Comprehensive Data Management Planning: Best Practices, Templates, and Strategies for Effective Research Data Stewardship

2025-09-13 Research Reports 1

Abstract

Data Management Plans (DMPs) represent indispensable instruments in contemporary research, meticulously articulating the strategies for data handling across the entire research lifecycle. Their fundamental role lies in ensuring that research data is collected, processed, stored, preserved, and disseminated in a manner that rigorously upholds principles of transparency, reproducibility, and long-term accessibility. This comprehensive report embarks on an exhaustive examination of Data Management Plans, delving into their profound significance, intricate components, the foundational best practices for their methodical creation and sustained maintenance, and offers nuanced, actionable guidance tailored for researchers spanning a diverse array of scientific and scholarly disciplines.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

1. Introduction

In the rapidly evolving and increasingly data-intensive landscape of modern research, the judicious management of data has ascended to become an indispensable cornerstone of scientific integrity, methodological rigor, and sustained progress. A Data Management Plan (DMP) serves not merely as a bureaucratic requirement but as a dynamic, strategic blueprint, meticulously detailing the methodologies and governance frameworks for every stage of the data lifecycle: from its initial collection and robust storage to its ethical sharing and enduring preservation. The contemporary research paradigm, characterized by a burgeoning emphasis on open science principles, data-driven discovery, and the imperative for verifiable reproducibility, has demonstrably amplified the criticality and widespread adoption of DMPs. Consequently, a growing consortium of prominent funding agencies, national and international regulatory bodies, and academic institutions worldwide have transitioned from recommending to mandating the inclusion of comprehensive DMPs as a prerequisite for research proposals and project initiation (National Science Foundation, n.d.; NIH, 2023). This report is specifically designed to provide an expansive and in-depth exploration of Data Management Plans, furnishing researchers across the academic spectrum with profound insights, practical frameworks, and comprehensive guidance essential for their effective development, rigorous implementation, and continuous stewardship throughout the lifespan of any research endeavour. By understanding and embracing the principles outlined herein, researchers can not only meet compliance requirements but also significantly enhance the impact, reliability, and enduring value of their scholarly contributions.

The exponential growth in data volume, velocity, and variety – often termed ‘Big Data’ – coupled with advanced computational capabilities, has transformed nearly every scientific discipline. This transformation underscores the necessity for structured data management, moving beyond ad-hoc approaches. The advent of the internet and digital storage has facilitated data sharing but simultaneously introduced complex challenges related to data security, privacy, and intellectual property. Furthermore, the global movement towards Open Science advocates for research outputs, including data, to be ‘as open as possible, as closed as necessary’ (FOSTER, n.d.). This philosophy is intrinsically linked to the FAIR principles – Findable, Accessible, Interoperable, and Reusable – which provide a guiding framework for scientific data management and stewardship (Wilkinson et al., 2016). DMPs act as the practical mechanism through which these lofty principles are translated into concrete actions, ensuring that research data is not only preserved but actively poised for future discovery, validation, and innovation.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

2. The Evolution and Importance of Data Management Plans

2.1 Historical Context

The genesis of Data Management Plans, while seemingly a modern phenomenon, can be traced back much further than commonly perceived, primarily emerging from fields demanding high levels of precision, accountability, and long-term data utility. Early iterations of systematic data handling protocols gained prominence in the 1960s, notably within the highly structured environments of aeronautical and engineering projects, where meticulous record-keeping for complex systems was paramount for safety, performance analysis, and iterative design improvements. Similarly, in the social sciences, pioneers recognised the enduring value of survey data and established early data archives. For instance, the Inter-university Consortium for Political and Social Research (ICPSR), founded in 1962, became a seminal institution for archiving and disseminating social science data, demonstrating an early commitment to data preservation and reuse (ICPSR, n.d.).

Over subsequent decades, the application and sophistication of data management practices progressively expanded across a broader spectrum of scientific disciplines. This expansion was driven by a confluence of factors: the increasing volume and complexity of scientific data, the growing necessity for interdisciplinary collaboration, and critically, the advent and rapid proliferation of digital computing technologies. The transition from physical data storage to digital formats, while offering unprecedented efficiencies, also presented new challenges regarding data integrity, format obsolescence, and long-term accessibility. The early 2000s marked a pivotal period, witnessing a significant acceleration in what became known as ‘e-research’ or ‘e-science’ – research paradigms heavily reliant on digital infrastructure, vast datasets, and computational analysis. Concurrently, policy-driven initiatives began to emerge, often spurred by public investment in research and a growing demand for transparency and accountability in scientific endeavours. Major funding bodies and governmental agencies started to recognise that robust data management was not merely good practice but a fundamental requirement for maximising the return on research investments, promoting reproducibility, and ensuring the enduring legacy of scientific output. This era saw the gradual embedding of DMPs into the fabric of research funding applications and project governance, evolving from ad-hoc recommendations into formal, often mandatory, components.

Key policy milestones include the UK’s E-Science Programme (2001-2006) which championed digital research infrastructure, and later, the European Commission’s Horizon 2020 framework (2014-2020) which strongly encouraged, and in some areas mandated, DMPs, eventually solidifying this into the Horizon Europe programme’s ‘Open Science Policy’ (European Commission, n.d.). In the United States, the National Science Foundation (NSF) implemented its data management plan requirement in 2011, making DMPs a mandatory component of grant proposals across all disciplines (National Science Foundation, n.d.). Subsequently, the National Institutes of Health (NIH) followed suit with its updated Data Management and Sharing Policy taking effect in January 2023, requiring all NIH-funded research generating scientific data to have a DMP (NIH, 2023). These policy shifts underscore a global consensus on the critical role of structured data management in fostering responsible and impactful research.

