Research Data Management

Data Sharing has multiple benefits which include increasing the visibility of your research, facilitating new discoveries by other researchers, and meeting funding requirements.

Benefits of Data Sharing

• Increases the visibility of your research
• Increases citations of related articles
• Datasets can be independently cited
• Facilitates new discoveries 
• Accelerate the pace of research
• Reduces research waste
• Encourages research transparency and reproducibility
• Creates greater public trust and engagement with science
• Complies with funder and journal policies

The FAIR principles are a set of guiding principles that have been largely accepted by researchers and the data-sharing community. They are a set of guiding principles that help ensure research data is shared in such a way that enhances reproducibility and transparency. They go beyond the act of just sharing data, and ensure data has value once it is shared. The four principles are that data should be Findable, Accessible, Interoperable, and Reusable.

A Simple Definition

Findable

Data needs to be findable by researchers for it to be used. Datasets should

• Have a persistent identifier (DOI)
• Have rich and descriptive metadata
• Have metadata that is findable in an online repository

Accessible

Potential users of the data need to be able to access the data and/or information about the data.

• Datasets should have DOIs
• Metadata should be publicly accessible even if the data is not
• Use a well-established, reputable repository to share you data

Interoperable

Data should be in recognized formats to make them easily reused by others and by different programs.

• Use open file formats such as .txt, .pdf, or .csv
  • Even common file formats like .docs or .xlsx may not function 5 or 10 years in the future
• Follow established metadata standards set by the data repository
• Use controlled vocabulary terms (such as MeSH) to describe your data when possible

Reusable

Data that is shared serves no purpose if it cannot readily be reused for future research.

• Data should have clear documentation that makes it understandable and usable by others. This can include things like README files, data dictionaries, file naming conventions ,etc
• Data should have clear licensing and copyright information attached to it, often dictated by repository choice

Types of Data Repositories

Institutional repositories are hosted by academic institutions. Data is accepted for all disciplines and are designed to ensure long-term preservation. Researchers outside the institution are unaware of their existence, making data less discoverable.  

Disciplinary repositories are curated by domain experts and have domain-specific metadata to allow for greater discovery. The NIH hosts many such repositories for health science topics.

Generalist repositories accept data regardless of field and are designed for wide discoverability across research domains.

Generally speaking, disciplinary and generalist repositories are best for sharing your data. Reach out to a librarian if you would like help identifying what repository is a good fit for your research.

Many research funders, journals, and publishers require data sharing as a condition of receiving research funding or publication. See the resources below for help identifying if funders have data-sharing policies. The best place to find information on a journal or publisher's policy is on their website. If you need help complying with a data-sharing policy, reach out to your liaison librarian for assistance.

Not all data can be shared freely. Sometimes legal, technical, or ethical concerns may limit or prevent data sharing. Common situations include

• Protecting personal health information
• Legal data protections such has HIPAA and FERPA
• Ethical concerns around sharing data related to children, prisoners, or vulnerable groups

When data cannot be shared freely, common approaches include

• Only making data available upon request
• Placing data in controlled access repositories
• Only sharing de-identified data