NSF Data Management Plan: Requirements and Writing Tips

The National Science Foundation requires a data management plan (DMP) as a supplementary document for all grant proposals. This plan describes how a project will manage, preserve, and share research data. NSF implemented this requirement in 2011 with the intent of facilitating the dissemination and sharing of research results. A study of NSF-funded researchers at Oregon State University found that sharing at both the project level and the journal article level was not carried out in the majority of cases, and when sharing was accomplished, the shared data were often of questionable usability due to access, documentation, and formatting issues. This article explains the NSF requirements, provides a template for writing a compliant plan, and offers practical tips for addressing each element.

At a Glance

The table below summarizes the core NSF DMP requirements and the practical actions you must take to satisfy each one.

Understanding the NSF Data Management Plan Requirement

The NSF data management plan requirement applies to all proposals submitted to the agency. The plan is a supplementary document, typically limited to two pages, that describes how the project will conform to NSF policy on the dissemination and sharing of research results. The DMP must be submitted with the proposal and is reviewed as part of the merit review process.

The NSF requirement went into effect in 2011. Since that time, researchers have been expected to implement the elements of the data management plans required for their grant proposals. The Oregon State University study evaluated data sharing practices of researchers funded by NSF at that institution by attempting to discover project-level research data using the associated DMP as a starting point and by examining data sharing associated with journal articles that acknowledge NSF support. The findings showed that sharing was not carried out in the majority of cases, and when sharing was accomplished, the shared data were often of questionable usability due to access, documentation, and formatting issues.

The practical implication is that writing a DMP is not a paperwork exercise. The plan you submit becomes the standard against which your data sharing performance will be evaluated. A plan that is vague about repositories, formats, or timelines creates ambiguity that can lead to poor sharing outcomes. A plan that is specific about these elements creates accountability.

Core Principles for Writing a Compliant DMP

Specificity Over Generality

A common failure in DMP writing is the use of generic language that could apply to any project. Reviewers and program officers look for plans that demonstrate the proposer has thought concretely about the data their project will generate. For each dataset, you should specify the file format, the estimated volume, the software needed to read the files, and the repository where the data will be deposited.

The Oregon State University study found that shared data were often of questionable usability due to access, documentation, and formatting issues. This finding points directly to the need for specificity in the DMP. If you do not specify the documentation standards you will use, the metadata schema, and the file formats, the data you deposit may be difficult or impossible for others to use.

Alignment With Disciplinary Norms

Different fields have different data sharing expectations and repository infrastructures. A DMP for a genomics project will reference different repositories and standards than a DMP for an ecological monitoring project. Your plan should reflect the norms of your discipline. If your field has established data repositories, name them. If your field has metadata standards, reference them.

The National Institute of Standards and Technology maintains the Research Data Framework, which provides a structure for understanding the components of research data management. This framework can help you identify the elements your DMP should address, including data description, preservation, access, and reuse.

Realistic Assessment of Effort and Cost

Data management requires personnel time and financial resources. Your DMP should reflect a realistic assessment of what it will take to manage the data your project produces. This includes time for documentation, quality control, and deposit. The budget justification should include these costs.

The NSF/NIH Effect study surveyed the effect of data management requirements on faculty, sponsored programs, and institutional repositories. The findings from this survey indicate that institutions have had to develop new infrastructure and support services in response to these requirements. Your plan should acknowledge the support available at your institution and the costs that will be borne by the project.

Practical Workflow for Writing Your DMP

Step 1: Inventory Your Data Types

Begin by listing every type of data your project will produce. This includes raw data, processed data, derived data, code, software, and documentation. For each data type, record the following information:

• File format and version
• Estimated volume in gigabytes or terabytes
• Software needed to read or process the files
• Whether the data are unique or can be regenerated
• Any confidentiality or privacy considerations

This inventory becomes the foundation for the data description section of your DMP.