2.2 Significance in Contemporary Research

In today’s intricate and interconnected research environment, Data Management Plans are not simply compliance documents; they are pivotal strategic tools that underpin the foundational pillars of modern scientific inquiry. Their significance emanates from a multitude of benefits they confer upon individual researchers, research teams, funding bodies, and the broader scientific community:

  • Ensuring Data Quality and Integrity: A well-articulated DMP establishes clear, standardised procedures for data collection, validation, cleaning, and quality control. By proactively defining these methods, researchers can minimise errors, reduce inconsistencies, and ensure the accuracy and reliability of their datasets. This meticulous approach to data handling is fundamental for bolstering the credibility of research findings and enhancing the trustworthiness of the data itself. Protocols for data validation, error checking, and outlier detection are detailed, ensuring that the analytical insights derived are robust and defensible (Gonzaga University, n.d.). Furthermore, DMPs often include provisions for tracking data provenance – the lineage of data from its source through various transformations – which is crucial for verifying its authenticity and reliability.

  • Facilitating Data Sharing and Reuse (FAIR Principles): DMPs are the primary mechanism for operationalising the FAIR principles: Findable, Accessible, Interoperable, and Reusable (Wilkinson et al., 2016). They provide explicit guidelines on how data will be described (metadata), stored (repositories), and licensed (usage rights) to ensure it can be readily discovered, accessed, understood, and reused by others. This commitment to transparent data sharing not only promotes collaboration and accelerates scientific discovery by preventing redundant data collection but also allows for independent verification of research findings, thereby strengthening scientific rigor. Shared data can lead to new research questions, interdisciplinary insights, and novel discoveries that extend beyond the original scope of the project, significantly increasing the return on investment for research funding.

  • Compliance with Funding Agencies and Institutional Policies: As highlighted, many prominent funding bodies, including the NSF, NIH, UK Research and Innovation (UKRI), and the European Commission, have made DMPs a mandatory component of grant applications (National Science Foundation, n.d.; NIH, 2023; UKRI, 2022). Institutions also increasingly implement their own data policies. Adhering to these requirements is critical not only for securing funding but also for ensuring responsible data stewardship, avoiding potential sanctions, and maintaining a positive reputation within the scientific community. DMPs demonstrate accountability to taxpayers and stakeholders who fund research, assuring them that public funds are being used to generate enduring and accessible knowledge assets.

  • Mitigating Risks and Enhancing Data Security: By systematically addressing potential vulnerabilities, DMPs serve as a critical risk management tool. They compel researchers to plan for data security measures, including access controls, encryption, and regular backups, thereby significantly reducing the risk of data loss, corruption, or unauthorised access. For sensitive data, such as personally identifiable information (PII) or confidential business data, DMPs outline protocols for anonymisation, pseudonymisation, and secure storage environments, safeguarding privacy and preventing legal or ethical breaches. This proactive approach minimises the potential for reputational damage and legal repercussions that can arise from data mismanagement.

  • Optimizing Resource Allocation: Developing a DMP forces researchers to thoughtfully consider the resources required for effective data management from the outset of a project. This includes anticipating storage needs, budgeting for personnel involved in data handling and curation, identifying necessary software tools, and accounting for potential repository fees. This foresight allows for more efficient allocation of financial, human, and technological resources, preventing unforeseen costs or resource bottlenecks during the project’s execution. It shifts data management from an afterthought to an integrated and budgeted component of research.

  • Promoting Transparency and Reproducibility: In an era where scientific reproducibility is under increasing scrutiny, DMPs play a pivotal role. By explicitly detailing the ‘what, why, where, and how’ of data collection, processing, and analysis, they provide a transparent roadmap that allows other researchers to understand, replicate, and validate experimental methods and computational workflows. This level of transparency fosters trust in scientific findings and contributes directly to the self-correcting nature of science.

  • Enabling Long-Term Preservation and Legacy: Beyond the immediate project, DMPs ensure that valuable research data is preserved in accessible and usable formats for future generations of researchers, policymakers, and the public. This long-term preservation extends the utility and impact of the data, allowing it to contribute to longitudinal studies, new research questions, and interdisciplinary syntheses years or even decades after its initial creation. It transforms ephemeral research outputs into lasting scientific assets, securing the legacy of research investments.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

3. Core Components of a Data Management Plan

A comprehensive and effective Data Management Plan is a multifaceted document that systematically addresses all critical aspects of data stewardship throughout the research lifecycle. While specific requirements may vary slightly depending on the funding agency or institution, the following core components are universally recognised as fundamental for a robust DMP:

3.1 Types of Data

This foundational section of a DMP requires researchers to thoroughly characterise the nature of the data they intend to collect, generate, or acquire. A detailed understanding of data types is crucial as it informs decisions regarding storage, formats, security, and preservation strategies.

  • Data Categorisation: Data can be broadly categorised by its origin and state: raw data (unprocessed output from instruments or observations), processed data (cleaned, transformed, or aggregated raw data), derived data (results of analysis or modelling), experimental data (from controlled studies), observational data (field measurements, surveys), and simulation data (from computational models). Qualitative data (interviews, focus groups, textual analysis) presents different management challenges compared to quantitative data (numerical measurements, statistics). Specifying these categories helps in identifying appropriate tools and methods for each.

