Emotionalized Social Media Environments: How Alternative News Media and Populist Actors Drive Angry Reactions

ABSTRACT

This study employs a comparative analytical framework to enhance our understanding of the conducive opportunity structures that foster emotionally charged political discourse. We examined 175,539 Facebook posts characterized by variations in content (in terms of themes and populist rhetoric), authorship (including populist politicians, traditional news, and alternative media), and geographic context (Belgium, France, Germany, Switzerland, UK, US). We then analyzed 360,000 emotional responses from Facebook users to determine which posts create the most conducive conditions for eliciting angry emotions. Our key findings show that posts from alternative and hyperpartisan media, as well as those from populist politicians and parties, tend to elicit elevated levels of angry reactions. These posts often use anti-elitist and exclusionary language. This finding has significant implications, as the anger generated by such accounts can propagate incivility and polarization and facilitate the spread of ideologically driven misinformation. A particular case is Donald Trump, who, as a populist governing figure, manages to elicit positive emotions, including “love,” despite delivering seemingly antagonistic messages. To strengthen the robustness of our findings, we conducted a replication analysis with 67,620 Facebook posts from three of the six countries and examined two different time periods. This analysis confirmed the persistence of our findings over time. Our opportunity-structure framework offers valuable insights for designing targeted strategies to improve the quality of public discourse and promote informed and constructive political engagement in diverse societies.

KEYWORDS:

• Emotions
• social media
• alternative media
• political actors
• political communication
• comparative analysis

Introduction

Social networking sites, such as Facebook, have become important platforms for political discussions, as they allow users to show support for a political party and to make their opinions known. News media and political actors compete for the attention of social media users and try to reach them with emotional content (Eberl et al., 2020; Heiss et al., 2019). Research has established that emotions such as anger and anxiety can increase attention to political campaign communication (Gerstlé & Nai, 2019; Neuman et al., 2018). Populist politicians, in particular, seem to have mastered this strategy of emotionalization particularly successfully, thus increasing their reach and attracting new followers (Ernst, Esser, et al., 2019).

Emotional social media posts attract attention and provoke emotional user reactions, mainly if controversial issues are addressed (Eberl et al., 2020). For example, posts that include hostility toward out-groups receive significantly more “angry” reactions than others (Rathje et al., 2021). In the longer term, such emotional reactions can influence political attitudes, for example, when emotionalized blame attributions are shared by populist actors (Hameleers et al., 2016). Similarly, alternative and hyper-partisan news outlets publish emotionalized content to attract high user engagement, which can increase the virality and visibility of their content (Hiaeshutter-Rice & Weeks, 2021; Larsson, 2019). Emotional user reactions also serve as a sentiment barometer for the account holders (i.e., politicians and media outlets) who seek to build strong follower networks. Furthermore, they can influence other users, as they suggest a particular climate of opinion, and can lead to emotional spill-over effects (Bene et al., 2022; Heiss et al., 2019).

In this study, we combine an analysis of emotions with an opportunity structure approach that distinguishes contextual conditions on the level of countries, actor types, media types, and message types. We are interested in how various countries differ regarding emotional user reactions to political posts on Facebook and what kind of content provokes these reactions. How do types of political actors (i.e., populist vs. traditional politicians) and news media actors (i.e., hyper-partisan vs. traditional media) differ regarding those emotional user reactions? Finally, what kinds of posts drive such emotional reactions, particularly anger? Studying these conditions will provide a more differentiated understanding of the communicative strategies employed by political actors and news media to attract large audiences.

Further, contextual conditions might explain how strongly citizens from different countries are exposed to different emotions, such as anger while using social media. To answer these questions, we study emotional reactions to accounts of hyper-partisan news media and populist politicians and parties in six countries (Belgium, France, Germany, Switzerland, the United Kingdom, and the United States) and compare them to traditional news media and traditional political actors and parties. We argue that analyzing audience engagement is decisive in understanding whether and how different actors contribute to the emotionalization of political discussion, with important implications for the quality of public discussions in various countries (Mason, 2018).

Emotions in Online Political Discussions

Research is increasingly focusing on the role of emotions in political discussions on social media (Hameleers et al., 2016; Kim & Kim, 2019). Political communication on social media is characterized by emotions, primarily negative ones (Hasell & Weeks, 2016; Majó-Vázquez et al., 2020). This is relevant insofar as political attitudes consist of both cognitive and emotional components, and emotions can have a strong influence on the perception and evaluation of topics and events (Kühne et al., 2011) and influence the processing of information (Kühne & Schemer, 2015). Against this backdrop, it is crucial to understand which actors purposefully elicit emotional responses from online users and what content they use.

Moreover, recent research has shown that users are increasingly exposed to negative emotions in online debates (D’Errico & Paciello, 2018; Humprecht et al., 2020; Kim & Kim, 2019), which can both influence their perception of the topic and reinforce the affective polarization of societies (Anderson et al., 2014; Yarchi et al., 2020). Moreover, rude, disrespectful, or unreasonable comments can make people leave a discussion, undermining participation and inclusion in public discourse (J. W. Kim et al., 2021; Majó-Vázquez et al., 2020). For example, a multilingual study in 2020 found that comments evoking hostile or negative emotions accounted for 21% of overall conversations on the COVID-19 pandemic on Twitter (Majó-Vázquez et al., 2020).

In addition to commenting, users can show their emotional reactions on Facebook by clicking the “like” button or using one of the emoji icon buttons (“love,” “haha,” “wow,” “sad,” and “angry”). These mutually exclusive icons allow for a more nuanced reaction than just “liking” content. This form of engagement can serve as an emotional cue to other users, allowing news media and political actors to derive an opinion climate among their followers (Blassnig & Wirz, 2019). Researchers have cautioned that the targeted provocation of hostile emotions, as expressed in Facebook’s “angry” icon, leads to an emotionalized social media environment where spirals of aggression and in-group/out-group thinking unfold, ultimately driving the polarization of societies (D’Errico & Paciello, 2018; Humprecht et al., 2020).

We adopt a context-sensitive approach and theorize that emotional responses are not universal but are dependent on favorable opportunity structures. In comparative media use and effects research, opportunity structures are examined at four levels (Boomgaarden & Song, 2019): country characteristics, actor characteristics, media characteristics, and message characteristics. We thus use this logic to introduce our research questions and hypotheses.

