Alternative Media Vary Between Mild Distortion and Extreme Misinformation: Steps Toward a Typology

Abstract

Social media has faced criticism for promoting misinformation. The role of alternative media in spreading misinformation, however, remains uncertain. We examined 1661 Facebook posts from 25 most popular alternative media outlets during the Covid-19 pandemic in five countries (France, Germany, Switzerland, the UK, and the US). Our codebook covered a wide range of categories, from mild misleading content to blatant misinformation. Through cluster analysis, we identified four reporting types in alternative media: Light distortion, heavy distortion, ideological misinformation, and extreme misinformation. Light and heavy distortion were most prevalent in popular alternative media, while ideological and extreme misinformation had smaller but more engaged audiences. In summary, alternative media found more success on Facebook with content categorized as light and heavy distortion. This typology recognizes complexity in misinformation and alternative media research, while simultaneously reduces complexity to contribute to comparative studies in the future.

Keywords:

• Fake news
• disinformation
• hyperpartisan media
• social media
• facebook
• comparative research

Given the rise of misinformation during the global COVID-19 crisis and election campaigns in recent years, the role of alternative media as purveyors of false and misleading information received considerable attention (Silverman et al. 2016). However, research linking alternative media and misinformation at the content level remains limited. Notable examples include a US-based study that classifies alternative media (calling them “fake sources”) based on the extent of published misinformation, with e.g., Gateway Pundit identified as major falsehood spreaders, and Breitbart leaning towards mild inaccuracies to promote a political agenda (Grinberg et al. 2019). Moreover, a German study found that alternative media tend to propagate anti-system and populist perspectives rather than outright falsehoods (Boberg et al. 2020). Those examples indicate a range of misinformative content. One may argue that even minor deviations from facts warrant serious consideration, as they are less likely to be detected and corrected, and more likely to be viewed as credible by their target audience (Hameleers et al. 2021).

To date, the relationship between alternative media and various types of misinforming content has not been systematically studied across countries. We therefore apply a broad spectrum of content analysis categories, ranging from mild distortions to drastic misstatements. We examine 25 alternative media from 5 countries to determine how typical the dissemination of misinformation is for a given alternative media segment. Based on the amount and kind of misinformation, we identify four alternative media types using cluster analysis.

We find that alternative media that succeed in building a larger following on Facebook are characterized by distortive rather than grossly false reporting. Eighteen of the 25 most popular alternative media we studied from five countries are characterized by a journalistic program we call light distortion or heavy distortion. Only five have a journalistic program of ideological misinformation, and only two have a program of extreme misinformation. The typology we developed for this study has the potential to enrich future research on misinformation and alternative media.

Defining Alternative and Hyperpartisan Media

In today’s media landscape—where many boundaries between media types and journalistic genres are dissolving—it is increasingly difficult to distinguish between alternative and mainstream media. Therefore, Holt, Ustad Figenschou, and Frischlich (2019) propose defining all media that position themselves in opposition to the mainstream as alternatives. The alternative, they argue, is relational to and normatively and empirically defined by the mainstream. One crucial aid in alternative media classification is, according to Holt, Ustad Figenschou, and Frischlich (2019), that these outlets either describe their counter-positioning in editorial mission statements or that their alternativeness is attributed to them by audiences and third parties.

From a research perspective, this alternative nature must be reflected in the content of their coverage. According to Holt, Ustad Figenschou, and Frischlich (2019), alternative media publish other political perspectives and topics to influence public opinion in a direction that the mainstream media supposedly underexposes or marginalizes. Studies that have demonstrated the “alternativeness” of certain outlets using content analysis have, for example, investigated how right-wing alternative media in Norway have portrayed the mainstream media as biased, elitist, leftist, or politically overcorrect (Figureenschou and Ihlebæk 2019). Other studies using content analysis have shown that the left-wing alternative media in the UK have portrayed the BBC as anti-labor, pro-establishment, and generally untrustworthy (Cushion 2021).

This fixation on accusations of bias explains why alternative media are also sometimes associated with hyper-partisan media. Hyperpartisan media tend to present strongly biased perspectives on issues and events; they may be affiliated with a political camp, movement, or ideology (Mayerhöffer 2021; Rae 2021). Researchers have expressed concerns that hyperpartisan media adopt “strategies to give misleading or fake content more visibility” (Soares and Recuero 2021, 2). Both hyperpartisan and alternative media are defined as part of the journalistic periphery (Eldridge 2018; Rae 2021), rely primarily on social media to propagate their messages (Peacock et al. 2021), and are conceptually close (Mayerhöffer 2021). In this study, we include hyperpartisan outlets among alternative media only if they meet Holt, Ustad Figenschou, and Frischlich (2019) definition: “Alternative news media represent a proclaimed and (self-) perceived corrective, opposing the overall tendency of public discourse emanating from what is perceived as the dominant mainstream media in a given system” (p. 862).

