s lead directorate and pull its specific guidance page (BIO, ENG, GEO, MPS, OPP, SBE, or EDU) alongside the general PAPPG II.D.2(ii) requirements
    • List every data type, sample, and software product the project will generate
    • Identify the metadata standard and, for BIO or GEO proposals, the target public repository (e.g. BCO-DMO for ocean sciences)
    • For SBE or human-subjects work, document anonymisation, consent, and PII-handling procedures explicitly
    • For OPP proposals, check whether the relevant Dear Colleague Letter policy supersedes the standard DMSP structure
    • Set an archiving and long-term preservation plan with a named repository or institutional data service
    • Submit through the Research.gov webform rather than attaching a standalone PDF

    Common Questions About NSF Data Management Plans

    Does NSF require a data management plan?

    Yes. NSF requires a data management and sharing plan as a mandatory component of every full proposal, per PAPPG II.D.2(ii). Proposals that will not produce data must instead include a written justification explaining why no plan is needed.

    What is included in a data management plan?

    A complete plan covers the types of data produced, the metadata standards applied, access and sharing policies, provisions for reuse and derivatives, and an archiving and preservation plan for long-term accessibility.

    Do I need a data management plan?

    Any NSF full proposal needs one unless the project genuinely generates no data, samples, or research products — in which case a short justification statement, not a full plan, satisfies the requirement under Policy Notice NSF 26-202.

    What This Means for Research Administrators

    Directorate-tailored webforms shift the compliance burden earlier in the proposal cycle. Research offices that previously offered a single boilerplate DMP template now need directorate-aware review checkpoints, because a plan that satisfies BIO’s repository expectations will not automatically satisfy SBE’s human-subjects requirements or OPP’s separate policy letter. Institutions supporting multi-directorate portfolios should update internal guidance documents to reference the correct directorate page rather than a single generic NSF DMP resource.

    The Outlook for NSF Data Management Requirements

    The move to a structured, directorate-tailored webform signals that NSF intends to enforce, rather than merely suggest, discipline-specific data practices. Investigators and research offices that continue treating the NSF data management plan as a single generic two-pager risk submitting plans that technically comply with PAPPG but miss the sharper, directorate-specific expectations now built into the submission system itself.