Country Characteristics

Cultural and political contexts matter. Previous research has established that country differences exist regarding the amount of hostile emotions in user comments (Humprecht et al., 2020), which can be partly attributed to different discourse cultures (Hellmueller et al., 2021). For example, one study compared the United States to Germany and found that hostile comments and negative emotions were more prevalent in U.S. comments. In contrast, German comments used more of a neutral tone when responding to Facebook posts (Humprecht et al., 2020). In other words, although negative emotions in online discussions are prevalent in most countries, the phenomenon does not seem to be driven purely by technology and depends on contextual factors, such as different norms in political discussions (Freelon, 2013; Humprecht et al., 2022; Kalogeropoulos et al., 2017). Against this background, this study explores the following research question (RQ1): How do emotional user responses on social media, particularly Facebook, differ across countries?

Previous research has shown that spill-over effects of emotional online content on user responses occur and showed that sentiment as a content characteristic is a good predictor of audience emotional engagement (Blassnig & Wirz, 2019; Eberl et al., 2020). For example, negative news about salient issues such as war or climate change can elicit negative affective responses from audiences (Zollo et al., 2015). Correspondingly, news with positive framing can evoke positive emotions, such as hope or excitement (Lecheler, Bos, & Vliegenthart, 2015). In terms of online communication, negative online messages are more likely to be shared on both Facebook and Twitter, leading to higher visibility of this kind of content (Trilling et al., 2017). On Facebook, negative posts by political actors can lead to an increase in “shares” and comments, while “likes” show no clear relationship with the sentiment of a post (Bene, 2017; Heiss et al., 2019).

Actor Characteristics

Research into the communication strategies of populist actors has shown that populists deliberately provoke negative emotions, such as anger or fear, to reach their supporters (Skonieczny, 2018; Wirz, 2018). Such a deliberate provocation of emotions serves different purposes, such as creating a sense of urgency among their followers. By provoking hostile emotions such as anger, fear, and resentment, populist actors can make their followers feel that they are in a shared struggle and that they need to act quickly and decisively to defend themselves (Mudde, 2009). Another purpose is to exacerbate social and political divides. By provoking negative emotions, populists can deepen existing divisions and attract supporters who feel alienated from the “establishment” and the political system more broadly (Hameleers et al., 2016).

Moreover, populist actors frequently use emotional appeals to connect with their followers rather than relying on logical arguments or facts (Moffitt, 2018). By provoking hostile emotions, they can tap into their follower’s deepest fears and anxieties, making them more likely to follow their lead without questioning their claims or motives. Finally, by provoking anger or fear, populists can simplify complex issues and present themselves as the solution to these problems. This often involves identifying a common enemy or scapegoat (e.g., immigrants, elites, or other groups) as the source of the problems, which can further strengthen the bond between the populist leader and their supporters (Rhodes-Purdy et al., 2021). Therefore, researchers have argued that such populist communication strategies can lead to increased polarization and scapegoating, making it more difficult to find common ground and work together to address complex issues, such as to mitigate a pandemic (Ringe & Rennó, 2022).

Negativity and emotionality have thus been conceptualized as stylistic elements of populist communications, particularly on social media (Ernst, Blassnig, et al., 2019). Anger, in particular, leads to increased support for populist actors (Rico et al., 2017). A study of Italian politicians has shown that negative emotions are associated with populistic ideology and that emotional appeals by populists evoke more interactions and reactions, both negative and positive. “Love” reactions, for example, were elicited primarily by blaming the “dangerous others” (Martella & Bracciale, 2022).

Given the empirical evidence discussed above, we expect populist politicians (H1a) and populist parties (H1b) to elicit more angry reactions to their posts than traditional politicians and parties.

Media Characteristics

Like populist actors, alternative and hyper-partisan news media have been found to purposefully provoke negative emotions among their readers, particularly in social media (Tuomola & Wahl-Jorgensen, 2022). By blaming elites, such as the government or traditional news media, alternative and hyper-partisan news media trigger negative emotions among their readers and gain visibility on social media (Figenschou & Ihlebæk, 2019; Ihlebæk & Holter, 2021). In a study on negative emotions toward journalism in Korea, Shin et al. (2021) show that alternative media deliberately stir up hostile emotions against journalists to strengthen their community and create a common enemy image. Negative emotions are thus deliberately used to attack traditional media and increase reach and legitimize themselves. Similar strategies can be observed in Norway (Figenschou & Ihlebæk, 2019), Germany (Boberg et al., 2020), and India (Bhat & Chadha, 2020). Studies from Germany (Müller & Freudenthaler, 2022) and Norway (Ihlebæk & Holter, 2021) further show that the most successful alternative media outlets have a far-right orientation.

Based on the research discussed above, we suggest that alternative and hyper-partisan news media provoke higher levels of angry reactions with their posts than traditional news media (H2).

Message Characteristics

In addition to the source, the content-related characteristics of media or political messages, such as the topics they address, may affect emotional engagement. For example, research has shown that salient topics such as immigration are associated with higher levels of emotional responses (Lecheler et al., 2015). Previous research on news stories has shown that topics such as racism, war, and political conflict lead to more hostile and toxic comments than news topics such as science and technology or arts and culture (Salminen et al., 2020). In a study of Brazilian news websites and Facebook pages, Rossini (2022) shows that policy-related news stories and stories on international affairs or civil society received significantly more angry comments than other topics, such as general politics or celebrities. However, there is still little empirical evidence on the topics alternative media and populist politicians use to evoke emotions in their audiences, especially anger. In this study, we explore research question 2 (RQ 2): Which type of content of Facebook posts by alternative/hyper-partisan media or populist politicians (candidates/parties) evoke particularly high levels of angry emotions among users?

Data and Methods

We examined the posts and user engagement levels of 148 Facebook pages of different actors and outlets from April 14, 2020, to June 30, 2020, in six countries, namely Belgium, France, Germany, Switzerland, the UK, and the US. During this period, when lockdowns were still in place in most countries studied, a particularly large number of people mostly stayed at home and used Facebook to obtain information (Newman et al., 2021). At the time of the study, countries were affected by the pandemic to varying degrees. Confirmed infections per day were highest in France (172.70) and Belgium (109.39), followed by the US (89.55), the UK (63.22), Germany (60.20), and Switzerland (44.33).¹ The countries also differed in terms of the political situation at the time of the study. Due to the different political and health systems, crisis management varied, ranging from top-down approaches (e.g., France), top-down with operational bottom-up (Italy) to negotiated and coordinated top-down with significant bottom-up approaches in multi-party systems (e.g., Belgium and Germany) (Bouckaert et al., 2020). The U.S. exhibited a mix of top-down approaches, such as federal directives and vaccine distribution strategies, and bottom-up responses, characterized by varied state policies, local initiatives and grassroot community actions (OECD, 2022). In almost all countries studied, populist parties assumed an opposition function and criticized the measures imposed as too far-reaching (e.g., Switzerland, Germany, Belgium) or not far-reaching enough (France) (Bolleyer & Salát, 2021). To compare the potentially exceptional period under study, we replicated our study in three countries (BE, FR, US) for two periods before the pandemic (four weeks in January and February 2019 and 2020; see Appendix).

