Combining Policies to Reduce the Spread of Viral Misinformation Online

Data were collected as part of the Election Integrity Partnership. Instances of potential misinformation were flagged as tickets. These were reviewed and categorized as misinformation if they made false claims related to election integrity. Details on the collection methods of the EIP can be found in our Report. These tickets were grouped together into qualitatively similar incidents. For example, tickets regarding false narratives about the use of Benford's law to detect fraud in Wisconsin became an incident. For each incident, search terms and appropriate data-ranges were determined to query our database. 

Our full database consisted of all tweets matching an evolving set of keywords, collected in real time, using the Twitter API. To maintain user privacy, we are providing data segmented into events and aggregated into 5-minute blocks of time. This should be sufficient for replicating our findings (predicated on the aggregation and segmentation). In order to permit analysis under various user-removal conditions, we have provided multiple versions of this dataset with users removed according to the conditions evaluated in the manuscript. We encourage anyone with the need for more granular data or alternate conditions to reach out to the University of Washington Center for an Informed Public.