The most widespread type of data sharing is the collaboration that happens within research groups and projects as part of the day-to-day research process. It is also increasingly common to share data more widely as a form of publication, in addition to more conventional research papers and books.

Both of these types of sharing are supported by research data management practices such as Organising and describing data, and Data preservation, but require additional tools as well.

Collaboration

Transfers of large volumes of data between Imperial and external sites, as well as sharing project data with third parties, is available via Globus.

More research support services from ICT

If using third-party collaboration tools, make sure you do so safely, see Sharing sensitive data.

Publication

All articles published by members of Imperial must include a "data access statement" containing the following key pieces of information about the data underlying the article:
  • How the data can be accessed: where it can be downloaded from or who must be contacted to request access. This should always include either a web link (a DOI or other persistent identifier if possible) or in rare cases an email address.
  • What terms use of the data is subject to: what embargo is in place, whether a general licence applies to all users, or whether a data sharing agreement must be entered in to before access to the data is granted.

If you cannot share your data, you must still include a statement briefly explaining why: you should be aware of your options for sharing sensitive data.

Writing a data access statement

At a minimum, Imperial and the research councils expect data underlying published results to be made available. You can either do this directly (if you have the right to share the data yourself) or indicate where readers can obtain the data for themselves.

If possible, try to use several of the methods described below, to improve the chances of your data being found.

Check specific requirements of your funder

Public repository / archive / data centre

The most straightforward way to make data available is to deposit it in a public repository (also known as an archive or data centre). In many cases, this will also meet the requirement to archive your data for the long term.

Most data repositories will give you a Digital Object Identifier (DOI) for your dataset, enabling it to be cited and tracked like a regular publication.

If your data is too sensitive to share publicly, you could deposit some summary statistics or an anonymised version to advertise your availability to collaborate on the full dataset.

Finding a data repository / archive

A comprehensive collection of domain specific data centres are listed in Recommended data repositories.

Supplementary information

If the data is very small, some journals may accept it as supplementary information and host it themselves. However, this is not ideal as supplementary information can be difficult to discover and some publishers unintentionally corrupt it as part of their publishing process.

Data journal

There are now a growing number of "data journals": journals dedicated solely to publishing brief descriptions of data (data papers or data descriptors) that link to the dataset itself.

Some leading data journals include Scientific Data (Nature) and the Geoscience Data Journal (Wiley), but the list is growing daily so it's probably easiest to ask colleagues or search the web for data journals in your area.

Available on request

If none of the options above are possible (perhaps because your data is sensitive), you can offer to make the data available on request. This allows you, for example, to enter into a data sharing agreement that binds the user to comply the consent granted by the participants who provided the data.

Do not use an individual person's email address for this. People change jobs or retire, but it must still be possible for people to obtain the data, so use a shared email address for your research group or department.