Alternative Media as Spreaders of Misinformation

Following Hameleers et al. (2021), we define misinformation as “erroneous, false, or misleading information that is deemed untrue based on relevant expert knowledge or empirical evidence” (p. 1700). We also follow these authors in using misinformation as an umbrella term for a range of weak to solid types of misinformation, assuming that all these types can be used intentionally and unintentionally. Because determining the exact intentions behind publishing misinformation when analyzing only media coverage content is difficult, we use misinformation as an “overarching term” (Hameleers et al. 2021, 1700). In our case, it ranges from severe manipulation (completely false information) to various forms of decontextualization (partially false information) and distortion (misleading information) to associated stylistic elements (related to conspiracy, populism, or sensationalism).

Alternative media sources may spread misinformation for several reasons, either intentionally or unintentionally (Frischlich, Klapproth, and Brinkschulte 2020; Holt 2020; Möller and Hameleers 2019; Wardle 2018):

• Profit motive: Some alternative media may spread misinformation to generate clicks, views, or revenue. They may do so intentionally, knowing that sensational or misleading information will attract more attention, thereby increasing profits.

• Ideological bias: Alternative media may spread misinformation to promote a particular ideology or political agenda. They may do so intentionally, believing that the information they share supports their cause or beliefs, even if it is inaccurate.

• Lack of fact-checking: Alternative media may spread misinformation unintentionally due to a lack of fact-checking or critical evaluation of the information they share. They may be influenced by their own biases or simply lack the editorial resources or expertise to verify the accuracy of the information they share.

• Dissemination of propaganda: Alternative media may also be targeted by politicians or groups that deliberately spread misinformation as a form of propaganda or to sow confusion and discord.

As the last point indicates, alternative media are not the only potential senders of misinformation. Previous research showed that other sources include foreign actors, such as the Russian “Internet Research Agency” during the 2016 US presidential campaign; ordinary citizens, such as opponents of protective measures during the COVID-19 pandemic; or well-known public figures with wide reach, such as politicians or celebrities (Boberg et al. 2020; Brennen et al. 2020; Lukito 2020; Silverman et al. 2016). Also, previous research suggests that not only alternative but so-called “mainstream media” have spread false or misleading claims e.g., Fox News as a prominent example (Yang and Bennett 2021). However, only a few researchers have devoted in-depth attention to potential variations in alternative media’s misinforming content (Boberg et al. 2020; Grinberg et al. 2019). To consider the full spectrum from false to misleading information as part of the definition of misinformation, we aim to illustrate the wide range of possible variants. Thus, we distinguish between three substantive core categories and three supplementary genre-typical categories of misinformation.

Substantive Core Categories of Misinformation

Previous research on the definition of mis- and disinformation has shown that the aspect of “false content” is rarely sufficient and that most definitions additionally consider “misleading information” (Altay et al. 2023; Freelon and Wells 2020; Wardle 2018)—but without precisely operationalizing “misleading content” and clarifying in detail how it manifests at the content level. Therefore, we take a broad range of misinformation into account, following Wardle’s (2018, 952) statement that the “information disorder is not black and white; it’s fluid.” We therefore distinguish three core categories of misinformation: fabricated falsehoods, false connections, and ideological biases.

The term fabricated falsehoods refers to objectively false stories in the sense of what Wardle (2018) called “fabricated content” and “manipulated content” (p. 954). These stories are often invented from whole cloth or include fictitious quotes, events, and numbers. In visual terms, this category of misinformation typically includes photoshopped images, doctored infographics, data visualizations, and AI-supported deepfake photographs or video footage (Weikmann and Lecheler 2022). For example, within the thematic context of bushfires in Australia, manipulation would be a picture of people with artificially inserted fires in the background. For detailed coding instructions and further examples, see our Codebook in Supplementary Appendix A, p. 2.