We collected data from a wide range of public accounts, including media outlets (alternative/hyper-partisan, public broadcasters, up-market, mass market, and tabloid news media), politicians, and political parties (government, opposition, populist; see Table 1). The selection of media accounts followed an audience-based approach in that we selected functional equivalents in terms of audience and reach (according to which Fox News, for example, is more similar to audience- and opinion-rich tabloids like Daily Mail and Bild than it is to niche papers in the alternative media sector; see Peck (2022).To ensure that we selected functional equivalents in all countries, we sampled the Facebook accounts based on 1) importance (rank of the actor or reach of the news outlet) and 2) the current literature (Engesser et al., 2016; Ernst, Esser, et al., 2019). In addition, we selected hyper-partisan/alternative news outlets based on data from the Digital News Report 2020 (Newman et al., 2020) and on their self-descriptions as “alternative” (e.g., on their websites). To reflect the diversity and heterogeneity of these outlets in the countries under study, we sampled five to eight Facebook accounts per country. We understand populism as a communication phenomenon that is expressed in content and style (de Vreese et al., 2018). Therefore, to sample functional equivalents of populist actors across different countries, we drew on studies that used communication content from an international set of actors to categorize populism. Ernst, Blassnig, et al. (2019) define core populist messages on the dimensions of anti-elitism, people-centrism, and preservation of sovereignty. Timbro (2019) uses similar dimensions and focuses on authoritarianism for categorization. Furthermore, we draw on data from the Global Populism Database, which contains over 1100 speeches by political actors. Various elements of populist communication were coded (with reference to text, tonality, quality of arguments, style, etc.), elements that can also be attributed to concepts such as popular centrism and anti-elitism. Although we could not identify a populist party in the United States based on our sources, we did identify a populist politician (i.e., Donald Trump). By applying a mixed-methods approach of automated and qualitative text analysis, we analyzed how various countries and actor groups differ in terms of emotional user reactions and what kind of content triggers anger on Facebook.

Table 1. Sample and number of posts.

		BE	n	CH	n	DE	n	FR	n	U.K.	n	U.S.	n
Alternative media		SCEPTR/PALNWS.be	662	Weltwoche	215	Junge Freiheit	624	Valeurs Actuelles	2878	Evolve Politics	71	Breitbart	4656
		Doorbraak	1231	Jan Walter	18	CompactOnline	444	Sputnik France	7293	Guido Fawkes	781	Epoch Times	4507
		De Wereld Morgen	231	ExpressZeitung	38	Tichy’s Einblick	953	FdeSouche	384	The Canary	902	CNS News	954
		Apache	225	ConvivaPlus	196	Watergate.TV	191	RT France online	3809	Double Down News	82	Daily Caller	4370
		Mondiaal Nieuws	614	INFOsperber	250	Epoch Times DE	1524	Basta	510	Epoch Times UK	482	Occupy Democrats	4689
				Daniele Ganser	55	RT Deutschland	2519			RT UK	1637
						Sputnik Deutschland	1304
Established News Media	Public broadcaster	VRTNWS	2433	SRF News	628	ARD News	1366	France 2	193	BBC News	2380	NPR	3625
	Up-market	De Standaard	2425	NZZ	1877	Süddeutsche Zeitung	2171	Le Figaro	3977	The Telegraph	4045	NY Times	4983
		De Morgen	2093	Tages-Anzeiger	1477	Frankfurter Allgemeine Zeitung	3361	Le Monde	4836	The Guardian	6046	Washington Post	3059
	Mass market/tabloid	HLN	3134	20 Minuten	1778	Der Spiegel	3216	L’Express	4157	Daily Mail	10640	Fox News	3622
		Nieuwsblad	2931	Watson	1121	Welt	1777	Le point	2008	Daily Express	5226	USA Today	3137
				Blick	1584	Bild	3765	Le Parisien	4813	Sun	8659	NY Daily News	5100
Politicians	Government and opposition politicians	Bart De Wever (N-VA)	41	Martin Candinas (CVP)	51	Markus Söder (CSU)	157	Emmanuel Macron (REM)	80	Boris Johnson (Tories);	184	Mike Pence (GOP)	110
		Theo Francken (N-VA)	227	Alain Beset (SP)	67	Jens Spahn (CDU)	122	Olivier Véran (REM)	99	Matt Hancock (Tories)	157	Andrew Cuomo (Dem)	46
		Maggie De Block (Open VLD)	72	Bastien Girod (Grüne)	16	Martin Schulz (SPD)	7	Jean-Luc Mélenchon (LFI)	191	Jeremy Corbyn (Labor)	156	Joe Biden (Dem)	585
		Conner Rousseau (Sp.a)	298	Cédric Wermuth (SP)	53	Christian Lindner (FDP)	103	Nicolas Bay (FN)	92	Sadiq Khan (Labor)	120	Bernie Sanders (Dem)	581
		Peter Mertens (PVDA)	172	Natalie Rickli (SVP)	14	Cem Özdemir (Grüne)	36	Olivier Faure (PS)	84	Rishi Sunak (Tories)	53	Nancy Pelosi (Dem)	159
		Dries Van langenhove (VB)	240			Sahra Wagenknecht (Linke)	45	Francois Bayrou (MODEM)	58	Dominic Raab (Tories)	25	Alexandria Ocasio-Cortez (Dem)	163
								Christian Jacob (LR)	38
								Yannick Jadot (EELV)	53
	Populist politician	Tom Van Grieken (VB)	229	Roger Koeppel (SVP)	346	Alice Weidel (AfD)	113	Marine Le Pen (RN/FN)	165	Nigel Farage (Brexit Party)	183	Donald Trump (GOP)	1284
Parties	Government and opposition parties	N-VA	103			CDU	192	REM	261	Conservative Party	347	Republican Party (GOP)	182
		Open VLD	184	CVP	96	SPD	123	MODEM	128	Labor Party	324	Democratic Party	199
		Cd&V	97	FDP	225	FDP	142	LR	134	Liberal Democrats	430
		Sp.a	287	GLP	59	Die Linke	127	PS	211	DUP	97
		groen	191	SP	113	Grüne	81			SNP	113
		PVDA	229	Grüne	138
	Populist party	Vlaams Belang	152	SVP	127	AfD	200	RN/FN	159	Brexit Party	71

The sample for Belgium reflects Flanders; the sample for Switzerland reflects the German-speaking part of Switzerland (based on the sampling criteria for reach).

