Finding and Using Data

Finding Data

Without logging in to Dataverse, users can browse Dataverse, search for dataverses, datasets, and files, view dataset descriptions and files for published datasets, and subset, analyze, and visualize data for published (restricted & not restricted) data files. To view an unpublished dataverse, dataset, or file, a user will need to be given permission from that dataverse’s administrator to access it.

A user can search within a specific dataverse for the dataverses, datasets, and files it contains by using the search bar and facets displayed on that dataverse’s page.

Browsing Dataverse

In Dataverse, browsing is the default view when a user hasn’t begun a search on the root dataverse page or on a specific dataverse’s page. When browsing, only dataverses and datasets appear in the results list and the results can be sorted by Name (A-Z or Z-A) and by Newest or Oldest. You can toggle the “Files” facet on the left to include files in the results list.

Using Data

View Dataverses + Datasets

After performing a search and finding the dataverse or dataset you are looking for, click on the name of the dataverse or dataset or on the thumbnail image to be taken to the page for that dataverse or dataset. Once on a dataverse page, you can view the dataverses, datasets, and files within that dataverse.

Once on a dataset page, you will see the title, citation, description, and several other fields, as well as a button to email the dataset contact and a button to share the dataset on social media. Below that information, the files, metadata, terms of use, and version information for the dataset are available.

View Files

Files in Dataverse each have their own landing page that can be reached through the search results or through the Files table on their parent dataset’s page. The dataset page and file page offer much the same functionality in terms of viewing and editing files, with a few small exceptions.

  • In installations that have enabled support for persistent identifers (PIDs) at the file level, the file page includes the file’s DOI or handle, which can be found in the file citation and also under the Metadata tab.
  • Previewers for several common file types are available and can be added by installation administrators.
  • The file page’s Versions tab gives you a version history that is more focused on the individual file rather than the dataset as a whole.

File Search within Datasets

Datasets containing multiple files offer a file search function. On the Dataset page, under the Files tab, you’ll see a search bar you can use to locate an individual file. It searches within the filename and file description. Performing a search will filter the file table to list only files matching your search. After you perform a search, if you’d like to return to the full list of files, just perform an empty search.

Under the search bar, you’ll see file search facets you can use to filter the dataset’s files by file type, access level, and file tags (see the example below).

image-file-search-facets

(To provide these search facets, we rely on the Solr search engine. Only the latest published version and any draft version of each dataset are indexed in Solr. Because of that, facets cannot be offered for older versions of a dataset.)

Tree View

Files can be organized in one or more folders (directories) within a dataset. If the folder structure is defined, the Dataset Page will present an option for switching between the traditional table view, and the tree-like view showing folder and file hierarchy, as in the example below:

image-file-tree-view

Cite Data

You can find the citation for the dataset at the top of the dataset page in a blue box. Additionally, there is a Cite Data button that offers the option to download the citation as EndNote XML, RIS Format, or BibTeX Format.

In installations that have added file-level citations, you can find and download the file’s citation on the file page in a similar manner.

Download Files

Within the Files tab on a dataset page, you can download the files in that dataset. To download more than one file at a time, select the files you would like to download and then click the Download button above the files. The selected files will download in .zip format that preserves any folder structure that the dataset owner had set up.

You may also download a file from its file page by clicking the Download button in the upper right corner of the page, or by Downloading via URL under the Metadata tab on the lower half of the page.

Tabular data files offer additional options: You can explore using any data exploration or visualization External Tools (if they have been enabled) by clicking the Explore button, or choose from a number of tabular-data-specific download options available as a dropdown under the Download button.

Tabular Data

Ingested files can be downloaded in several different ways.

  • The default option is to download a tab-separated-value file which is an easy and free standard to use.
  • The original file, which may be in a proprietary format which requires special software
  • Rdata format if the instalation has configured this
  • The variable metadata for the file in DDI format
  • A subset of the columns of the data

Downloading via URL

Dataverse displays a plaintext URL for the location of the file under the Metadata tab on the file page. This Download URL can be used to directly access the file via API (or in a web browser, if needed). When downloading larger files, in order to ensure a reliable, resumable download, we recommend using GNU Wget in a command line terminal or using a download manager software of your choice.

Certain files do not provide Download URLs for technical reasons: those that are restricted, have terms of use associated with them, or are part of a dataverse with a guestbook enabled.

Downloading a Dataverse Package via URL

Dataverse Packages are typically used to represent extremely large files or bundles containing a large number of files. Dataverse Packages are often too large to be reliably downloaded using a web browser. When you click to download a Dataverse Package, instead of automatically initiating the download in your web browser, Dataverse displays a plaintext URL for the location of the file. To ensure a reliable, resumable download, we recommend using GNU Wget in a command line terminal or using a download manager software of your choice. If you try to simply paste the URL into your web browser then the download may overwhelm your browser, resulting in an interrupted, timed out, or otherwise failed download.

Downloading a Dataverse Package via rsync

rsync is typically used for synchronizing files and directories between two different systems. Some Dataverse installations allow downloads using rsync, to facilitate large file transfers in a reliable and secure manner.

rsync-enabled Dataverse installations offer a new file download process that differs from traditional browser-based downloading. Instead of multiple files, each dataset uploaded via rsync contains a single “Dataverse Package”. When you download this package you will receive a folder that contains all files from the dataset, arranged in the exact folder structure in which they were originally uploaded.

In a dataset containing a Dataverse Package, the information to download and/or access is in two places. You can find it on the dataset page under the Files tab, and on the file page under the Data Access tab. If the data is locally available to you (on a shared drive, for example) you will find the folder path to access the data locally. To download, use one of the rsync commands provided. There may be multiple commands, each corresponding to a different mirror that hosts the Dataverse Package. Go outside your browser and open a terminal (AKA command line) window on your computer. Use the terminal to run the command that corresponds with the mirror of your choice. It’s usually best to choose the mirror that is geographically closest to you. Running this command will initiate the download process.

After you’ve downloaded the Dataverse Package, you may want to double-check that your download went perfectly. Under Verify Data, you’ll find a command that you can run in your terminal that will initiate a checksum to ensure that the data you downloaded matches the data in Dataverse precisely. This way, you can ensure the integrity of the data you’re working with.

Explore Data

Please see the Data Exploration Guide to get started.