The second category, false connections, refers to overall false content that is not manipulated and can contain true information. For instance, a particular event actually happens, and a correct image is shown, but a connection is made to untrue information, interpretations, or contexts (Hameleers et al. 2021; Wardle 2018). In visual terms, this corresponds to the decontextualization category, in which information is combined with mislabeled images, misattributed videos, or visualizations taken from other contexts as alleged evidence for a false underlying message (Weikmann and Lecheler 2022). A typical example here is the combination of text pointing at a current (often polarized debate) together with an older picture that stems from a different context: A photo from 2014 showing an art performance in Germany where dozens of people were lying on a street were misinterpreted (intentionally or unintentionally) with the caption that people were dying of coronavirus in the streets (Reuters 2020). Regarding text, false connections often manifest in clear misinterpretations (concluding something false from an actual event). For coding instructions of this category, see Supplementary Appendix A, p. 3.

The third category, ideologically biased messages, is closely related to the second category; however, in this case, information is not falsely connected but is somewhat biased, stretched, and presented misleadingly, with specific political or ideological goals in mind. Such a piece of information is neither completely false nor entirely correct, but somewhat misleading to serve a political narrative, for example, by generalizing (creating false impressions), implying (suggestive questions), exaggerating, or simplifying (Cook 2020). In terms of visuals, the “reframing or cropping” (Weikmann and Lecheler 2022) of Donald Trump’s 2017 inauguration speech is an example in which the number of people in the audience was presented in a misleading way to send a politically motivated message. The distinction from the first category is that the photo in this category is not manipulated by inserting more people artificially in the picture (complete falsehood), or taking the picture from an earlier inauguration with a bigger audience (false connection), but here the ideologically biased message would stem from cropping the edges of the picture to not show empty spaces and imply a huge crowd of people (Swaine 2018) (see also Supplementary Appendix A, p. 5).

Supplementary Genre-Typical Features of Misinformation

In addition to these core misinformation categories, previous research has identified various features typical of the misinformation genre. These refer to production conventions that have often been observed in the context of misinformation. Most of these features also match social media’s economic logic, which emphasizes users’ limited attention spans; thus, misinformation can spread rapidly through specific social network structures (Bakir and McStay 2018; Trilling, Tolochko, and Burscher 2017).

First, a misinforming reporting style is characterized by methods that Cook (2020) called “common ploys in disinformation campaigns.” They include conspiracy rhetoric, calls for skepticism, and pseudo-expert use (Cook 2020). Second, a clickbait style is often used to encourage misinformative content consumption, for example, by means of sensationalist and emotional writing or visually eye-catching modes of presentation, such as capital letters, extreme punctuation, or fear-mongering images (Damstra et al. 2021; Humprecht 2019; Mourão and Robertson 2019). Third, adopting a counter-position to certain societal groups or elites is not only a programmatic characteristic of alternative media but also a common element of populist communication (Blassnig et al. 2019) and a key stylistic component of misinforming communication (Humprecht 2019). For this purpose, we consider in- and out-group references, as well as targets of criticism, as a third stylistic indicator of misinformation.

In summary, alternative media are generally characterized by their differentiation from the mainstream and their critiques of established media and political actors (Cushion 2021; Figenschou & Ihlebæk, 2019; Frischlich, Klapproth, and Brinkschulte 2020). They also attempt to differentiate themselves through their topic range (Müller and Freudenthaler 2022). However, there is considerable ambiguity in the research as to whether the use of misinformative coverage is a universal element that characterizes the alternative media sector across countries. Using a differentiated content analysis of the core and supplementary genre-typical elements of misinformation, our study aims to provide deeper insights into the “general characteristics of alternative news media across contexts” (Holt, Ustad Figenschou, and Frischlich 2019, p. 866) and whether misinformation is one of them. Given the lack of attention paid to this topic in extant scholarly literature, the following two research questions guide our empirical analysis.

• RQ1: What types of alternative media can we identify based on their potential use (or non-use) of the core categories of misinformation?

• RQ2: How do these types of alternative media differ in their use of the supplementary genre-typical features of misinformation?