  • Data Formats: Detailing the expected file types is critical for ensuring data compatibility, longevity, and interoperability. Researchers should specify formats such as CSV (Comma Separated Values) or TSV (Tab Separated Values) for tabular data, JSON (JavaScript Object Notation) or XML (Extensible Markup Language) for structured data, HDF5 (Hierarchical Data Format) or NetCDF (Network Common Data Form) for complex scientific datasets, DICOM (Digital Imaging and Communications in Medicine) for medical images, TIFF (Tagged Image File Format) or JPEG for general images, and MP3 or WAV for audio. A strong preference should be given to open, non-proprietary, and widely supported formats to mitigate the risk of format obsolescence and ensure long-term accessibility. Proprietary formats often rely on specific software that may become unavailable or unsupported over time, hindering future reuse.

  • Data Volume and Velocity: Estimating the anticipated amount of data (e.g., in gigabytes, terabytes, petabytes) is essential for planning storage infrastructure, network bandwidth, and computational resources. Projects generating ‘big data’ may require specialised distributed storage systems and high-performance computing. Additionally, understanding the velocity of data generation (e.g., real-time sensor streams, daily experimental runs) impacts decisions on data ingestion pipelines, processing speed, and backup strategies. Underestimating data volume is a common pitfall that can lead to significant budgetary and logistical challenges.

  • Data Sources and Collection Methods: Identifying whether the data will be originally generated by the research project (e.g., through experiments, surveys, simulations), acquired from existing public datasets (e.g., government archives, open repositories), or sourced from third-party commercial providers is crucial. This distinction has profound implications for intellectual property rights, licensing agreements, ethical approvals, and compliance requirements. Describing the data collection methods (e.g., laboratory instruments, questionnaires, interviews, remote sensing, web scraping) provides context and helps to define data quality control procedures.

3.2 Standards and Metadata

Adhering to established data and metadata standards is paramount for enhancing data interoperability, discoverability, and long-term reusability, which are core tenets of the FAIR principles.

  • Data Standards: This involves utilising recognised formats, vocabularies, and protocols relevant to the specific research field. For example, in genomics, standards like the Minimum Information About a Microarray Experiment (MIAME) or the Guidelines for Minimum Information about a Proteomics Experiment (MIAPE) ensure consistent reporting. In environmental sciences, ISO 19115 is a common metadata standard for geospatial information. Using such standards ensures that data from different sources can be integrated, compared, and analysed effectively by the wider scientific community. The DMP should specify which domain-specific standards will be adopted and why.

  • Metadata Standards: Metadata – ‘data about data’ – provides essential descriptive information that makes datasets understandable and usable. Implementing consistent metadata schemas is critical. These schemas can be general-purpose, like Dublin Core, which provides a simple yet effective set of elements for resource description, or domain-specific, such as Ecological Metadata Language (EML) for environmental data, Data Documentation Initiative (DDI) for social science data, or DICOM for medical imaging. Metadata typically includes descriptive elements (title, author, keywords, abstract), structural elements (data relationships, file organisation), administrative elements (creation date, access rights, licenses), and preservation elements (checksums, file formats). The DMP should detail which metadata standards will be used, how metadata will be created and stored, and how it will be linked to the data. The use of Persistent Identifiers (PIDs), such as Digital Object Identifiers (DOIs) for datasets, ensures that data can be uniquely identified and cited even if its storage location changes (DataCite, n.d.).

  • Controlled Vocabularies and Ontologies: Beyond formal metadata schemas, the use of controlled vocabularies, thesauri, and ontologies (e.g., Gene Ontology, SNOMED CT) helps to standardise the terminology used to describe data, making it more consistently searchable and machine-readable. This reduces ambiguity and improves the precision of data discovery and integration across different studies and platforms.

3.3 Policies for Access and Sharing

Clear, ethical, and legally compliant guidelines on data accessibility and sharing are fundamental for responsible data stewardship and maximising the impact of research. This section requires careful consideration of various factors.

  • Access Levels and Embargo Periods: The DMP must define who will have access to the data and under what conditions. This can range from fully ‘open access’ (publicly available with minimal restrictions), to ‘restricted access’ (available upon request with specific terms, e.g., for sensitive data), or ’embargoed access’ (temporarily restricted, often until publication or patent filing). Justifications for any restrictions must be clearly articulated. For instance, data that contains sensitive personal information (e.g., health records, qualitative interviews) must be handled with appropriate access controls and potentially anonymisation or pseudonymisation to protect individual privacy (GDPR, 2016; HIPAA, 1996).

  • Sharing Mechanisms and Repositories: The DMP should outline the platforms or repositories intended for data dissemination. These can include institutional repositories (maintained by universities), disciplinary repositories (specific to a field, e.g., GenBank for genetic sequences, PDB for protein structures, ICPSR for social sciences), or generalist repositories (e.g., Zenodo, Figshare, Dryad) that accept data from any discipline (Bik & Poisot, 2022). The choice of repository should be guided by its ability to assign persistent identifiers (like DOIs), its commitment to long-term preservation, and its compliance with relevant disciplinary standards and certifications (e.g., CoreTrustSeal). The plan should specify when the data will be deposited and how it will be prepared for deposit.