Emotional Reactions

To create the dataset, we accessed public Facebook data (posts and comments) through the CrowdTangle API and the Facebook Graph API using the Facepager application (Jünger & Keyling, 2013) from April 14 to June 21, 2020. CrowdTangle is a Meta owned analytic platform that tracks posts published by verified public profiles and groups. The resulting dataset consisted of N = 175,539 Facebook posts (BE: n = 18,501; CH: n = 10,542; DE: n = 24,663; FR: n = 36,611; U.K: n = 43,1211; U.S: n = 46,011). The metadata included emotional reactions (love, haha, wow, sad, and angry) as well as numbers of likes, shares, and comments for each post. We compared emotional reactions across countries and actor groups. Prior to the analysis, we normalized the counts of emotions against the number of followers of each Facebook account to allow for comparability between accounts, some of which varied greatly in their number of followers. This approach was chosen to holistically capture the diverse emotional reactions across various account sizes and ensure that smaller accounts, which might have fewer responses to individual posts, were accurately represented. It is vital to note that while the numbers may not intuitively align with an absolute scale, they are indicative of the relative engagement efficiencies of posts within and across countries. Additionally, the normalization was based on the number of followers an account had at the time of our data sampling. Given platform constraints, this method offers the closest representation of potential post engagement. A high normalized number of emotion expressions indicates what proportion of an account’s followers have actively left emojis. This value has the advantage over the total count that the accounts can be compared directly, even though their followers (and thus the engagement potential) differ greatly. A normalized count of .7, for example, means that 70% of the followers have expressed emotions.

Posts

In accordance with our theoretical expectations regarding favorable opportunity structures, we were particularly interested in posts circulated by accounts of populist political actors and alternative/hyper-partisan news media that generated the highest levels of angry reactions. To analyze the data, we relied on relative measurements. First, we filtered for all posts with angry reactions and subsequently selected the highest quartile in terms of the normalized number of angry reactions (i.e., angry reactions divided by the number of followers). To pinpoint the content that elicits notably high angry reactions, we adopted a relational approach, focusing on the identification of “overused” words. These are terms that a specific actor or media outlet uses far more frequently than the reference group’s average. By blending ’corpus-based’ and ’corpus-driven’ techniques (Rayson, 2008), we spotlight essential words that can be compared across individual accounts. Our method involved contrasting a chosen post’s content against all other posts from the same actor category (e.g., traditional politicians vs. news media). For each word or symbol (referred to as ’tokens’), we assessed its observed versus expected appearance rate, focusing on the most anger-inducing posts. To refine our analysis, we removed common words, punctuation, single letters, numbers, and URLs. This process yielded a “weight” for each token, revealing its prominence in anger-evoking posts as either ’overused’ or ’underused.’ Importantly, our metric also integrated the absolute frequency of each token, adjusted logarithmically for scale. Through this method, we could discern which content elements – expressed as specific words – were most associated with heightened angry responses.

In a second step, we conducted a qualitative content analysis of the 10 most anger-provoking posts per account to be able to interpret the results of the automated analysis (for similar approaches, see Hellmueller et al., 2021). We inductively identified emerging themes by a close reading of the posts, summarized the posts into groups of themes, and interpreted the themes. This step was conducted independently for each country, arriving at a set of granular themes that were summarized and refined until data saturation was reached. We then compared themes and communication styles based on Engesser et al. (2016) study of populist rhetoric. Those authors have defined five ideological key elements of populist rhetoric (advocating for the people, attacking the elites, ostracizing others, invoking the heartland, and emphasizing the sovereignty of the people). We used these elements as heuristic categories for our text analysis and classified the posts accordingly. Finally, we compared the distribution of identified themes across actor types and countries.

Results

To answer RQ1, we first compared different emotional reactions between the countries under study (see Table 2). Regarding emotions, “angry” is the most common and is especially widespread in Belgium, Germany, and the United States. In most countries, “angry” reactions represent the most common form of user engagement, except for the United Kingdom, where “love” is about as common as anger. In Switzerland, Germany, and France, “haha” is the second most frequently used emotion, while in Belgium, the United Kingdom, and the United States, “love” is significantly more frequent than “haha.” “Sad” and “wow,” in contrast, occur comparatively rarely in all countries. Overall, we found that the use of emojis differs greatly between countries. While the most used emoji, “angry,” has a normalized count of .07 in Switzerland, it reaches a normalized count of .38 in Belgium. The use of “love” also varies from .03 in Germany to .33 in Belgium (normalized values). Belgian users use all available emojis comparatively often, while Swiss users are generally more reserved in their use of emojis. In the remaining countries, only “angry” and “love” or “haha” are used more often (see Figures 1–3).

Figure 1. Proportion of emotional user reactions in traditional vs populist politicians per country.

Figure shows normalized values of individual accounts per country. Normalized values indicate the number of emotions per outlet in relation to the number of followers of the respective account.

Figure 2. Proportion of emotional user reactions in traditional vs populist parties per country.

Figure shows normalized values of individual accounts per country. Normalized values indicate the number of emotions per outlet in relation to the number of followers of the respective account.

Figure 3. Proportion of emotional user reactions in mainstream vs alternative media per country.

Figure shows normalized values of individual accounts per country. Normalized values indicate the number of emotions per outlet in relation to the number of followers of the respective account.

Table 2. Means of Facebook reactions per country.

	CH		DE		FR		BE		UK		US
	M	Mnormalized	M	Mnormalized	M	Mnormalized	M	Mnormalized	M	Mnormalized	M	Mnormalized
Like	50404.25	0.76	371363.12	0.82	494770.12	0.58	273516.63	1.51	1775881.22	1.07	11279741.40	1.42
Love	3799.13	0.04	23491.92	0.03	48074.96	0.06	54384.17	0.33	259588.93	0.12	1728053.50	0.23
Wow	2026.21	0.02	12794.27	0.02	39715.60	0.02	24713.04	0.07	86570.37	0.02	375518.55	0.05
Haha	6477.71	0.04	55743.77	0.12	84531.40	0.09	31463.58	0.18	150709.37	0.06	1131097.40	0.16
Sad	4736.63	0.03	36783.35	0.04	93577.00	0.05	38616.54	0.10	224407.41	0.03	790118.15	0.11
Angry	6814.71	0.07	80500.92	0.19	180144.16	0.19	73331.08	0.38	189963.93	0.12	2092711.15	0.33
Sum		0.95		1.22		0.99		2.58		1.42		2.29

Note: Mean values represent the total emotional reactions of each account in a country. Normalized mean values refer to the number of reactions normalized by the number of followers of the respective account. The sum of the normalized means represents the aggregated normalized counts for each account, subsequently reflecting the average for all accounts within a given country.