Alternative Media and Misinformation in Cross-National Comparisons

Our analysis differs from the vast majority of existing single-country studies of alternative media in that we are not interested in the specifics of any one country, but rather focus on many democracies in the West to broaden the data set and variation range. To reduce complexity, we will only include media systems whose mainstream media can be classified as belonging to the same type of journalistic culture in our analysis. We will examine countries whose established media, according to Hanitzsch et al. (2019) World of Journalism Project, belong to the so-called monitorial journalistic culture, which is prevalent mainly in Western Europe and North America. Politically, these are liberal democracies; economically, they are highly developed, affluent industrialized countries where the mainstream news media are committed to watchdog and participatory journalism. This journalistic culture combines the fourth estate and empowerment role models by holding those in power to account and enabling a critically thinking citizenry. Simultaneously, entertaining and advisory news formats are becoming increasingly important in these media systems (Hanitzsch et al. 2019, 295–298). In these countries, established media are thus within the independent journalism tradition. An independent journalist aspires to inform the general public about diverse and particularly relevant issues and events while adhering to professional and ethical standards (RSF 2021; Waisbord 2022). However, these media systems also differ in terms of three variables that are crucial to the opportunity for alternative media to develop and spread misinformation. These three variables are (1) the degree of social polarization, (2) the degree of media fragmentation, and (3) the degree of public distrust.

A high degree of polarization in a media system can make individuals entrenched in their opinions, less open to differing views, and more prone to confirmation bias. Alternative and hyperpartisan media can worsen polarization by providing biased content that reinforces existing beliefs (Ladd 2011; Steppat, Castro, and Esser 2021). As people become increasingly polarized, they may more readily accept and share information aligned with their beliefs, regardless of its accuracy (Humprecht, Esser, and van Aelst 2020).

A high degree of fragmentation allows alternative and partisan media to cater to niche audiences with content that challenges mainstream narratives or supports specific ideologies (Allern & Blach-Ørsten 2011; Steppat, Castro, and Esser 2021). This can facilitate misinformation spread within like-minded communities. In a fragmented, competitive media landscape, emotionally and ideologically charged stories—often more attention-grabbing than accurate ones—can also contribute to misinformation dissemination (Humprecht, Esser, and van Aelst 2020).

A high degree of media distrust is the third facilitating condition. As confidence in mainstream sources wanes, individuals may increasingly seek alternative or partisan sources, potentially facilitating the spread of misinformation sources (Steppat, Castro, and Esser 2021; Tsfati and Cappella 2003). These alternative and partisan outlets may not adhere to established journalistic standards, resulting in the dissemination of inaccurate or misleading information. Moreover, some purveyors of misinformation capitalize on this distrust, presenting their deceptive content as a remedy for perceived biases and inaccuracies in mainstream media.

Against this background, we assume that it is likelier that alternative media will spread misinformation in countries with highly polarized, highly fragmented, low-trust media environments. However, because previous research is very scarce in this regard, and we are entering a new research area in comparing various categories of misinformation in alternative media across countries, we aim to adopt an explorative approach to answer our final research question:

• RQ3: How do alternative media types, as potential spreaders of misinformation, differ across countries?

Data and Methods

Sample

To investigate alternative media as potential misinformation producers, we conducted a quantitative content analysis of the most popular and active alternative media accounts on Facebook (FB) in five countries. To set our sample, we first selected countries that share a similar idea and tradition of mainstream news media and alternativeness and, therefore, that share the same culture of journalism (i.e., a monitorial journalistic culture; Hanitzsch et al. 2019). These democracies share similar levels of press freedom (RSF 2021) and, historically, dominance on the part of neutral and professional journalism in the mainstream news media (Holt, Ustad Figenschou, and Frischlich 2019; Waisbord 2022). Within these democracies, we defined the following country sample: Germany, France, Switzerland, the United Kingdom, and the United States. In these countries, we saw differences regarding media trust. Research has found lower levels of trust in France, the US, and the UK, and higher levels of trust in Switzerland and Germany (Newman et al. 2020). In terms of news audience polarization and media fragmentation, the US showed the highest levels, while France, the UK, Germany, and Switzerland showed moderate levels (Fletcher, Cornia, and Nielsen 2020; Humprecht, Esser, and van Aelst 2020; Newman et al. 2020).

We further focused on the FB-profiles of alternative media because misinformation was identified primarily as a problem of the social media ecosystem (Tucker et al. 2018) and Facebook was still the social media platform most used for news at the time of data collection (Newman et al. 2020). Finally, we identified five alternative media per country using three steps. First, we consulted previous studies to create a list of outlets (Guess, Nagler, and Tucker 2019; Höllig and Hasebrink 2018; Holt, Ustad Figenschou, and Frischlich 2019; Newman et al. 2018; Schwaiger 2021; Schweiger 2017), media coverage, and expert knowledge and then checked for cross-references to those outlets to build a list of alternative media that was as complete as possible for each country. Second, we reviewed the outlets’ online mission statements or “about us” sections to determine whether each outlet met the alternative media criteria (Holt, Ustad Figenschou, and Frischlich 2019), either implicitly or explicitly. The relevant words and expressions included, for example, being an alternative, presenting other views/voices, covering what others overlook/ignore/under-report, and being independent from, critical of, and/or different from the established/large/powerful mainstream. Third, we sampled five alternative media per country and chose those with the largest numbers of FB-followers and that were engaged in regular activity at the time. By doing so, we ensured that the selected outlets had sufficient reach.