  • Ethical, Legal, and Intellectual Property (ELIP) Considerations: This is a critical and often complex aspect. The DMP must explicitly address:

    • Privacy and Confidentiality: For data involving human subjects, researchers must detail strategies for anonymisation or pseudonymisation, secure data storage, and the process of obtaining informed consent that covers data sharing. Compliance with regulations such as the General Data Protection Regulation (GDPR) in Europe or the Health Insurance Portability and Accountability Act (HIPAA) in the US is mandatory.
    • Intellectual Property Rights (IPR): The plan must clarify who owns the data (e.g., individual researcher, institution, funding agency, sponsor) and how IPR will be managed. This includes discussing copyright, database rights, and patenting intentions. Using appropriate data licenses, such as Creative Commons licenses (e.g., CC BY, CC BY-SA) or Open Data Commons licenses, defines the terms under which others can reuse the data, ensuring both protection for creators and maximum permissible reuse (Creative Commons, n.d.).
    • Export Controls and National Security: In certain fields, particularly those involving advanced technologies or dual-use research, data sharing might be subject to national export control regulations, which must be explicitly addressed.

3.4 Provisions for Reuse and Preservation

Ensuring the long-term viability and reusability of research data is a core objective of any DMP. This section details the strategies for making data enduringly valuable.

  • Archiving and Long-Term Preservation Strategies: The DMP must specify how data will be preserved beyond the active phase of the project. This involves selecting appropriate, trusted digital repositories (TDRs) that are committed to long-term data curation and preservation (e.g., through certification like CoreTrustSeal). Distinguish between active working storage and archival storage. The plan should outline the timeline for data deposit, the costs associated with preservation (if any), and the responsible party for ensuring data migration to new formats or technologies as they evolve. Digital curation involves active management and appraisal of data over time to maintain its usability and value (BSI, 2014).

  • Data Formats for Preservation: Reiterate the importance of choosing non-proprietary, open, and widely supported file formats for preservation. The plan should also consider migration strategies for data originally collected in proprietary formats, detailing how these might be converted to more stable, open formats for archival purposes to facilitate future use, even if the original software becomes obsolete.

  • Documentation for Reuse: Beyond metadata, comprehensive documentation is crucial for enabling future researchers to understand and reuse the data correctly. This includes:

    • README files: Providing a general overview of the dataset, file structure, contents, and any specific instructions.
    • Data Dictionaries/Codebooks: Detailed explanations of all variables, their definitions, units of measurement, valid ranges, and any coding schemes used.
    • Methodological Documentation: A clear description of data collection methods, experimental protocols, data processing steps, quality control procedures, and software/scripts used for analysis. This includes any scripts, code, or workflows that transform raw data into processed outputs.
    • Data Provenance: A clear record of the data’s origin, any transformations it has undergone, and who made those changes, ensuring traceability and trustworthiness.
    • Version Control: Implementing systems (e.g., Git) for tracking changes to both data files and associated documentation.

3.5 Budget Considerations

Effective data management is not without cost, and a robust DMP must realistically account for the financial and resource implications throughout the project lifecycle. This section ensures that adequate resources are allocated for all data-related activities.

  • Personnel: Budgeting for dedicated personnel is often overlooked but critical. This includes data managers, data stewards, research software engineers, statisticians, IT support staff, and data curators who are responsible for data quality, organisation, metadata creation, and repository interaction. Training costs for researchers to acquire necessary data management skills should also be considered.

  • Infrastructure and Tools: This encompasses various costs associated with technology:

    • Storage Solutions: Costs for active working storage (e.g., cloud storage, institutional network drives) and long-term archival storage (e.g., national data centres, disciplinary repositories). These costs can vary based on volume, security requirements, and data access frequency.
    • Computational Resources: Costs for high-performance computing (HPC) or cloud computing for data processing and analysis, especially for large or complex datasets.
    • Software Licenses: Expenses for specialized software tools for data collection, cleaning, analysis, visualisation, metadata creation, anonymisation, and version control.
    • Networking: Costs associated with high-speed data transfer between collaborators or to repositories.

  • Compliance and Dissemination Costs: Adhering to regulatory requirements may incur costs, such as legal consultation for data privacy, security audits, or fees for depositing data in certified repositories that ensure long-term preservation. Open access publication fees for data papers or supplementary data accompanying journal articles should also be factored in, as these enhance data visibility and impact.

  • Data Transfer and Ingest Costs: For very large datasets, the costs associated with transferring data to an archive (e.g., network bandwidth, shipping physical media) or paying for data ingestion services might be significant.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

4. Best Practices for Developing and Maintaining Data Management Plans

The utility and effectiveness of a DMP are directly correlated with the rigor of its development and the diligence of its maintenance. Adhering to best practices transforms the DMP from a bureaucratic obligation into a dynamic and indispensable tool for successful research.

4.1 Planning Ahead

Proactive and early planning is the bedrock of effective data management, setting the stage for smooth project execution and successful data stewardship.

  • Early Integration into Project Lifecycle: Data management strategies should not be an afterthought but rather integrated from the project’s inception. This means incorporating DMP development into the initial grant writing phase, aligning it with research methodology design, and considering data implications during ethical review applications. An iterative approach is beneficial, where the DMP is a living document that evolves with the project rather than a static plan (Case Western Reserve University, n.d.). This early engagement helps identify potential challenges and solutions before they become critical issues.