Next, we compared the levels of emotional reactions among various actor types. We assumed that populist politicians (H1a) and parties (H1b) receive more angry reactions compared to traditional politicians and parties. As Figures 4a and 4b show, populist politicians in most countries under study provoked significantly higher levels of “angry” reactions relative to the number of their followers compared with traditional politicians. Donald Trump in the United States was an exception, as he received mostly “love” reactions from his followers. Traditional politicians also received mostly “love” reactions, except for Germany, where “angry” reactions predominated (but not as strongly as on the Facebook account of the German populist politician).

Figure 4a. Emotional user reactions (angry and love) of traditional vs. populist politicians per country (Belgium, France, Germany).

Boxes show the distribution per actor group. Points indicate means. Scales adjusted per country due to widely varying follower and user reactions.

Figure 4b. Emotional user reactions (angry and love) of traditional vs. populist politicians per country (Switzerland, UK, US).

We found a similar pattern for the Facebook accounts of populist parties, where “angry” reactions predominated (see Figures 5a and 5b). Since we could not sample a distinct populist party in the United States, we compared the Democratic and the Republican Parties instead. While on the account of the Democratic Party “angry” was the most common emotional reaction, “love” was found most frequently on the account of the Republican Party. On the accounts of populist parties in the remaining countries, “angry” was again the dominant emotional reaction, particularly in Germany and Switzerland. Traditional parties elicited mostly “love” reactions, except for Germany, where those parties provoked more “haha” reactions. Two-sided t-tests showed that populist politicians and parties provoked significantly more “angry” reactions than other politicians and parties across all countries under study (p = .000). Based on these findings, we accept our hypotheses H1a and H1b.

Figure 5a. Emotional user reactions (angry and love) of traditional vs. populist parties per country (France, Belgium, Germany).

Boxes show the distribution per actor group. Scales adjusted per country due to widely varying follower and user reactions.

Figure 5b. Emotional user reactions (angry and love) of traditional vs. populist parties per country (Switzerland, UK, US).

In addition, we assumed that hyper-partisan and alternative news media would elicit more “angry” reactions from their Facebook followers compared with traditional news media. As Figures 6a and 6b show, this is true for all countries under study. In Germany, France, and Belgium, alternative and hyper-partisan news media provoked the highest levels of “angry” reactions. Sadness, in contrast, was more frequent in the accounts of traditional news media. Finally, “love” did not occur as an emotional user reaction in either traditional or alternative/hyper-partisan media accounts – unlike in the accounts of political actors. Based on these findings, we accepted H2.

Figure 6a. Emotional user reactions (angry and love) of mainstream vs. alternative media per country.

Figure 6b. Emotional user reactions (angry and love) of mainstream vs. alternative media per country.

Boxes show the distribution per actor group. Scales adjusted per country due to widely varying follower and user reactions.

Further, we were interested in the posts that elicited the highest levels of angry reactions (RQ2). To answer this question, we first identified the quartile of posts from populist actors and alternative/hyper-partisan media that received the highest number of angry reactions. In a second step, we examined which words occurred more frequently in these posts than in other posts within the respective actor group. In a third step, we examined the 10 posts with the highest number of angry reactions per account using qualitative content analysis to identify dominant themes.

Table 3 shows the 10 most overused words for populist politicians and parties per country. After a close reading of all messages, we assigned the most overused words to two categories: immigration and the blaming of political opponents. While in the European countries under study, immigration seems to be a highly important topic for populist actors, the accounts of Donald Trump and the GOP focus more on blaming political enemies. That right-wing populists actors focus on immigration is far from new (Bene et al., 2022), but this topic was pivotal and provoked much anger even in a period when a global health-crisis was affecting people all over the world.

Table 3. Top 10 key words in anger-provoking Facebook posts by media outlet.

Category	CH	BE	DE	FR	UK	US
	ExpressZeitung	SCEPTR	Watergate.TV	fdesouche	politicalukblog	Breitbart
Immigration		migrants		burned	asylum	alien
		asylum seekers		créteil job	bow restore
		illegal		asylum-seeker	citizenship
		muslim		virginity
		refugees		honest
		aggressive		sudanese
		riots		Napoleon
Elite-blaming	German	leftist	EU commission president	condemns	owen	orange
		looter	environment minister		fisherman	far-left
			ruthless		lords	leftist
					integrity	disgusting
					khan	soros
					impartial	insane
						Democrat
						phony
						fact-checking
Covid-19 vaccination	gates		immunity card
	test run		mandatory vaccination
	production		back door
	vaccine		coercive measures
	operate		sovereignty
	bill		gavi
	baby
Other		spit	share	motorcyclists

Note: The 10 disproportionately often used words per account compared to the other accounts in the reference group (“overused” words per account compared to the reference group). Words were translated into English; multi-word results are an artifact of the translation.

Similar patterns can be observed regarding alternative/hyper-partisan media (Table 4). These outlets also focus on immigration and blaming national governments, for example, accusing migrants of being responsible for the spread of the coronavirus or accusing governments of trying to force the population to undergo a supposedly dangerous vaccination against COVID-19. In the United States, the hyper-partisan news outlet Breitbart focuses exclusively on attacking and blaming the Democrats for current problems. In Belgium and the United Kingdom, the alternative/hyper-partisan media combine the topics of immigration and elite blaming. These media outlets attack leftist or liberal parties and politicians and report on the alleged aggressive or illegal behavior of migrants. In Switzerland, Germany, and France, however, the COVID-19 pandemic was more present. These posts were vaccine-skeptical and included allegations of possible planned forced vaccinations of the general population. The German outlet is an outlier because of its sensationalist style. It started almost every post with the words “unbelievable” or “incredible” written in capital letters and asked its followers to “please share the article.”

Table 4. Top 10 key words in anger-provoking Facebook posts by politicians (top) and parties (bottom).

