Data repositories allow creators to deposit datasets to preserve and make the data available to others. A repository will provide a record for each dataset that includes descriptive information about the data (metadata). This information enables others to understand what the dataset is about and if it’s relevant to them.

Repositories can make data available under different access conditions. Some repositories are open access, meaning that all data stored there can be freely accessed by anyone. Other repositories require people to meet specific conditions, such as proof of approval for the proposed research from an ethics board, before access to data will be granted. Some repositories will have both openly-accessible datasets as well as datasets with access conditions. Repositories may also allow metadata only records to be published.

Discipline-specific data repositories contain data relating to a specific subject or field of research. When looking for data from a specific research area, discipline-specific repositories are a great place to start your search. Use the Registry of Research Data Repositories to find a data repository for your field of research.

General data repositories accept data from any discipline. If you’re looking for data resulting from multi-disciplinary research or your research area doesn’t have any discipline-specific repositories, then a general repository, such as Zenodo, figshare or Dryad, would be a good place to search.

Institutional data repositories contain data produced by researchers working at a specific institution. If you’re looking for data from a particular researcher or research group, and you know what university or organisation they work for, then it can be worthwhile to search the organisation’s institutional repository. The University of Sydney’s institutional repository contains datasets, publications and theses produced by researchers at the University.

Data portals and registries are useful tools to help you find datasets stored in repositories. Data portals allow you to search across multiple data repositories from a single search interface. They can be a great place to start searching if there are a large number of discipline-specific repositories in your research area. Data registries provide lists or catalogues of datasets that are stored elsewhere. Descriptive information about the datasets, usually in the form of metadata only records, is provided by the registry, along with a link to where the data is stored. Researchers may choose to publish metadata only records in several registries or repositories to ensure their data is as findable as possible.

Many government and other organisations choose to make data that they have collected or produced available for others to reuse. Government and NGO datasets are often well described and high quality, making them easy to reuse. Data containing sensitive information, such as Australian Bureau of Statistics microdata, may require you to apply or meet certain conditions before you can be granted access. Some data providers charge a fee or require a subscription to access data. The Library subscribes to some data sources, making the data available to University of Sydney staff and students.

Some researchers choose to make their data available on personal, research group, or project websites. This is an ad hoc form of data sharing, resulting in variable data and metadata quality, which can make some datasets difficult to find and reuse. Long term access to data isn’t guaranteed, as it relies on the continuing investment of time and resources by individual researchers. As there is no central authority for data made available on websites, internet search engines are the best way of finding this data.

Increasingly, data access statements are being required in research publications to let readers know where the data underlying the research can be found. Ideally, this will point to data published in a repository, however, some journals allow data to be made available as part of the supplementary material to the article. This means that access to the data is limited to people or institutions with access to the article, generally through a paid subscription to the journal.

Data journals can be good sources of well-described datasets. These journals provide a platform for researchers to publish their data that mirrors how traditional journal articles are published, including the peer review process. A data paper will contain the information necessary to understand and reuse the data, but generally won’t include any findings based on the data. The paper will include the dataset itself, or a link to it. Data journals can be discipline-specific, such as Geoscience Data Journal, or general or multidisciplinary, such as Scientific Data. Search for data journals as you would any other academic journal, using the Library’s search tools, and internet search engines.