Data Management Plans That Reviewers Actually Want to Read

By Dr. Amara Osei-Mensah · Feb 6, 2026 · 10 min read

Data management plans used to be an afterthought—a two-page document that reviewers skimmed and rarely commented on. That era is over. NSF's new Data Management and Sharing requirements (effective since 2023) and NIH's strengthened Data Management and Sharing Policy have made the DMP a document that can genuinely affect your score. Here's how to write one that works.

Why Data Management Plans Matter Now

Three forces have elevated DMPs from bureaucratic checkbox to scored criterion:

• Reproducibility concerns: Funders are responding to the replication crisis by requiring transparent data practices
• Open science mandates: The 2022 OSTP memo requiring free public access to federally funded research data has real teeth
• Reviewer emphasis: Both NIH and NSF review criteria now explicitly evaluate data management practices

The Framework: Six Sections That Cover Everything

1. Data Types and Sources

Describe exactly what data you'll generate, collect, or reuse. Be specific about formats, estimated volumes, and whether data includes human subjects information. "Survey responses from 500 participants collected via REDCap in CSV format, approximately 2GB total" is far more useful than "survey data."

2. Data Standards and Metadata

Identify the metadata standards relevant to your field. If you're in genomics, reference MIAME or MINSEQE. If you're in social sciences, reference DDI. If no standard exists for your data type, describe the documentation you'll create: variable definitions, codebooks, data dictionaries, and README files.

3. Data Access and Sharing

This is where reviewers pay the most attention. Address:

• When will data be shared? (Timeline relative to publication or project end)
• Where will it be deposited? (Name the specific repository)
• What format will shared data be in? (Prefer open, non-proprietary formats)
• Are there restrictions? (Legitimate reasons include human subjects protections, tribal data sovereignty, or national security)
• How will others find it? (DOIs, persistent identifiers, repository search functions)

4. Data Preservation

Specify where data will be stored long-term and for how long. NIH expects data to be available for at least the duration of the project plus any applicable retention requirements. Name the repository, describe its sustainability, and mention any backup procedures.

5. Data Security and Privacy

If your data includes personally identifiable information, protected health information, or other sensitive data, describe your security measures: encryption, access controls, de-identification procedures, and compliance with relevant regulations (HIPAA, FERPA, etc.).

6. Roles and Responsibilities

Who manages data day-to-day? Who ensures compliance with the DMP? What happens to data stewardship if key personnel leave? Reviewers want to see that data management is someone's explicit responsibility, not an assumption.

Repository Selection: Getting It Right

Choosing the right repository signals competence. Use discipline-specific repositories when possible:

• Genomics: GEO, SRA, dbGaP
• Clinical trials: ClinicalTrials.gov, Vivli
• Social sciences: ICPSR, Dataverse
• General purpose: Dryad, Zenodo, Figshare

Avoid vague statements like "data will be made available upon request." Reviewers interpret this as code for "we don't actually plan to share our data."

Common Mistakes That Cost You Points

1. Generic boilerplate: Copy-pasting a generic DMP from the internet is obvious to reviewers. Tailor every section to your specific project.
2. Ignoring costs: Data management costs money. Include repository fees, storage costs, and personnel time in your budget. A DMP that requires resources you haven't budgeted for lacks credibility.
3. Overpromising on sharing timelines: Promising to share data "immediately upon collection" when you have human subjects data is unrealistic and signals unfamiliarity with the requirements.
4. Forgetting software and code: If your research produces analysis code, scripts, or software, include plans for sharing those too. GitHub with a DOI from Zenodo is a standard approach.

A thoughtful data management plan tells reviewers that you understand modern research practices, respect the investment of public funds, and have the organizational skills to manage a complex project. Invest the time to get it right, and it will strengthen your entire application.