  • Stakeholder Engagement and Defined Roles: Effective data management is a team effort. Involving all relevant stakeholders – principal investigators, co-investigators, post-docs, graduate students, data managers, IT support staff, librarians, and legal/ethics advisors – from the outset ensures comprehensive coverage and shared understanding. The DMP should clearly define roles and responsibilities for each aspect of data management (e.g., who is responsible for data entry, quality control, metadata creation, backups, and repository submission). This avoids ambiguity and ensures accountability.

  • Utilising DMP Tools: A variety of online tools can assist in DMP creation, such as DMPTool (USA) and DMPonline (UK and international). These tools provide templates, guidance, and examples tailored to specific funding agency requirements, making the process more streamlined and ensuring all necessary components are addressed (DMPTool, n.d.; DMPonline, n.d.). They also often allow for collaborative editing and version control.

  • Alignment with Institutional Policies: Researchers should familiarise themselves with their institution’s research data management policies, which often provide resources, infrastructure, and guidelines. Aligning the project’s DMP with institutional frameworks can streamline compliance and access to support services.

4.2 Organizing and Documenting Data

Effective data organisation and thorough documentation are critical for data usability, reproducibility, and long-term value, preventing ‘data archaeology’ where future users struggle to interpret the data.

  • Consistent Naming Conventions: Establishing clear, systematic, and descriptive file naming conventions from the beginning is paramount. File names should include elements like project ID, date (YYYYMMDD), experiment name, sample ID, version number, and file type (e.g., ‘ProjectX_ExpA_20230315_Sample001_v1.csv’). This practice aids in file retrieval, avoids confusion, and supports version control. The DMP should explicitly detail these conventions.

  • Structured Directories and Folder Hierarchies: Creating logical, intuitive, and consistent folder hierarchies helps to organise data files and related documentation. A common structure might include separate folders for raw data, processed data, analysis scripts, documentation (e.g., READMEs, codebooks), and publications. This systematic arrangement facilitates data retrieval and ensures that all associated materials are easily locatable (Gonzaga University, n.d.).

  • Comprehensive Documentation: Beyond basic metadata, detailed documentation is essential. This includes:

    • README Files: For each dataset or folder, a README file (e.g., in plain text or Markdown) should provide an overview, describe the contents, explain the file structure, list any dependencies, and offer instructions for use.
    • Data Dictionaries and Codebooks: These define every variable in a dataset, including its name, description, data type, units of measurement, valid ranges, and explanation of coded values. For qualitative data, a codebook explains the coding scheme used for thematic analysis.
    • Methodological Notes: Detailed descriptions of data collection protocols, instrumentation, experimental setups, quality control procedures, and data processing steps. This should be granular enough for another researcher to replicate the methodology. Digital lab notebooks can be instrumental here.
    • Version Control: Implementing a version control system (e.g., Git, SVN) for both data files (if feasible for size) and, more importantly, for code, scripts, and documentation ensures that changes are tracked, previous versions can be retrieved, and collaborative work is managed effectively (Himmelfarb Health Sciences Library, 2022).

  • Quality Control (QC) Procedures: The DMP should outline specific QC steps to ensure data accuracy and reliability. This includes data validation rules (e.g., range checks, consistency checks), error detection mechanisms, and procedures for handling missing values or outliers. Documenting these procedures is as important as implementing them.

4.3 Storing and Preserving Data

Reliable storage and long-term preservation are fundamental to safeguarding research investments and ensuring data availability for future generations.

  • Reliable Storage Solutions: Differentiate between active working storage and archival storage. Active storage should be secure, accessible to the research team, and ideally institutionally managed (e.g., network drives, cloud storage services with institutional agreements) to ensure security and IT support. Long-term archival storage should be in a trusted digital repository (TDR) committed to preservation standards and certified against frameworks like CoreTrustSeal (CoreTrustSeal, n.d.). The DMP should specify the chosen solutions and justify their suitability based on data volume, sensitivity, and access requirements.

  • Regular Backups and Recovery Plans: Implementing a robust backup strategy is non-negotiable to prevent data loss. The ‘3-2-1 rule’ is a widely accepted best practice: keep at least three copies of your data, store two copies on different storage media, and keep one copy offsite. Automated backup protocols are preferable. Crucially, the DMP should detail a data recovery plan, including how backups will be tested periodically to ensure they are functional and data can be restored effectively (Oak Ridge National Laboratory, n.d.).

  • Data Security: For all data, especially sensitive or confidential information, the DMP must detail security measures. This includes access controls (role-based permissions), encryption (for data at rest and in transit), secure physical storage for hardware, and audit logs to track data access and modifications. Regular security assessments and adherence to institutional IT security policies are vital.

  • Data Format Selection for Preservation: Reiterate the emphasis on open, non-proprietary, and well-documented file formats (e.g., CSV, TIFF, XML, JSON, NetCDF, HDF5) that are likely to remain readable and usable in the long term. The DMP should address strategies for migrating data from proprietary formats to open formats if necessary, to ensure future accessibility, even as technology evolves.

4.4 Sharing Data Responsibly

Sharing data responsibly encompasses not only making it available but doing so ethically, legally, and in a manner that maximises its utility and impact while protecting sensitive information.

  • Open Access Repositories and Data Publication: Depositing data in publicly accessible, curated repositories is a key mechanism for promoting transparency, increasing research impact, and facilitating reuse. The DMP should specify the chosen repository, which may be disciplinary (e.g., NCBI, PDB, UK Data Service) or generalist (e.g., Zenodo, Figshare, Dryad). Criteria for selecting a repository include its sustainability, adherence to FAIR principles, assignment of persistent identifiers (DOIs), and subject-matter expertise. Data publication, often accompanied by a ‘data paper,’ provides a formal citation and peer review for the dataset itself, increasing its visibility and potential for reuse.