Step 2: Identify Your Repositories

For each dataset, identify the repository where you will deposit the data. Consider the following factors:

• Whether the repository is the standard choice in your discipline
• Whether the repository accepts your file formats
• Whether the repository provides persistent identifiers such as DOIs
• Whether the repository has any costs associated with deposit or storage
• Whether the repository has any restrictions on data access

If your discipline does not have an established repository, identify a general-purpose repository that meets your needs.

Step 3: Define Your Documentation Standards

Documentation is the element that most often determines whether shared data are usable. Your DMP should specify the documentation you will create, including:

• Readme files that describe the dataset structure
• Metadata that follows a recognized schema
• Codebooks that define variables and values
• Protocols that describe data collection methods

The EQUATOR Network provides reporting guidelines for health research that can inform the documentation standards for clinical and epidemiological data. For microbiome research, the STREAMS guidelines provide a checklist for the reporting of study information, experimental design, and analytical methods in a standardized and machine-actionable manner.

Step 4: Describe Your Preservation Strategy

Preservation means more than storing files on a server. Your DMP should describe how data will be preserved for the long term, including:

• The repository that will hold the data after the project ends
• The backup strategy during the project
• The file formats that will be used for long-term preservation
• The retention period for the data

Step 5: Address Access and Restrictions

NSF policy expects data to be shared, but it also recognizes that some data cannot be shared openly. Your DMP should describe any restrictions on access and the reasons for those restrictions. Common restrictions include:

• Human subjects protections that limit data sharing
• Proprietary data from industry partners
• National security considerations
• Embargo periods for dissertation research

For each restriction, describe the mechanism that will allow appropriate access. This might include a data use agreement, a de-identification process, or a controlled access repository.

Step 6: Assign Roles and Responsibilities

Name the individuals who will be responsible for data management tasks. This includes the principal investigator, co-investigators, research staff, and students. For each person, describe their specific responsibilities. The plan should also identify who will be responsible for the data after the project ends.

Step 7: Review Against NSF Requirements

Before submitting your proposal, review your DMP against the NSF requirements for your specific directorate or program. Some NSF programs have additional data management requirements beyond the general policy. Check the program solicitation for any specific instructions.

Options and Tradeoffs in Data Management

Repository Selection

The choice of repository involves tradeoffs between disciplinary norms, cost, and functionality. A disciplinary repository may offer the advantage of being the expected choice in your field, but it may have limitations on file formats or data volume. A general-purpose repository may offer more flexibility but may not provide the specialized services that a disciplinary repository offers.

The National Center for Biotechnology Information provides access to a range of databases for biomedical and genomic data. For researchers in these fields, depositing data in NCBI databases is often the expected practice. The NCBI Literature Resources provide access to the scientific literature that can help you identify the standards and practices in your field.

Data Sharing Timing

NSF policy expects data to be shared in a timely manner, but the policy does allow for embargoes in some circumstances. The tradeoff is between the benefit of immediate sharing and the need to protect the ability of the research team to publish their findings. Your DMP should specify the timing of data release and the justification for any embargo.

Data Format Choices

The choice of file formats affects both usability and preservation. Open formats that do not require proprietary software are generally preferred for preservation. However, some data types are only available in proprietary formats. Your DMP should address this tradeoff by specifying the formats you will use and the software needed to read them.

Documentation Depth

The level of documentation you commit to in your DMP affects both the usability of your data and the effort required to prepare it for sharing. Minimal documentation may be sufficient for simple datasets, but complex datasets require detailed documentation to be usable by others. The Oregon State University study found that documentation issues were a common reason for shared data being of questionable usability.

Observations and Measurements for DMP Compliance

Tracking Data Sharing Outcomes

Once your project is funded, you should track your data sharing activities against the commitments in your DMP. The following records will help you document compliance:

• Dates of data deposit to repositories
• Persistent identifiers assigned to datasets
• Documentation files created for each dataset
• Data access requests and how they were handled
• Data use agreements executed for restricted data

These records serve two purposes. They provide evidence of compliance if questions arise, and they help you identify gaps in your data management practices.