Table 1 presents our final sample of alternative media per country. We then collected FB-posts from those 25 alternative media accounts in 5 countries (n = 25,018) during 11 wk of the initial stages of the COVID-19 pandemic (April 14, 2020–June 30, 2020) and drew a sample of the 70 most popular¹ FB-posts per outlet to investigate using quantitative content analysis (n = 1661).

Table 1. Sample overview.

Country	Outlet	Original number of most popular posts retrieved	Exclusion of hour-long live video posts	Exclusion of posts that became unavailable after a short time^a	N (for analysis)
CH	ConvivaPlus	71	–	1	70
	ExpressZeitung	37	–	1	36
	InfoSperber	71	–	1	70
	JanWalter/Legitim.ch	18	–	3	15
	Weltwoche	71	–	1	70
DE	EpochTimesDE	126	–	56	70
	JungeFreiheit	71	–	1	70
	RT-DE	98	26	2	70
	TichysEinblick	70	–	–	70
	NEOPresse	70	–	–	70
FR	Basta!	70	–	–	70
	FDSouche	70	–	–	70
	RT-FR	82	12	–	70
	Sputnik-FR	88	18	–	70
	ValeursActuelles	70	–	–	70
UK	AnotherAngryVoice	70	–	–	70
	DoubleDownNews	70	–	–	70
	EpochTimesUK	96	–	26	70
	RT-UK	151	81	–	70
	TheCanary	73	–	3	70
US	Breitbart	89	19	–	70
	CNSNews.com	71	–	1	70
	DailyCaller	72	1	1	70
	EpochTimesUS	89	3	16	70
	OccupyDemocrats	86	–	16	70
n		1950	160	129	1661

Two Swiss outlets published fewer than 70 Facebook posts during the sampling period. Country codes: CH = Switzerland, DE = Germany, FR = France, UK = United Kingdom, US = United States.

^a Because the owner restricted who could see the posts or deleted the posts.

Measurements

Substantive core categories of misinformation were measured as follows² (Supplementary Appendix A): fabricated falsehoods refer to completely false statements (i.e., researchable false claims, including quotes, numbers, and events) that combine Wardle’s (2018) subcategories of fabricated content and manipulated content. The reliability score (Brennen and Prediger’s Kappa) for this category was κ = .90. This score was computed using two coders after three rounds of training. The next category, false connections, refers to false representations, interpretations, or descriptions of actual events, numbers, or circumstances (κ = .85); this category combines Wardle’s (2018) false connection and false context subtypes. The third core category, ideological bias, refers to a correct event or fact that is presented in a biased manner to support a certain political position or ideology (κ = .70). What distinguishes this category from the second is that at least one of the following strategies (as identified by Cook (2020)) is used to misrepresent reality through inadmissible generalization, intimations or suggestive questions, over- or understatements, cherry picking, anecdotal evidence, or strategic irony or sarcasm.

Capturing additional genre-typical features of misinformation, we measured conspiratorial rhetoric (κ = 0.91), calls for skepticism (κ = 1), and the use of pseudo-experts (κ = 1), adapted from Cook (2020). To operationalize clickbait journalism, we measured emotionality (κ = 0.75), sensationalist content (κ = 0.70), a sensationalist format (κ = 0.90), and fear-mongering images (κ = 0.80), as suggested by Damstra et al. (2021), Humprecht (2019), and Mourão and Robertson (2019). To operationalize counter-positioning, we measured ingroup references (κ = 0.80), outgroup references (κ = 0.75), and criticizing targets (κ = 0.77), as suggested by Blassnig et al. (2019) and Humprecht (2019). Finally, we also recorded political topics (α = 0.75) and COVID-19 topics (α = 0.81) to capture the preferred thematic associations in the FB-posts. These nominal variables’ reliability scores were calculated using Krippendorff’s alpha. A detailed breakdown of frequency distributions of each variable across countries and outlets can be found in Supplementary Appendix D (see Table D1–D5).