  • Data Licensing: Clearly defining usage rights and restrictions through appropriate licenses is crucial. Creative Commons licenses (e.g., CC BY for maximum reuse with attribution, CC BY-NC for non-commercial use) are commonly used for data. For public domain data, CC0 (Public Domain Dedication) can be applied. The DMP must specify the chosen license and explain its implications for data reuse, ensuring that data creators receive appropriate attribution while enabling broad dissemination (Creative Commons, n.d.). For sensitive data, specific data use agreements or restricted licenses may be necessary.

  • Attribution and Data Citation: Ensuring proper credit is given to data creators is essential for academic recognition and promoting data sharing. The DMP should outline how persistent identifiers (like DOIs) will be used to make datasets citable, adhering to community-recognised data citation standards. This allows data to be tracked and acknowledged as a valuable research output, similar to scholarly articles.

  • Embargo Periods and Controlled Access: If data cannot be immediately made fully open, the DMP should outline the conditions for any embargo periods (e.g., to protect intellectual property, allow for publication, or process patent applications) and their duration. For sensitive data, controlled access mechanisms (e.g., secure data enclaves, data access committees, anonymised microdata files) should be detailed, ensuring that access is granted only to authorised researchers under strict terms and conditions, often via a formal data access agreement.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

5. Discipline-Specific Considerations and Templates

While the core components of a DMP are universally applicable, the practical implementation and emphasis on certain aspects vary significantly across different research disciplines due to distinct data types, ethical considerations, regulatory landscapes, and community norms. Tailoring the DMP to specific disciplinary requirements is crucial for its effectiveness.

5.1 Biomedical Research

Biomedical research, particularly involving human subjects, presents some of the most stringent data management challenges due to the highly sensitive nature of health information and the complex regulatory environment.

  • Clinical Data Management and Regulatory Compliance: Adhering to Good Clinical Data Management Practices (GCDMP) is paramount (Wikipedia, 2022b). This involves meticulous procedures for data collection, validation, query management, coding, and quality assurance to ensure data accuracy, reliability, and completeness in clinical trials. Compliance with regulatory bodies such as the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) is mandatory, often requiring extensive documentation, audit trails, and secure data handling. Electronic Health Records (EHRs) data presents specific challenges related to integration, standardisation, and de-identification for research use.

  • Data Sharing Policies and Ethical Considerations: The NIH Data Management and Sharing Policy (2023) is a prime example of a funder mandate. Data sharing for human subjects research requires robust anonymisation or pseudonymisation strategies to protect patient privacy, adhering strictly to regulations like HIPAA (USA) and GDPR (Europe). Informed consent forms must explicitly state how participant data will be shared, with whom, and under what conditions. Sensitive genomic data requires careful management, often with tiered access models (e.g., dbGaP, EGA). Repositories like GEO (Gene Expression Omnibus), SRA (Sequence Read Archive), and ArrayExpress are domain-specific for molecular biology data. Imaging data (e.g., MRI, CT scans) requires adherence to standards like DICOM for interoperability.

  • Data Security: Due to the high sensitivity of patient data, stringent data security measures are essential, including strong encryption, multi-factor authentication, secure data enclaves, and regular security audits. Data breaches in this domain carry severe legal and ethical penalties.

5.2 Environmental Sciences

Environmental sciences often deal with vast, heterogeneous datasets originating from various sources, presenting unique challenges for standardisation and integration.

  • Geospatial Data and Remote Sensing: This field heavily relies on geospatial data, including satellite imagery, drone data, GIS layers, and GPS coordinates. Implementing standards for geospatial data, such as those from the Open Geospatial Consortium (OGC) and ISO 19115/19139 for metadata, is crucial for interoperability. The DMP must detail coordinate reference systems and how they will be managed. Remote sensing data often comes in very large volumes, requiring specialised storage and processing capabilities.

  • Sensor Data and Time Series: Data from environmental sensors (e.g., weather stations, oceanographic buoys, air quality monitors) generates continuous time-series data. The DMP needs to address real-time data ingestion, quality control, time-stamping accuracy, and data aggregation methods. Specific challenges include managing diverse sensor formats and ensuring long-term accessibility of continuous streams.

  • Data Repositories: Domain-specific repositories are prevalent, such as NASA Earthdata, NOAA National Centers for Environmental Information (NCEI), PANGAEA for marine and environmental data, and the Long-Term Ecological Research (LTER) network. The DMP should specify which repositories will be used and how data will be prepared for submission, often including specific metadata standards like Ecological Metadata Language (EML).

  • Reproducible Workflows: Given the complexity of environmental models and data processing, documenting computational workflows and providing code/scripts (e.g., in R, Python, Matlab) alongside the data is vital for reproducibility and transparency.

5.3 Social Sciences

Social sciences frequently involve human subjects, leading to significant ethical considerations regarding privacy, confidentiality, and informed consent, particularly for qualitative data.

  • Qualitative Data Management: Managing qualitative data (e.g., interview transcripts, focus group recordings, ethnographic field notes) presents unique challenges in anonymisation, especially for rich, descriptive narratives. The DMP must outline rigorous processes for pseudonymisation or redaction, secure transcription, and ethical storage to ensure participant confidentiality. Challenges include preserving the context and richness of qualitative data while protecting privacy.

