Skip to main content

Facebook opens up ‘overwhelming data set’ for election research

Facebook opens up ‘overwhelming data set’ for election research

/

Part of a broader effort to study social media’s impact on elections

Share this story

Illustration by Alex Castro / The Verge

Researchers will soon have a chance to study every link shared on Facebook, thanks to a new data set released through a research partnership with Social Science One. Announced earlier this year, the partnership brings together independent academics with data from Facebook and funding from independent foundations, hoping to provide new insight into the impact of social media on elections.

The first data set shared under the new initiative will be a massive and continuously updated database of all public links shared on Facebook, starting in January 2017. The data set includes information on how often a given link was viewed, whether it spread through person-to-person sharing or group pages, and how often it was flagged as spam, hate speech, or misinformation.

“It’s an overwhelming data set”

The data set also includes information on whether a given link was coded as “hard news” by Facebook’s open graph and when and how it was vetted by Facebook’s internal fact-checkers. That information could be crucial as researchers look into the effectiveness of Facebook efforts to slow the spread of misinformation on its network.

The data will be updated continuously, and is expected to include 2 million new and unique URLs shared across 300 million posts each week. “It’s an overwhelming data set for people interested in social media and democracy,” said commission co-chair Nathaniel Persily.

The new data will only be available to specific research proposals, and only after they pass through anonymous peer review and IRB approval. Facebook has also pledged to not review findings prior to publication.

Facebook has drawn broad criticism for its role in the 2016 election, which saw Russian propaganda widely shared and often promoted as advertisements in violation of federal election law. Facebook has also faced blowback over improperly sharing user data with third parties like Cambridge Analytica, much of which was collected in the guise of social science research. On Tuesday, a UK privacy regulator hit Facebook with its maximum possible fine in response to related privacy violations.

A public request for proposals has been released inviting social scientists to pitch research projects using the data, which the commission believes will jump start a new wave of research into how Facebook is shaping our political landscape. “Social scientists have the tools to find this out,” said Harvard professor Gary King, who co-chairs the new commission, “but we need the data, or no one will ever really know.”