This post features a working paper that the Illuminating team presented at the 2020 SBP-BRiMS Conference. We describe and compare two computational techniques that we have used to classify Facebook and Twitter posts by message type.
In anticipation of a contentious presidential election, Facebook announced in September 2020 that it would make two changes to its treatment of ads about “social issues, elections or politics” in the US. This post provides an overview of these changes.
This report summarizes the challenges and concerns that the Illuminating project’s team encountered when using Facebook’s Ad Library to conduct research on the 2020 U.S. presidential election. We provide examples of each problem and explain why hindering the transparency of the online campaigning ecosystem poses a threat to campaign advertising research by academics and journalists. Our report concludes with recommendations that would help make Facebook’s Ad Library more accessible and transparent to journalists, researchers, and the general public.