Assessing Data Usability

The Oregon State University study assessed data sharing effectiveness by examining whether shared data were usable. The study found that shared data were often of questionable usability due to access, documentation, and formatting issues. To avoid these problems, you should assess your own data from the perspective of a potential user. Ask the following questions:

• Can someone outside your research group find the data?
• Can they open the files with standard software?
• Can they understand the variables and their values?
• Can they determine how the data were collected and processed?
• Can they verify the quality of the data?

If you cannot answer yes to these questions, your data are not ready for sharing.

Measuring Compliance Rates

Institutional studies of data management plan compliance provide useful benchmarks. A study of an institutional implementation of the NIH Data Management and Sharing policy found that 79.3 percent of submitted plans addressed all required elements. Element-level compliance ranged from 98.9 percent for data type to 82.7 percent for tools and software. Sub-element scores showed greater variability, with 98.9 percent completion for data description and 49.3 percent for data generation. The study also found that unstructured plans consistently underperformed compared to structured plans.

These findings have direct implications for NSF DMP writing. The use of a structured template improves the likelihood that all required elements will be addressed. The study recommended using structured templates to enhance the quality and consistency of data management plans.

Records and Documentation Standards

Metadata Standards

Metadata is the structured information that describes your data. Different disciplines have different metadata standards. Your DMP should identify the metadata standard you will use and describe how you will create metadata that conforms to that standard.

For biomedical research, the NCBI provides databases that require specific metadata formats for submission. For environmental research, the STREAMS guidelines provide a checklist for reporting study information, experimental design, and analytical methods in a standardized and machine-actionable manner.

Codebooks and Data Dictionaries

A codebook or data dictionary defines the variables in a dataset, their values, and their meanings. This documentation is essential for data usability. Your DMP should specify that you will create codebooks for each dataset and describe the level of detail they will contain.

Readme Files

A readme file provides an overview of a dataset, including its purpose, structure, and any special considerations for use. The readme should be deposited with the data and should be written for an audience that is not familiar with your research project.

Version Control

If you will be sharing multiple versions of a dataset, your DMP should describe your version control practices. This includes how versions will be identified, how changes between versions will be documented, and how users will be able to access previous versions.

Quality Controls for Data Management

Data Quality Checks

Your DMP should describe the quality checks you will perform on your data before deposit. These checks might include:

• Validation of data against collection protocols
• Checks for missing or out-of-range values
• Verification that file formats are correct
• Confirmation that documentation matches the data

Peer Review of Data

Some repositories offer peer review of data deposits. This review can identify problems with documentation, formatting, or completeness before the data are released. Your DMP should indicate whether you will seek peer review of your data deposits.

Reproducibility Checks

For computational research, reproducibility is a key quality measure. Your DMP should describe how you will ensure that others can reproduce your analyses from the data and code you deposit. This includes documenting the software environment, version numbers, and parameters used in the analysis.

The Experimental Design Assistant from NC3Rs provides a tool for designing experiments that can improve the quality and reproducibility of research. While this tool is focused on animal research, the principle of designing experiments with reproducibility in mind applies broadly.

Common Failure Patterns in DMP Implementation

Vague Repository Commitments

A common failure is naming a repository without specifying the details of the deposit. A plan that says data will be deposited in a repository without specifying the format, the timeline, or the documentation is not actionable. The Oregon State University study found that shared data were often of questionable usability due to access, documentation, and formatting issues.

Overcommitment to Sharing

Another failure pattern is promising to share data that cannot be shared. This happens when the DMP does not account for human subjects protections, proprietary data, or other restrictions. The result is either a failure to share as promised or a need to renegotiate the plan after funding.

