Archive and share data
On this page
- What is a data management plan
- Funder requirements
- Data management plan tools & examples
- Find data
- Collect data ethically
- File formats
- File naming, organization, versioning
- Document & describe
- Storage & backup
- Analyze & visualize data
- Prepare data for archiving, sharing
- Where to share data
- Data licenses
- Cite data
Archive and share data
Backing up your data is different from archiving. Backups can be discarded after a certain amount of time. Archiving is used to preserve a file as-is at the end of a project and acts as a static, final record. (Source: Oregon State University).
What data to keep?
Data can be archived and preserved locally or shared in a public data repository. Note that archiving can be costly and there may not be enough space to archive everything. Researchers should carefully identify which data to preserve. Consider the following:
- Does the data support published research?
- Are the data likely to be reused?
- Are the data unique or historically significant?
- Are there funder or institutional requirements?
- Are the data difficult to reproduce?
Examples of data that should be kept by discipline (from Stanford University).
Best practices when preparing data for archiving and sharing
|File formats||Choose long-term storage file-formats, preferably non-proprietary, to overcome software obsolescence. More information|
|Documentation||Add it alongside your data to make it understandable, reusable. More information|
|Ownership and privacy||If sharing data, make sure that:|
|Data integrity||If keeping a local copy, avoid bit rot through refreshment (copy data on a new drive every 2-5 years) and replication (maintain three copies of the data, on two forms of storage with one in an external location).|
Preparing sensitive data for sharing
Some data cannot be shared for legal or ethical reasons. However, if sharing the dataset is required, ensure that this has been stated in consent forms and cleared with the Research Ethics Unit. More information.
It may also be possible to retain multiple versions of the data: one for public release that has been de-identified, and one that is available on a highly restricted basis.
De-identification is the process used to remove identifying data. Identifiers can be direct, which point directly to an individual, or indirect, which point to an individual when combined with other data (see examples from Stanford University)
Below are two methods of data de-identification, with their benefits and drawbacks. (Source: UBC)
- Anonymization (removing identifiers altogether)
- Pseudonymization (replacing identifiers with pseudonyms or other identifiers)
- Anonymisation: managing data protection risk code of practice (UK Information Commissioner's Office).
- Anonymisation (UK Data Service)
Where to share data
There are many ways that researchers can share their data. These include:
- Depositing in a discipline-specific data repository
- Depositing in a general purpose repository
- Depositing in an institutional or recommended repository
- Publishing a data paper
Criteria in selecting a data repository
Source: University of Iowa
- FAIR Principles: FAIR means that data publishing platforms should enable data to be Findable, Accessible, Interoperable, and Re-usable. The FORCE11 FAIR Principles (simplified here) are:
- To be Findable any Data Object should be uniquely and persistently identifiable (have an identifer, such as a DOI)
- Data is Accessible in that it can be always obtained by machines and humans, upon authorization, through a well-defined protocol
- Data Objects are Interoperable (i.e. interpretable by a computer, so that they can be automatically combined with other data) if metadata and data use community agreed formats, language, vocabularies, and standards.
- Data Objects are Re-Usable if the above are met, if the data can be automatically linked or integrated with other data sources, with proper citation of the source, and have a clear machine-readable licence.
- Cost: Is there a cost to depositing data? Is it ongoing? Are these costs budgeted for?
- Discoverability: Are there adequate metadata fields to describe your data? Is the repository indexed by Google?
- Persistent identifiers: Does the repository register your data to create a persistent identifier (eg. a DOI)? These are necessary for citing your data.
- Policies and licenses: Are data use agreements and/or licensing (Creative Commons) clearly presented, to allow depositors to state explicitly up front what uses they would be willing to allow?
- Scholarly impact: Does it track data citation or download?
- Certification: It is possible for repositories to get certification (eg. CoreTrust Seal of Approval) which indicates how well they preserve digital content. Although good to have, note that very few repositories have achieved certification.
Discipline-specific data repository
These accept datasets related to either a specific discipline (e.g. genomics) or a broad subject-area (e.g. social sciences). Some repositories allow for self-archival and will provide limited or no curation service; others, like ICPSR, will provide in-depth curation services to subscribing institutions (Concordia is an ICPSR member) provided that the data fits within their collection development policy.
- Search for a disciplinary data repository:
If a discipline specific repository is not available, general-purpose repositories are the next best option. They typically accept a wide range of data types, and are suitable for cross-disciplinary data. Below are some examples:
|Concordia University's Dataverse||
Institutional or recommended repository
Depositing in discipline-specific or general-purpose repositories is encouraged, as they are generally better suited for data curation and dissemination. However, if there is no suitable discipline-specific repository for your dataset, consider using Spectrum, Concordia University's institutional repository.
Alternatively, some journals are requiring that researchers make the data associated with their papers publicly available to facilitate verification and replication of results. If there is a cost to depositing data, it is covered either by the submitter or by a sponsoring organization. See the following publishers for examples:
Data papers describe datasets, and do not typically include any interpretation or discussion. Data papers are published either in a journal’s “Data Papers” section, or in a journal that exclusively publishes data papers (for example, see Nature’s Scientific Data).
"The purpose of a data journal is to provide quick access to high-quality datasets that are of broad interest to the scientific community. They are intended to facilitate reuse of the dataset, which increases its original value and impact, and speeds the pace of research by avoiding unintentional duplication of effort."
Source: Oregon State University
A license defines what others may or may not do with your data.
When sharing your data in a repository, for a example, you can choose to apply a broad license to your data that allows anyone to do whatever they like with it. Alternatively, you can choose a narrower license that restricts use to strictly non-commercial activities and requires attribution to the data creator when it is used.
There are two primary sources for data licenses:
NOTE: Be sure you own the data! You can only publish data that you own or for which you've received permission to share. More information on data ownership and licenses.
FORCE11's Data Citation Principles indicate that data should be considered legitimate, citable products of research and should be accorded the same importance in the scholarly record as citations of other research objects, such as publications.
A data citation should try to include the same elements as a publication citation:
- Publication date
- Publisher (this is often the archive where it is housed.)
- Edition or version
- Resource type (eg. dataset or database)
- Access information (a URL or other persistent identifier)
Data Citation Generator:
If you have a dataset's DOI, use CrossCite's DOI Citation Formatter to create a data citation for you based on your selected style.
|APA||Cool, H. E. M., & Bell, M. (2011). Excavations at St Peter's Church, Barton-upon-Humber [Data set]. doi:10.5284/1000389|
(Footnote) H. E. M. Cool and Mark Bell, Excavations at St Peter’s Church, Barton-upon-Humber (accessed May 1, 2011), doi:10.5284/1000389.
(Bibliography) Cool, H. E. M., and Mark Bell. Excavations at St Peter’s Church, Barton-upon-Humber (accessed May 1, 2011). doi:10.5284/1000389