The European election through the lens of Twitter Journées d’études « Réseaux et TIC ; réseaux et innovation » Labex SMS Nikos Smyrnaios, LERASS
The European election through the lens of Twitter
Journées d’études « Réseaux et TIC ; réseaux et innovation »Labex SMS
Nikos Smyrnaios, LERASS
Research questions
Coming from journalism and media studies
Interested in how the online public sphere functions
Can network analysis help understanding political debate online ?
Which indicators better reveal significant political « relations » established on Twitter?
What is the structure of online political debate in a multilingual environment such as the European Union ?
ContextThe European election of May 25 2014 was a political
event of great magnitude:
1. The economic and social crisis raised political stakes especially in Southern Europe
2. First time that European parties had declared candidates for the presidency of the EU Commission
3. First time that a candidate debate on TV was organized and broadcast throughout Europe
4. Twitter was implemented massively by EU institutions as a tool of citizen participation
5. Success of extreme right and radical left parties
Method1. Collection of tweets including two different hashtags
related to two distinct events in May 2014. Data was collected with DMI Twitter Analytics (digitalmethodsinitiative/dmi-tcat on GitHub)
- #TellEurope: livetweeting of TV debate between the 5 candidates to the presidency of the Commission (30k tweets by 20K different users on May 15-16)
- #EP2014: livetweeting election day and commenting results (346k tweets by 138k different users on May 25-27)
2. Investigation on the political and electoral context
Method3. Production of two network graphs with Gephi using the Open Ord algorithm in order to distinguish clusters
- Dots = user accounts, lines = interactions between users
- Topology of the graph depends on interaction intensity
- Dot size depends on incoming mentions and RTs
- Color of dots depends on clustering
4. Manual classification of dots/user accounts according to nature (media, activist etc.), political position & nationality => comparison of information from the graph with the political context
#TellEurope#TellEurope
#TellEurope- Political differences and affinities reflected in
the structure of the network- Particularity of the cluster around Alexis Tsipras- While 4 clusters (Schulz, Keller, Verhofstadt,
Juncker) are relatively close and connected strongly to each other, Tsipras’ cluster interacts only with those of Schulz and Keller
- Simultaneously it strongly associates with another sizable cluster of Greek users
- Southern Europe (Greece, Italy, France & Spain) more present than the rest
- Greece organized campaign on social media in favor of Tsipras because of national stakes
#TellEurope #EP2014
#EP2014- Clusters mainly organized by country- Some strong inter-cluster relations by political
affinity e.g. Alexis Tsipras & Jean-Luc Mélenchon
- Top countries again from Southern Europe (France, Italy Spain) + GB, Germany, Nethelands & Belgium
- Transnational & national media play a central role in the discussion
- Mix of national and European politics- The most cited official accounts: Juncker, Renzi,
Mélenchon, Schulz, Le Pen, Grillo, Tsipras, Farage, Hollande, Keller and Verhofstadt
Network analysis shows protagonists, affinities, specific cluster dynamics that reflect realworld politics
Political debate on Twitter structured by political homophily but what is more important is language
Communication actions such as RT and mentions are more pertinent in such analysis than following/follower
Twitter and SNS begin to form a public arena for Europeans who don’t have common media
Shortcomings: hashtag dependency (e.g. Ireland)Representative of over-politicized segments and media
but not of the public opinion as a wholeTool and data opacity
Just a point of departure for further qualitative research
Both case studies available on http://ephemeron.eu
Merci !
@smykos