  • Survey and Quantitative Data: For survey data, careful attention must be paid to anonymising demographic variables and protecting personally identifiable information (PII). Managing longitudinal or panel data requires robust version control and consistent data cleaning over time. The DMP should detail methods for data cleaning, aggregation, and the creation of synthetic or anonymised microdata files suitable for sharing.

  • Data Repositories: Established repositories like the Inter-university Consortium for Political and Social Research (ICPSR), the UK Data Service, and QualBank (for qualitative data) are critical for sharing social science data. These repositories often provide expert curation services and secure access mechanisms, including controlled access for sensitive datasets.

  • Legal and Ethical Compliance: All human subjects research must adhere to strict protocols set by Institutional Review Boards (IRBs) or ethics committees. The DMP must explicitly reference these approvals and detail how informed consent covers data storage, sharing, and potential future reuse.

5.4 Digital Humanities

Digital Humanities projects often involve heterogeneous data, from text corpora and digitised cultural artifacts to complex networks, posing unique challenges for metadata and long-term preservation of unique digital objects.

  • Text Corpora and Cultural Heritage Data: Managing large collections of digitised texts, manuscripts, images, audio, and video requires specialised approaches. The DMP must address digitisation standards, optical character recognition (OCR) accuracy, and methods for text encoding (e.g., Text Encoding Initiative – TEI XML for literary and linguistic texts). Metadata for cultural heritage objects needs to be rich and descriptive (e.g., using CIDOC CRM, MODS) to capture their unique provenance and context.

  • Data Types: Data can range from structured databases of historical events to intricate social networks, 3D models of archaeological sites, or interactive digital editions. The DMP must detail how these diverse formats will be managed, cross-referenced, and preserved.

  • Annotation and Curation: Many DH projects involve extensive manual annotation of texts or images. The DMP should outline annotation schemes, tools, and how annotations will be stored and linked to the source data. Digital curation is crucial for ensuring the long-term interpretability and usability of complex digital objects.

  • Intellectual Property: Given the frequent use of copyrighted materials (e.g., modern texts, artworks), the DMP must meticulously address intellectual property rights, fair use, licensing for digitised content, and any associated access restrictions.

5.5 Engineering and Computational Sciences

These fields often generate massive datasets from simulations, sensor arrays, and experimental rigs, frequently involving proprietary software and complex data structures.

  • Simulation and Experimental Data: Large-scale simulations produce vast amounts of numerical data, often in custom binary formats. Experimental data from sensors (e.g., in materials science, robotics, aerospace) can be high-frequency and multi-dimensional. The DMP needs to specify how these data will be stored, validated, and linked to the computational models or experimental protocols.

  • Proprietary Formats and Software Dependencies: Engineering data often relies on proprietary software (e.g., CAD files, specific simulation outputs). The DMP must address strategies for converting these to open formats for long-term preservation or ensuring access to the necessary software environment for future reuse. Documenting the computational environment (operating system, libraries, software versions) is critical for reproducibility.

  • Intellectual Property and Commercialisation: Given the close links to industry and potential for commercialisation, IP protection is a major concern. The DMP must clearly outline how data will be protected, licensed, and potentially embargoed to support patent applications or commercial development, balancing open science with industrial competitiveness.

  • High-Performance Computing (HPC) Data: Projects leveraging HPC resources generate data at immense scales. The DMP must detail strategies for efficient data transfer, distributed storage, and parallel processing, often in conjunction with institutional HPC facilities and data storage solutions.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

6. Common Pitfalls and How to Avoid Them

While the concept of DMPs is well-established, their effective implementation can be hindered by common pitfalls. Awareness of these challenges and proactive strategies to mitigate them are crucial for enhancing DMP effectiveness and ensuring successful data stewardship.

  • Neglecting Regular Updates: A DMP should be a living document, not a static one written solely for a grant proposal. Research projects are dynamic, evolving in scope, methodology, team members, and data types. A common pitfall is to write a DMP at the outset and then never revisit or update it. This can lead to a plan that no longer accurately reflects current project realities, creating discrepancies and making compliance difficult. Avoidance Strategy: Schedule regular reviews of the DMP (e.g., quarterly, or at key project milestones). Designate a ‘DMP Steward’ within the team responsible for tracking changes and initiating updates. Implement version control for the DMP itself, noting dates and reasons for revisions (Himmelfarb Health Sciences Library, 2022).

  • Overlooking Comprehensive Compliance: Researchers may focus solely on funder mandates, neglecting other critical compliance aspects such as institutional policies, ethical guidelines (e.g., IRB approvals), legal requirements (e.g., GDPR, HIPAA, export controls), and intellectual property laws. Failure to comply can lead to grant revocation, legal repercussions, significant fines, reputational damage, and project delays. Avoidance Strategy: Conduct a thorough audit of all relevant policies and regulations at the project’s inception. Engage with institutional legal counsel, research ethics boards, and library data services early in the planning process. Ensure all consent forms explicitly address data sharing. Clearly document all compliance requirements and how they will be met within the DMP.

