Introduction
As the European Parliament elections approached in 2024, there were significant developments in the dynamics of the digital campaign landscape. With political, social, and economic challenges mounting, digital platforms emerged as the primary public opinion battlegrounds, prompting the EU to implement a comprehensive legal framework – including the Digital Services Act (DSA), the EU Code of Practice on Disinformation, and the EU AI Act – to address these issues. As the campaign unfolded, however, the effectiveness of these mechanisms was tested repeatedly as they were being implemented for the first time.
This report aims to provide a brief overview of the digital campaigns during the 2024 EP elections and of the broader discourse surrounding them, with a focus on the level of toxicity in discussions and the main narratives driving them. The analysis, based on work by researchers from eight key EU Member States, provides insights into these trends, using data from Facebook and Instagram for the period between 1 January and 29 May.
In this research toxicity can be understood as “rude, disrespectful or unreasonable” content “that is likely to make someone leave the discussion”. Please refer to our methodology for further information.
Key findings
Toxicity in public discourse
- The analysis identified low levels of toxicity in online political discourse. Overall, only 140 posts, or 0.46 per cent of all posts related to the EP elections that were analysed, were classified as toxic, with this level varying across the countries. These findings can be interpreted positively, suggesting that this restraint in the discourse around the EP elections can be attributed to fear of regulatory repercussions, indicating that the policies in place to curb toxic online behavior may be working as intended.
- At the same time, engagement with toxic content frequently exceeded that of engagement with non-toxic content, in terms of interactions, shares, comments, and likes.
- Hungary and Poland stood out with highest absolute numbers of toxic posts.
- Toxic content generally revolved around personal attacks on EP candidates and MEPs, and criticism of the European Union, Israel, and Russia.
Narratives
- Political campaigning differed significantly across the countries, with the conflicts in Ukraine and Gaza, and agriculture and farmer’s protests being some of the few common topics. Most discussions focused on different locally relevant issues.
- Within the overall discourse examined, campaigning for the EP elections and attitudes towards the EU itself represented the most common topics by a wide margin, followed by immigration, current conflicts, and foreign policy.
- Sentiment in the discourse regarding the elections and the EU was generally positive, whereas that in posts related to immigration, current conflicts, and foreign policy was more polarising.
Methodology
For this analysis, we used a dataset comprising 30, 625 posts collected from Instagram and Facebook via CrowdTangle, covering the period between 1 January and 29 May. We monitored the activity of over 700 entities from eight key EU Member States: France, Germany, Hungary, Italy, Poland, Romania, Spain, and Sweden. The sample includes posts from leading national politicians from across the political spectrum, EP candidates, non-institutional political influencers, media outlets, prominent political journalists, and civil society organisations (CSOs). The full study, which will be published in July, will reflect data collected through 15 June.
To examine the level of toxicity in public discourse, we employed Jigsaw’s Perspective API, combined with a qualitative assessment of the output. Per the API documentation, toxicity can be understood as “rude, disrespectful or unreasonable” content “that is likely to make someone leave the discussion”. We explored the distribution of toxic content across different entity types, its prevalence over time, and resulting engagement levels.
For the identification of dominant narratives, we relied on BERTopic – a topic modeling technique that leverages BERT embeddings. Furthermore, we analysed how each topic developed over time, the sentiment surrounding it, and which entities were the main drivers of the discourse on each topic. Understanding the main narratives that dominated the online space is crucial for comprehending the broader political climate during the elections. By identifying key topics, opinion leaders, and their messaging strategies, we can better understand how public opinion was shaped and swayed in the digital arena. This analysis presents only a part of our findings.
Low levels of toxicity marked public discourse surrounding the EP elections
In none of the countries studied did the percentage of toxic posts exceed one per cent of the total posts. The highest absolute number of toxic posts identified was in Hungary, where 47 posts were labeled as toxic, or just over .5 per cent of the 8, 744 posts analysed. The lowest level was in Italy, where only 2 out of the 2, 327 posts analysed were labeled as toxic (Graph 1)
In Hungary, toxic posts ranged from criticisms of the EU and its perceived dominance over Hungary to criticisms of Fidesz, the governing party, Prime Minister Viktor Orbán, and various EP candidates, with political candidates generating most of the toxic posts.
Toxic post criticising the EU
“’274c We will not let Brussels brainwash Hungarian children with gender propaganda! ‘2757 There is no place for LGBTQ propaganda in Hungarian schools! #fidelitas”
In Poland, which had the second highest absolute number of toxic posts, these were frequently aimed at the EU, its climate policies, and the promotion of liberal values, as well as at various MEPs and EP candidates. As in Hungary, a majority of toxic posts came from political entities, and many of these were from the official Facebook page of the far-right Confederation (Konfederacja) party alliance. The hashtag #WeWantToLiveNormally, meaning in opposition to perceived EU-imposed values, was often associated with toxic posts.
Toxic post promoting Confederation and criticising the EU and the PiS opposition party
“Stanislaw Tyszka vs. MEP Anna Zalewska, or #ploughed ‘d83d’de0e This is how you expose the hypocrisy of PiS, which is now trying to tell people that it is against von der Leyen and the Green Deal, meanwhile…. ALL PiS MEPs voted for von der Leyen 🙂 Shame and disgrace. That’s why we’re voting for Confederation on 9 June, to really stop these climate crazies, and not just talk about it and lie to the Poles! #WeWantToLiveNormally”
In Germany, political entities were also the main contributors of toxic posts, with this discourse being engaged in by politicians and parties across the political spectrum, including VOLT, AfD, and SPD. Toxic posts ranged from criticisms of the rise of the right-wing in Europe to criticisms of the EU and its policies on migration, as well as of liberal values.