Underestimating Documentation Effort

Documentation is the most time-consuming part of data sharing, and it is often underestimated in DMPs. A plan that does not allocate time for documentation is unlikely to produce usable data. The Oregon State University study found that documentation issues were a common reason for shared data being of questionable usability.

Ignoring Institutional Support

Many institutions have data management support services, including librarians, research data specialists, and repository infrastructure. A DMP that does not reference these services may miss opportunities for support and may not align with institutional practices. The NSF/NIH Effect study surveyed the effect of data management requirements on faculty, sponsored programs, and institutional repositories, finding that institutions have had to develop new infrastructure and support services in response to these requirements.

Treating the DMP as a One-Time Document

A DMP is not a static document. As the project evolves, the data management practices may need to change. A failure to update the DMP when the project changes can lead to a mismatch between the plan and the actual data management practices.

Limitations of the DMP Approach

The Gap Between Plans and Practice

The Oregon State University study demonstrated a significant gap between the commitments in DMPs and the actual data sharing practices of researchers. The study found that sharing at both the project level and the journal article level was not carried out in the majority of cases. This finding suggests that the DMP requirement alone is not sufficient to ensure data sharing.

The Challenge of Data Usability

Even when data are shared, they may not be usable. The Oregon State University study found that shared data were often of questionable usability due to access, documentation, and formatting issues. This finding highlights the need for attention to documentation and formatting in DMPs.

The Variability of Disciplinary Practices

Data management practices vary significantly across disciplines. A DMP that is appropriate for one field may not be appropriate for another. The NSF approach recognizes this variability by allowing proposers to describe practices that are appropriate for their field.

The Cost of Compliance

Data management has real costs, including personnel time, storage, and repository fees. These costs may be significant for large or complex projects. The DMP should address these costs, and the budget justification should include them.

Safety and Regulatory Context

Human Subjects Data

If your research involves human subjects, your DMP must address the protections required by the Common Rule and your institutional review board. This includes describing how data will be de-identified, how access will be controlled, and how data use agreements will be managed. The DMP should be consistent with the informed consent documents and the IRB approval for the project.

Animal Research Data

If your research involves animals, your DMP should address the data related to animal studies. The NC3Rs Experimental Design Assistant provides a tool for designing experiments that can improve the quality and reproducibility of animal research. The tool can help you document the experimental design in a way that supports data sharing.

Controlled Data

Some data are subject to export controls or other regulations that restrict access. Your DMP should identify any controlled data and describe the mechanisms that will be used to comply with applicable regulations.

Data Security

For sensitive data, your DMP should describe the security measures that will be used to protect the data during the project and after deposit. This includes encryption, access controls, and secure storage.

Professional Escalation Criteria

When to Consult Your Institution

You should consult your institutional research office or data management support services in the following situations:

• When you are uncertain about the data management requirements for a specific NSF program
• When your project involves data that may be subject to restrictions
• When you need assistance selecting a repository or metadata standard
• When you are preparing a data management budget

When to Seek Expert Advice

You should seek expert advice from a data management specialist or librarian in the following situations:

• When your data are complex or heterogeneous
• When your field does not have established data sharing practices
• When you are working with sensitive or controlled data
• When you are depositing data to an unfamiliar repository

When to Request a DMP Review

Before submitting your proposal, you should request a review of your DMP by a colleague or a data management specialist. The review should check for completeness, specificity, and alignment with NSF requirements. The study of institutional implementation of the NIH DMS policy found that 20.7 percent of plans omitted one or more required elements, indicating a need for improved policy conformance.

Frequently Asked Questions

What is the maximum length for an NSF data management plan?

NSF typically limits the data management plan to two pages. The plan is a supplementary document that is submitted with the proposal. The specific length requirement may vary by program, so you should check the program solicitation for any specific instructions.

Does every NSF proposal require a data management plan?