  • Inadequate Documentation and Metadata: Providing insufficient metadata and documentation is a pervasive issue that severely hampers data reuse. Without clear explanations of variables, methodologies, data cleaning steps, and contextual information, data can become unintelligible even to the original creators after some time, let alone to external researchers. This makes reproducibility impossible and diminishes the long-term value of the dataset. Avoidance Strategy: Prioritise creating rich, standardised metadata and comprehensive documentation (README files, data dictionaries, codebooks, methodological protocols) throughout the project. Integrate metadata creation into the daily workflow rather than leaving it until the end. Utilise automated metadata generation tools where possible and ensure that documentation is version-controlled and stored alongside the data (Gonzaga University, n.d.).

  • Underestimating Resource Requirements: Researchers often underestimate the time, effort, skills, and financial resources needed for effective data management. This can lead to understaffing, inadequate storage, reliance on unsuitable tools, and insufficient budget allocation for long-term preservation, leading to project inefficiencies and potential data loss. Avoidance Strategy: Conduct a realistic assessment of all data management needs at the proposal stage. Budget for dedicated data management personnel, appropriate storage solutions, software licenses, and potential repository fees. Seek advice from institutional data stewards or librarians for accurate cost estimations. Consider training for team members in data management best practices.

  • Lack of Data Security Planning: Failing to adequately plan for data security, particularly for sensitive data, exposes the project to significant risks of data breaches, unauthorised access, and loss of public trust. This is especially critical in fields handling human subject data or proprietary information. Avoidance Strategy: Implement a robust data security plan from the outset. This includes defining access controls, using encryption for data at rest and in transit, securing physical storage, maintaining audit trails, and adhering to institutional IT security policies. Conduct regular risk assessments and security audits, and train staff on data protection protocols. For highly sensitive data, consider secure data enclaves or privacy-preserving technologies.

  • Choosing Inappropriate Repositories or Formats: Selecting a repository that is not aligned with disciplinary norms, does not offer long-term preservation, or is not trusted can undermine the findability and reusability of data. Similarly, relying on proprietary or niche data formats without a clear migration strategy risks making data inaccessible in the future. Avoidance Strategy: Research and select trusted digital repositories early in the planning process, considering disciplinary relevance, certification (e.g., CoreTrustSeal), persistent identifier assignment, and long-term sustainability. Prioritise open, non-proprietary, and widely used data formats for preservation and sharing. Plan for data migration from proprietary to open formats if necessary (Oak Ridge National Laboratory, n.d.).

  • Poor Collaboration and Communication: Data management is inherently collaborative. A lack of clear communication channels and shared understanding among team members regarding data management responsibilities can lead to inconsistencies, duplicated efforts, or neglected tasks. Avoidance Strategy: Establish clear lines of communication and define roles and responsibilities within the DMP. Regular team meetings should include discussions on data management progress and any emerging issues. Foster a culture of data stewardship within the research team.

  • Ignoring Ethical Dimensions Beyond Privacy: While privacy is crucial, other ethical considerations might be overlooked, such as potential biases in data collection or algorithms, the implications of data reuse for vulnerable populations, or issues of digital colonialism when sharing data from indigenous communities. Avoidance Strategy: Engage with ethical review boards and community representatives. Develop clear ethical guidelines within the DMP that go beyond legal compliance to address broader societal impacts. Ensure informed consent processes are culturally sensitive and comprehensive.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

7. Conclusion

A meticulously crafted and diligently maintained Data Management Plan is unequivocally integral to the success, integrity, and enduring impact of any contemporary research endeavour. Far from being a mere administrative burden, a DMP serves as a sophisticated strategic framework that not only facilitates efficient and rigorous data handling throughout the research lifecycle but also profoundly underpins the core principles of transparency, scientific reproducibility, and the invaluable long-term preservation of knowledge assets. In an increasingly data-intensive and interconnected research landscape, the proactive development and continuous stewardship of a DMP empowers researchers to navigate complex data challenges, mitigate significant risks, and significantly enhance the trustworthiness and utility of their scientific outputs.

By systematically addressing the diverse components detailed in this report – from the precise characterisation of data types and the implementation of robust metadata standards to the careful consideration of ethical access, secure preservation strategies, and realistic budgetary allocations – researchers can transform raw data into a valuable, discoverable, and reusable scientific legacy. Adhering to the outlined best practices, such as proactive planning, meticulous documentation, secure storage, and responsible sharing, ensures that research data remains of high quality, readily accessible, and fully compliant with the evolving demands of funding bodies and institutional policies. Moreover, tailoring the DMP to specific disciplinary requirements acknowledges the unique nuances of different research fields, from the stringent regulatory landscape of biomedical research to the ethical complexities of social science data or the vast volumes encountered in environmental and computational sciences.

Looking ahead, the landscape of data management will continue to evolve, driven by advancements in artificial intelligence for data curation, the development of increasingly sophisticated persistent identifiers, and the integration of semantic web technologies to enhance data interoperability. Researchers who embrace DMPs as dynamic, indispensable tools, continuously revisiting and refining them, will be best positioned to meet these future challenges and maximise the societal and scientific impact of their work. Ultimately, a well-executed DMP is not just about managing data; it is about building trust in science, fostering collaboration, and securing the enduring value of human knowledge for generations to come.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

References

1 Comment

  1. So, DMPs! Who knew organizing digital breadcrumbs could be so strategic? This report’s thoroughness sparks a question: How can we ensure that *all* researchers, regardless of resources, can craft and implement gold-standard DMPs? Maybe AI could analyze research proposals and auto-generate a starter DMP? Food for thought!

    Reply

Leave a Reply to Mason Poole Cancel reply

Your email address will not be published.


*