Category	CH	BE	DE	FR	U.K.	U.S.
	Koeppel	VanGrieken	Weidel	LePen	Farage	Trump
Immigration	united front	caste	visa	drop	escort	illegally
	free movement of persons	immigrant	bamf [Ministry for Migration and Refugees]		dover
	stop	suspect	family reunion		immigrant
		plunder	truck		illegal
		headscarf	unaccompanied		boat
		radical	pensioner		coast
		arrest			shocking
					landing
Enemy-blaming	session	disgusting	awo [Arbeiterwohl-fahrt]	denounce	Priti	hostile
	politicians	obituary	Cancer patient	barbarism	Patel	rioters cop terrorist
	arbitrariness	scum	Roth	laxity		bases
	killing			scandal		anarchists
	grant			fight		angry
	get			massively		scream
	stuttgart			shame		spy
Other			nationwide	kill	shock
				worst
	SVP	Vl Belang	AfD	RN	Brexit Party	GOP
Immigration	asylum seekers	agent	refugee	immigration	illegal
	gypsies	repatriate	mosque		channel
	foreigners	cell	clans		film
	illegal	hallucinatory	deport		english
	stop	immigrant			french
	criminals	asylum seeker			water
	liberty
Enemy- blaming	police	net contributor	Borchardt	tax	viral	hunter
	penny	clean up	climate control	scum	mainstream	dems
	levrat	Monaco	Wendt	acronie	willing	bank
		scum	Scheuer	clandestine	activists	left-wing
				suspicion		anarchist
				lie		subsidiary
						misinformation
						radical
						son
Other			Italian	voilà		China
			departure	communicate
				support us

Note: The 10 disproportionately often used words per account compared to the other accounts in the reference group (“overused” words per account compared to the reference group). The upper part of the tables shows results for populist politicians, and the lower part shows results for populist parties. Words were translated into English; multi-word results are an artifact of the translation.

To test whether our findings were influenced by the exceptional period of the pandemic, we conducted a replication analysis using data from 2019 and 2020, as documented in the Appendix. This analysis was performed in three countries (BE, FR, U.S.) prior to the outbreak of the pandemic (three weeks in January 2019 and in January 2020). We focused on two reaction types (love and anger) and four actor types (alternative media, traditional media, populist politicians, and traditional politicians). Our results confirm our initial findings, specifically regarding the patterns observed across media accounts. The data demonstrates consistently high levels of angry reactions generated by alternative media in comparison to traditional media in all three countries. The differences in emotional reactions to posts by political actors were even more pronounced. Notably, at the outset of the pandemic, reactions to posts made by traditional politicians were generally more positive. In France, emotional reactions were at a comparatively low level in 2019. In January 2020, especially the love reactions increased, which was probably related to the upcoming important local elections in March 2020. During the pandemic (May 2020), anger reactions also increased sharply, especially in the accounts of populist politicians. In Belgium, anger reactions increased significantly at the beginning of 2020, especially on the account of populist politicians, and, in addition, love reactions toward traditional politicians increased during the pandemic. In the U.S., love reactions to accounts of traditional politicians in particular had increased in the May 2020 study period compared to the two comparison periods.

Overall, the findings of our replication analysis indicate that the type and amount of emotional reactions are influenced by the political context and concrete events. There is little doubt that the pandemic has led to an increase in the emotional intensity of reactions, particularly in terms of anger and love. This trend was observed across all three countries, with heightened emotional reactions recorded in 2020, particularly during the pandemic.

Our in-depth qualitative analysis confirmed these patterns. The analysis showed that the 10 posts per outlet that received the highest numbers of “angry” reactions consisted of populist elements, as described by Engesser et al. (2016), such as anti-elitism (e.g., blaming the government or political opponents for current problems) and exclusion (e.g., discrediting specific social groups, such as immigrants). In one of her posts, which received more than 9,000 “angry” reactions, the German populist politician Alice Weidel claimed that a German pensioner couple was thrown out of their apartment so that asylum seekers could move in. Weidel used an emotional style in her posts, for example, by calling reported events “outrageous.” This style is adopted in the users’ reactions, which points to spill-over effects between post content and user engagement. Further, the French far right populist party Rassemblement National (RN) posted a message claiming that illegal immigrants would be allowed to demonstrate in France and called for support for a petition against refugees. A similar rhetoric can be found in the posts of the German far right populist party AfD (Alternative für Deutschland). One post, which received more than 8,000 reactions, referred to an increase in broadcasting fees for public media in Germany. In this context, the AfD spoke of a “compulsory fee” and called on users to donate to a counter-campaign. Similarly, the United Kingdom’s Brexit Party posted a video claiming that the mainstream media would deliberately hide criminal actions by refugees, which elicited 2,142 angry user reactions. In the United States, former President Donald Trump addressed similar topics as populist politicians in other countries but received mostly “love” from his followers. For example, his post stating that “The United States of America will be designating ANTIFA as a Terrorist Organization” generated 122,646 love reactions. We will elaborate on the remarkable difference in terms or reactions in the next section.

Regarding alternative and hyper-partisan media, we also find populist elements, such as exclusion and anti-elitism. For example, in one of its posts with 1,088 “angry” reactions, the German alternative news media Tichys Einblick attacked then German Chancellor Angela Merkel by accusing her of buying her power retention at the European level with German taxpayers’ money. The Belgian alternative outlet SCEPTR falsely claimed in a post that former minister Bert Anciaux wanted to introduce Arabic as an official language in Belgium, which provoked 1,696 “angry” reactions.

In summary, we found that while the frequency of various emotional user reactions on Facebook differs between countries, anger dominates across most countries (see Table 5). In some countries (Belgium, the United Kingdom, and the United States), “love” is also a frequently used emotional reaction. These country differences can be attributed primarily to a few actors who provoke a high number of emotional reactions, particularly anger, namely populist candidates, parties, and alternative/hyper-partisan news media. Our analyses of the most anger-eliciting posts showed that these messages frequently contain populist rhetoric, including elements such as anti-elitism and exclusion, focusing on themes such as immigration and blaming political opponents.

Table 5. Comparisons of normalized mean scores for “love” and “angry” emotions between account types (media, politicians, parties) across countries.

