Studying News Content Engagement in the 2018 Italian Election

The next iCS Symposium session starts with Fabio Giglietto, presenting his team’s results on the use of social media in the March 2018 Italian election. The project’s aim was to comprehensively examine the role of social media during the election, focussing especially on social media audience engagement with the various media sources available.

The project drew first on data from Twitter, capturing all retweets of Italian parties’ and politicians’ posts and assessing the political leaning of the accounts contributing to this datasets. It then captured the tweets by the top 5,000 contributors to this dataset, to examine which news sources they had shared on the platform.

Further, in order to establish a dataset of relevant news articles, the project also captured all Italian political news stories during the six months leading up to the election from the GDELT database, from Google News, and from Twitter (in the latter case, this included any tweets including links as well as mentions of an Italian party or politician).

Finally, then, the project used the Facebook API to capture the reactions, comments, and shares for any of these story URLs on a regular basis, in order to observe the level of engagement via Facebook with these Italian news stories.

Next, the project adapted the Media Partisanship Attention Score (MPAS) developed for the U.S. context to the Italian multi-party political environment. The MPAS assesses, first, the partisanship of the Twitter users engaging with politicians and parties on the basis of which parties and politicians they retweet most frequently, and second, the media preferences of these partisan users by examining which news sites’ articles they share most often. From this, the MPAS in turn infers a partisanship rating for each news source.

The project adapted this one-dimensional, left-right MPAS by creating multiple partisanship scores for each of the ten parties in the Italian system, and thereby deriving a multidimensional Multi-Party Media Partisanship Attention Score (MP-MPAS) for each news source. This enables the assessment of each news source’s insularity: the extent to which it is used by only one partisan group, or shared by supporters of multiple parties. Each news source can then also be assigned to the party whose supporters shared it most enthusiastically.

The news sources used most frequently by the populist parties Lega and, especially, Movimento Cinque Stelle tend to show the greatest level of insularity; these news sources also tend to receive a substantial number of Facebook shares, while the less insular news sources receive more Facebook comments rather than shares.

It would have been interesting to further investigate the content of these comments, yet unfortunately the increasing limitations of the Facebook API make this impossible. This also makes it difficult to assess whether the substantial volume of comments for non-insular news sources is organic or driven by inauthentic actors. Such information may be available in the datasets promised by the problematic Social Science One initiative, if indeed they ever eventuate.