Yes, the NSF data management plan requirement applies to all proposals submitted to the agency. The requirement went into effect in 2011. Some NSF programs may have additional data management requirements beyond the general policy, so you should check the program solicitation for any specific instructions.

What happens if my data management plan is not compliant?

The data management plan is reviewed as part of the merit review process. A plan that does not address the required elements may negatively affect the review of your proposal. The plan should be complete and specific to avoid this outcome.

Can I use a template for my NSF data management plan?

Using a structured template is recommended. A study of institutional implementation of the NIH Data Management and Sharing policy found that structured templates demonstrated greater alignment with policy requirements and that unstructured plans consistently underperformed compared to structured plans. The same principle applies to NSF DMPs.

What should I do if my data cannot be shared openly?

Your DMP should describe any restrictions on access and the reasons for those restrictions. Common restrictions include human subjects protections, proprietary data, and national security considerations. For each restriction, describe the mechanism that will allow appropriate access, such as a data use agreement or a controlled access repository.

How should I budget for data management costs?

Your budget justification should include the costs of data management, including personnel time for documentation and deposit, storage costs, and repository fees. The DMP should describe the data management activities that these costs support.

What is the difference between data storage and data preservation?

Data storage refers to the short-term storage of data during the project, including backup systems. Data preservation refers to the long-term retention of data after the project ends, typically in a repository. Your DMP should address both storage and preservation.

How do I know which repository to use for my data?

The choice of repository depends on your discipline, the type of data, and the requirements of your funder. If your field has an established repository, that is often the best choice. If not, a general-purpose repository may be appropriate. Your institutional data management support services can help you identify suitable repositories.

Related Articles

• Research Data Management Plan: From File Naming to Long-Term Sharing
• Protein Data Bank
• Protein Data Bank
• Protein Data Bank
• Pull Down Assay

References and Further Reading

• Research Data Framework. National Institute of Standards and Technology.
• EQUATOR Network. EQUATOR Network.
• Experimental Design Assistant. NC3Rs.
• NCBI Literature Resources. National Center for Biotechnology Information.
• PubMed. National Library of Medicine.
• Misdiagnosis of olfactory neuroblastoma.. Neurosurgical focus, 2002.
• Water, Water, Everywhere: Defining and Assessing Data Sharing in Academia.. PloS one, 2016.
• Development of an evidence-based individualized transition plan for spina bifida.. Neurosurgical focus, 2019.
• Unpacking weight management interventions measuring eating disorder risk in adults: coding of components of interventions in a systematic review.. Journal of eating disorders, 2026.
• Motion blur microscopy: in vitro imaging of cell adhesion dynamics in whole blood flow.. Nature communications, 2024.
• Radiation therapy and CyberKnife radiosurgery in the management of craniopharyngiomas.. Neurosurgical focus, 2008.
• Structured Data Management and Sharing Plan (DMSP) templates outperformed non-structured ones in an institutional implementation of the NIH Data Management and Sharing (DMS) policy.. 2025.
• STREAMS guidelines: standards for technical reporting in environmental and host-associated microbiome studies.. 2025.
• PEX11 mediates intralumenal vesicle formation in peroxisomes.. 2026.
• Directed evolution of LaccID for cell surface proximity labeling and electron microscopy.. 2025.
• Directed evolution of the multicopper oxidase laccase for cell surface proximity labeling and electron microscopy. 2024.
• Data Management Plans. 2012.
• Interview with Deborah Winslow of the National Science Foundation. Anthropological Data in the Digital Age, 2019.
• WNV-BOT portal project summary. 2004.
• Analyzing data management plans: Where librarians can make a difference. ASEE Annual Conference and Exposition Conference Proceedings, 2015.
• The NSF/NIH Effect: Surveying the Effect of Data Management Requirements on Faculty, Sponsored Programs, and Institutional Repositories. Journal of Academic Librarianship, 2014.

This article is educational and does not replace institutional policy, professional advice, or applicable safety and regulatory requirements.