Country	Account type	Emotion	Emotion reaction count	Mean	SD	t(df)	p	n
BE	alternative media	Love	22802	0.000	0.0011	t(3084.24) = 8.72	0.001	2963
	traditional media	Love	637501	0.000	0.0003			13016
	alternative media	Angry	136136	0.002	0.0045	t(2972.46) = 16.80	0.001	2963
	traditional media	Angry	818857	0.000	0.0004			13016
DE	alternative media	Love	125610	0.000	0.0002	t(9450.65) = 13.59	0.001	7559
	traditional media	Love	401295	0.000	0.0001			15656
	alternative media	Angry	448579	0.000	0.0012	t(7698.63) = 26.47	0.001	7559
	traditional media	Angry	1286141	0.000	0.0002			15656
UK	alternative media	Love	135604	0.000	0.0006	t(4455.71) = 15.27	0.001	4419
	traditional media	Love	5720621	0.000	0.0001			36996
	alternative media	Angry	369091	0.000	0.0012	t(4420.26) = 19.10	0.001	4419
	traditional media	Angry	3926255	0.000	0.0001			36996
CH	alternative media	Love	4977	0.000	0.0003	t(837.03) = 4.60	0.001	772
	traditional media	Love	73983	0.000	0.0002			8465
	alternative media	Angry	6083	0.000	0.001	t(781.09) = 7.60	0.001	772
	traditional media	Angry	139093	0.000	0.0003			8465
FR	alternative media	Love	198280	0.000	0.0002	t(18263.40) = 16.07	0.001	14874
	traditional media	Love	637930	0.000	0.0001			19984
	alternative media	Angry	913711	0.000	0.0007	t(16716.49) = 24.76	0.001	14874
	traditional media	Angry	3240081	0.000	0.0002			19984
US	alternative media	Love	11654740	0.000	0.0008	t(20251.73) = 10.30	0.001	19176
	traditional media	Love	9107123	0.000	0.0002			23526
	alternative media	Angry	22717781	0.000	0.0006	t(25004.50) = 33.98	0.001	19176
	traditional media	Angry	15751824	0.000	0.0003			23526
BE	populist politician	Love	128475	0.001	0.0028	t(1276.99) = −7.56	0.001	229
	traditional politician	Love	362469	0.005	0.0129			1050
	populist politician	Angry	391163	0.004	0.0056	t(315.15) = 4.88	0.001	229
	traditional politician	Angry	222629	0.002	0.0051			1050
DE	populist politician	Love	21550	0.001	0.001	t(132.46) = 3.43	0.001	113
	traditional politician	Love	32897	0.000	0.0006			470
	populist politician	Angry	121370	0.004	0.0056	t(113.34) = 5.86	0.001	113
	traditional politician	Angry	54001	0.001	0.0009			470
UK	populist politician	Love	119035	0.001	0.0015	t(796.59) = −9.56	0.001	183
	traditional politician	Love	797276	0.003	0.0041			695
	populist politician	Angry	652639	0.004	0.0051	t(189.52) = 9.12	0.001	183
	traditional politician	Angry	90583	0.000	0.0014			695
CH	populist politician	Love	3141	0.001	0.0007	t(228.80) = −3.20	0.002	346
	traditional politician	Love	6012	0.001	0.0021			201
	populist politician	Angry	5557	0.001	0.0021	t(528.49) = 4.04	0.001	346
	traditional politician	Angry	1325	0.000	0.001			201
FR	populist politician	Love	149327	0.001	0.0008	t(290.28) = −0.27	0.79	165
	traditional politician	Love	161010	0.001	0.0009			603
	populist politician	Angry	218127	0.001	0.0017	t(234.03) = 3.82	0.001	165
	traditional politician	Angry	64395	0.000	0.0015			603
US	populist politician	Love	11050519	0.000	0.0006	t(2029.98) = −13.75	0.001	1284
	traditional politician	Love	2682133	0.001	0.0018			1644
	populist politician	Angry	1609508	0.000	0.0002	t(1780.16) = −11.86	0.001	1284
	traditional politician	Angry	1671096	0.000	0.001			1644
BE	populist party	Love	55580	0.001	0.0009	t(304.34) = −1.81	0.072	152
	traditional party	Love	98393	0.001	0.0016			1091
	populist party	Angry	138693	0.002	0.0033	t(157.78) = 4.99	0.001	152
	traditional party	Angry	52468	0.000	0.0013			1091
DE	populist party	Love	14964	0.000	0.0003	t(328.62) = 1.57	0.118	200
	traditional party	Love	14474	0.000	0.0003			665
	populist party	Angry	166469	0.002	0.0027	t(201.54) = 7.89	0.001	200
	traditional party	Angry	16464	0.000	0.0004			665
UK	populist party	Love	8111	0.001	0.0009	t(73.24) = 3.00	0.004	70
	traditional party	Love	228254	0.000	0.0007			1311
	populist party	Angry	9651	0.001	0.0019	t(69.31) = 2.75	0.008	70
	traditional party	Angry	80807	0.000	0.0004			1311
CH	populist party	Love	1239	0.000	0.0006	t(157.74) = 3.53	0.001	127
	traditional party	Love	1827	0.000	0.0005			628
	populist party	Angry	10031	0.003	0.0077	t(126.21) = 4.44	0.001	127
	traditional party	Angry	1464	0.000	0.0005			628
FR	populist party	Love	34339	0.001	0.0013	t(159.51) = 3.54	0.001	159
	traditional party	Love	10556	0.000	0.0002			734
	populist party	Angry	37481	0.001	0.0011	t(158.13) = 5.77	0.001	159
	traditional party	Angry	2776	0.000	0			734
US	populist party	Love	38049	0.000	0.0003	t(283.42) = 0.39	0.696	182
	traditional party	Love	28506	0.000	0.0002			199
	populist party	Angry	9015	0.000	0.0001	t(203.44) = −7.03	0.001	182
	traditional party	Angry	94999	0.000	0.0006			199

M = normalized mean; SD = standard deviation; df = degrees of freedom. “Emotion Reaction Counts” refers to the total number of specific emotional reactions recorded for each account type within a country.

Conclusion

This study uses a comparative analytical approach to improve our understanding of the favorable opportunity structures that facilitate emotionally charged political discourse. Specifically, we examine the systematic differences in emotional responses exhibited by users and how these differences are related to country of origin, types of political actors, types of media, and characteristics of messages. Our research takes into account the event environment, specifically a pandemic health crisis, and how differently configured political systems and diverse constellations of political actors, media actors, and social media users respond to this crisis.

To establish the robustness of our observations, we conducted a replication analysis using data from 2019, which shows that the underlying patterns persist over time. However, it is noteworthy that the pandemic has exacerbated the situation in certain countries. Specifically, we find that populist politicians and alternative media have skillfully used their established narrative frameworks and applied them to the health crisis to elicit emotional responses, particularly anger and affection (love), from their supporters. This finding underscores the adaptability of these actors in applying their communication strategies even in times of crisis, aligning the crisis context with their core themes, such as blaming the government, the opposition, and foreigners, to reinforce their ideological narratives.

