Junk news during the EU parliamentary elections: Lessons from a seven-language study of Twitter and Facebook

For this study, we collected 584,062 tweets pertaining to the European parliamentary elections from 187,743 unique users between 5 April and 20 April using election-related hashtags in English, Catalan, French, German, Italian, Polish, Spanish, and Swedish. From this sample, we extracted 137,658 tweets containing a URL link, which pointed to a total of 5,774 unique domains.The set of election-related hashtags (see Table 1) was compiled by a team of nine coders with expert knowledge of the sociopolitical landscape of the countries they were assigned to. Prior to launching the data collection, the list of hashtags was refined in a trial run, which revealed the most frequently used election-related hashtags, and the list was revised accordingly. We split up the Twitter data based on hashtag use, with the language of a hashtag acting as proxy for the ad hoc public discussion forming around the European parliamentary elections. Therefore, a source shared in a tweet containing at least one Italian election hashtag (eg,# elezionieuropee) would be classified as part of the Italian language sphere. Using this technique, we found a small but not negligible number of sources associated with more than one hashtag, which were accounted for in multiple language spheres (980 sources, or 13.1% of total traffic).