Romania is one of the few examples where media outlets were the main contributors to toxic discourse, with the outlet 60m.ro being the most frequent offender.
Toxic post from 60m.ro
“’d83d’dd25Today, Ursula offered Grindeanu a basque. The next time TAROM is one step into the pit they give him a wheelchair, then bankruptcy will be recommended. Has the West airlines and for the poor in the East. Good thing we are ruled by fools and servants.”
In Spain, we identified toxic discourse only among the posts of by far-left politician Manu Pineda, most of which focused on the conflict in Gaza and criticisms of Israel and its allies.
Toxic post from politician Manu Pineda
“There are no words to describe such sadism and inhumanity. Israel in a criminal, terrorist entity, sponsored by #USA. The #EU has become a necessary accomplice to the crimes committed by these Zionist Nazis against the Palestinian people in #Gaza…[quote continues].”
Despite its relatively low prevalence, toxic content generated more engagement on average, including more interactions, shares, and comments, which highlights the troubling reality that provocative and negative posts often garner more attention and interaction than positive or neutral content (Graph 2). Non-toxic content drew more likes on average, but only by a small margin. These findings suggest that, while the overall discourse remains largely respectful, the more incendiary and divisive messages have a disproportionate impact on public engagement, potentially skewing perceptions and amplifying polarisation in political conversations.
The analysis of public discourse ahead of the EP elections reveals that, despite the presence of toxic posts, the overall toxicity remained low across all countries studied. This suggests a relatively respectful public dialogue, even within contentious political climates. The instances of toxicity that did occur, however, highlight ongoing tensions and polarisation, particularly surrounding EU policies, national politics, and various social issues.
Hungary and Poland stood out with higher absolute numbers of toxic posts, often driven by political entities leveraging critical and provocative rhetoric against the EU and other political opponents. In Germany, the diversity of contributors to toxic discourse, spanning the political spectrum, underscores a broader discontent with various EU policies and internal political dynamics. Romania’s media-driven toxicity and Spain’s isolated political instances reflect the influence of localised issues and international conflicts on public sentiment.
Overall, while toxicity did not dominate discourse, its presence and reach serve as a reminder of the underlying divisions and the need for continued efforts to foster more constructive and respectful political conversations.
Given these findings, we can conclude that the DSA, the Code of Practice on Disinformation, and various efforts to reduce the risks of social media discourse with regard to elections seem to be having an impact. In the future, educational campaigns to improve digital literacy and initiatives to encourage positive engagement across political divides are essential. By leveraging the DSA and other policy tools, the EU can address the root causes of toxic discourse and foster a more inclusive and respectful political dialogue across its member states.
The main narratives in EP discussions
In our analysis of the main narratives shaping discussions ahead of the EP elections, we distinguish between all content in general and that solely from political entities. We made this distinction to identify any meaningful differences between these two.
In the content posted by political entities, we found little overlap in the dominant topics within countries. The most frequently occurring topics across countries include issues of national/European sovereignty and identity, campaigning for the EP elections (in Poland, this topic made up 50 per cent of all discourse), the climate, agriculture/farmer’s protests, the war in Ukraine, and the conflict in Gaza. The top 10 topics discussed by politicians across all countries and their prevalence are shown below (Graph 3). A small cluster of posts on civil rights and freedoms represented the smallest topic in the top 10 topics discussed.
In Sweden, the main topics included campaigning for the EP elections (22 per cent), the climate (16 per cent), and the war in Ukraine (9 per cent). In Poland, the main topics included the campaigning for the elections (52 per cent), the European Green Deal (9 per cent), and the government of Prime Minister Donald Tusk (6 per cent). In Romania, most online discourse directly or indirectly related to Romania within the EU and its future, as well as neighbouring Moldova’s accession to the EU. In Italy, the EP campaigning (32 per cent) and discourse about the government of Prime Minister Georgia Meloni (26 per cent) together accounted for more than half of all discourse surrounding the elections, with topics like the war in Ukraine, agriculture, and immigration making up less than 5 per cent each of the total discourse. In Germany, no one topic dominated discourse, with topics such as agriculture, NATO, migration, the war in Ukraine, violence against women, and the role of the CDU/CSU alliance in Europe being among the many issues covered by politicians. In France, discourse was similarly diverse, ranging from agriculture and free trade, immigration and asylum, to national sovereignty. In Spain, topics like the Catalan elections, the Koldo corruption scandal, and unemployment were among the most relevant topics.
A deep dive into EU attitudes, immigration, and war and foreign policy
In the discourse from all entities monitored, including political entities, media outlets, CSOs, and relevant political influencers, we looked for the prevalence of three topics: attitudes towards the EU (including EP elections/campaigning discourse), immigration, and war and foreign policy. Unsurprisingly, the EP elections and attitudes towards the EU dominated discussions in nearly all countries, with the exception of Spain, where war and foreign policy was the most discussed topic. In all other countries, except for Sweden, the elections and EU attitudes were the most discussed topic by a significant margin. The second most discussed topic was war and foreign policy, with immigration being prominent only in Poland, France, and Sweden (Graph 4).
Further, we analysed the sentiment of posts about these topics. Based on the average percentage of positive, negative, and neutral posts across all eight countries, we can see that discourse regarding the EP elections and EU attitudes was predominantly positive, with, on average, 58.6 per cent of conversations classified as positive, and only 25.9 per cent as negative. Discourse on war and foreign policy and on immigration was more polarised, leaning more towards positive than negative, but by only by a small margin in both cases (graph 5).
In addition to these topics, we further examined other trending topics in the overall discourse across each country. Apart from notable exceptions, however, such as the GDP and the European AI Act in Spain, or national identity grievances in France, most of the overall discourse in other countries was largely similar to the discourse seen in the political campaigning (Graph 3).