In some countries, alternative and hyper-partisan media have also contributed to the spread of misinformation and conspiracy theories about the COVID-19 vaccine. Nevertheless, our study shows that emotional expressions are more prevalent in certain countries, such as Belgium and the United States, than in others, regardless of the contextual background of the pandemic.

In light of these observations, we conclude that the emotionalization of national Facebook spheres is primarily driven by a few high-profile accounts characterized by predominantly hyper-partisan and populist content. The anger generated by these accounts can generate uncivil forms of polarization and facilitate the spread of ideologically motivated misinformation (Rossini, 2022; Staender et al., 2021). The anger-driven engagement by their followers provides an impetus for alternative and hyper-partisan news outlets and populist figures to disseminate further contentious content, increasing the potential to translate misleading and hostile information into tangible political action.

It is important to recognize that not all forms of anger-inducing messaging produce the same results. Some actors are able to evoke positive emotions, even “love,” through seemingly hostile messages because of their loyal and like-minded followers. Former US President Donald Trump, for example, stands out as a notable outlier in this context. His messages predominantly elicited positive responses from his followers, especially when he targeted political opponents. This finding is consistent with previous research suggesting that attacking opponents is a successful strategy for pleasing populist supporters (Martella & Bracciale, 2022). Furthermore, Trump held the position of an incumbent at the time of our data collection, which significantly influenced the emotional responses of his followers. This may explain why he received numerous positive responses in support of his policies, despite criticism from his opponents. As a leader, Trump bore more resemblance to traditional politicians than to oppositional populist figures in Europe. Encouragingly, our replication analysis shows the persistence of this pattern across two pre-pandemic time points.

Future research would be well-served by examining how the emotional tenor of these responses evolved as Trump lost the 2020 election and assumed an “outsider” role. Unfortunately, this investigation was not feasible due to Meta’s prolonged suspension of Trump’s account, which was only reinstated (in February 2023) after the completion of this study.

What does the frequency of angry emotions tell us about political communication and the quality of public discourse in different countries?

First, it is imperative to acknowledge that polarizing issues, assigning blame, and using attack rhetoric are commonly used strategies to captivate audiences and elicit emotional responses in different countries. As Ihlebæk and Holter (2021) argue, the anger generated by alternative media often becomes an integral part of users’ identities as oppositional truth-seekers. This anger is fueled by their confrontation with mainstream sources and validated by alternative sources. This phenomenon highlights the complex relationship between anger, identity, and political communication.

Second, it is crucial to consider the role of cultural and political contexts in explaining the variation in the prevalence of hostile emotions, such as anger, in online discourse across countries (Hellmueller et al., 2021; Humprecht et al., 2020). The extent to which populist political figures and alternative, hyper-partisan news outlets achieve success and visibility in a given country appears to correlate directly with the prominence of hostile emotions, particularly anger, in social media discourse. In such a national environment, the pervasive presence of angry emotions can contribute to the degradation of public discourse and distort citizens’ perceptions of prevailing public opinion. Furthermore, hostile conversations can undermine trust in the political process (Mutz, 2015) and lead individuals to disengage from political information (Kim et al., 2021; Majó-Vázquez et al., 2020).

Nonetheless, it is worth noting that negative emotions, including anger, can be intertwined with more positive feelings of perseverance and resilience among populist voters and alternative media readers. These individuals often find affirmation of their beliefs and camaraderie with like-minded users.

In sum, the prevalence of angry emotions in political communication has profound implications for the quality of public discourse in different countries. By examining the strategies employed by political actors, the cultural and political contexts, and the potential interplay of emotions, our opportunity-structure approach can be helpful for developing targeted strategies to improve the quality of public discourse and promote informed and constructive political engagement in diverse societies.

There are several limitations to our study. Most importantly, based on our analysis, we cannot determine whether user reactions are related to the content of the post or to the author. For example, a populist post mostly leads to angry reactions because of the claim (e.g., that immigrants misbehave), but the anger could also be directed to the populist politicians because of the offensive language used. Reactions like “haha” and “wow” can also be ambiguous and express irony or cynicism, for example. To better understand the underlying motivations behind these reactions, future studies should examine the effects of different types of content by various actor groups. In addition, for the categorization of populist actors, we drew on the current research literature, which understands populism as political communication that manifests itself in content and style. However, this approach ignores the fact that a populist communication style consists of elements that are also sporadically used by traditional politicians. Such nuances are particularly important for in-depth analyses of individual countries or case studies and should be considered in further research. Another limitation relates to our sample of alternative and hyper-partisan media outlets. This category is very broad, and their Facebook accounts vary widely. To address this issue, we based our sampling decisions on several indicators, including reach and self-description of the outlets. Future research should collect more data on these media outlets, develop typologies, and identify functional equivalents across countries. Finally, more cross-national research is needed on the nature of Facebook reactions. In particular, the question arises as to why users in some countries generally express emotional reactions more often than users in other countries and how emotionalization on Facebook is perceived in different countries. Thus, future research on the emotionalization of social media spheres should not only focus on negative emotions in posts and comments to describe information environments but also examine discourse cultures and the general emotional expression of citizens in various countries on- and offline.

Our analysis of users’ emotional reactions belongs to a recent line of comparative research that understands social media environments as opportunity structures that favor or disfavor certain consumption patterns and media effects. Our findings provide a more nuanced view of the conditions that foster political misinformation and emotional contagion on social media (Kramer et al., 2014). The similarities across countries show how some actors with extreme positions effectively utilize the mechanisms Facebook provides. Through evidence from different nations, actor types, media accounts, and message characteristics, our study contributes to a better understanding of these mechanisms.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2024.2350416

Additional information

Funding

The work was supported by the Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung [100017L_182253].

Notes on contributors

Edda Humprecht

Edda Humprecht is professor of Digitalization & the Public Sphere at the Institute of Communication Science at the University of Jena. Her research focuses on digital political communication from a cross-national perspective.

Michael Amsler

Michael Amsler is a postdoc in the Department of Communication and Media Research at the University of Zurich. His research focuses on computational methods.

Frank Esser

Frank Esser is professor of International & Comparative Media Research in the Department of Communication and Media Research at the University of Zurich. His research focuses on cross-national studies of news journalism and political communication.

Peter Van Aelst

Peter Van Aelst is a research professor in the Department of Political Science at the University of Antwerp and a founding member of the research group ‘Media, Movements and Politics’ (M2P). His research focuses on political communication.

Notes

1. Data obtained from: https://ourworldindata.org/, numbers refer to daily new confirmed COVID-19 cases per million people as of April 14, 2020.