Results

We employed hierarchical cluster analysis to address the first RQ—Identifying alternative media types based on their use (or non-use) of core misinformation-categories. We measured distances between outlets (using Squared Euclidean distance) across fabricated falsehood, false connection, and ideological bias variables. Following Backhaus et al. (2021, 480–491), we conducted cluster analysis in two rounds. First, employing the single-linkage method, we examined if outliers impacted cluster quality. This revealed a subgroup of two cases, markedly different from others in misinformation metrics (see Figure B1, Supplementary Appendix B).

Figure 1. Multidimensional scale plot and cluster solution. The discrepancies between cases are represented using three categories in a two-dimensional space, which is why the two axes are not informative. The outlets’ loads on the three misinformation-factors are relevant. Thus, regarding the vectors, it is a matter of direction in multidimensional space. Country-codes: CH = Switzerland; DE = Germany; FR = France; UK = United Kingdom; US = United States.

Second, after excluding the two extreme cases and switching to Ward’s method, the cluster analysis yielded three cohesive and well-separated alternative media groups. While we primarily focus on these three clusters (see elbow scree plot in Figure B2, Supplementary Appendix B), we also give attention to the fourth group, containing the two outlier cases. Their internal similarity and external distance from others warrant their status as a distinct cluster, demonstrated by the initial analysis step. Moreover, their highly distinct reporting style give them theoretical relevance.

To illustrate the relationships among these four alternative media groups, we arranged them on a two-dimensional map in Figure 1. However, it’s important to envision these outlets in a three-dimensional context (across the three misinformation-categories). Therefore, discrepancies in this two-dimensional representation do not directly translate to their actual distances.

Developing an Alternative Media Typology on Misinformation Core-Categories

Clusters 1 and 2, encompassing nine alternative outlets each, are the most extensive both in terms of media entities and content volume, with 630 published posts per outlet (Table 2). Conversely, cluster 3 comprises five media organizations, and cluster 4 comprises the two aforementioned outliers. Clusters 1 and 2 exhibit more homogeneous reporting styles than clusters 3 and 4, evident in lower standard deviations (Table 2) and Supplementary Appendix B’s dendrograms.

Table 2. Clusters based on misinformation-spread.

	1 Light distorters (n = 630)		2 Heavy distorters (n = 630)		3 Ideological misinformers (n = 350)		4 Extreme misinformers (n = 51)
	M	SD	M	SD	M	SD	M	SD
Fabricated falsehood	0.03^a	0.171	0.03^a	0.167	0.15^b	0.353	0.65^c	0.483
False connection	0.03^a	0.175	0.03^a	0.175	0.13^b	0.341	0.27^b	0.451
Ideological bias	0.16^a	0.366	0.40^b	0.491	0.53^c	0.500	0.16^a	0.367
Outlets	Basta!, ConvivaPlus, EpochTimesDE, EpochTimesUK, EpochTimesUS, InfoSperber, RussiaTodayFR, RussiaTodayUK, SputnikFR		Breitbart, CNSNews, DailyCaller, DoubleDownNews, FDSouche, JungeFreiheit, OccupyDemocrats, RussiaTodayDE, TheCanary		AnotherAngryVoice, TichysEinblick, ValeursActuelles, NEOPresse, Weltwoche		ExpressZeitung, JanWalter/Legitim.ch

Unit of analysis = FB-post; N = 1661. Analysis of variance with post-hoc Games-Howell test was used for unequal group variances. Cluster means with different superscript letters (^a,b,c) are significantly different at the 0.05%-level. The means range from 0 to 1. In total, we found M = 0.07 (SD = 0.260) fabricated-falsehoods, M = 0.06 (SD = 0.239) false-connections, M = 0.33 (SD = 0.470) ideological-biases.

The typology emerging from the analysis highlights distinctive characteristics of the clusters: Cluster 1 and 2 rarely rely on fabricated falsehoods (3% of FB-posts) or false connections (3% of posts). Instead, they predominantly spread ideological biases (16% in cluster 1 and 40% in cluster 2). Cluster 3 combines an elevated proportion of fabricated falsehoods (15% of FB-posts) and false connections (13% of posts) with heavy ideological biases use (53% of posts). In theoretical terms, cluster 4 offers crucial insights by showing a small group relying heavily on misinformation. Within this cluster, the spread of fabricated falsehoods (65% of FB-posts) and false connections (27% of FB-posts) is considerably higher than in the other three clusters (Table 2). The two outlets in this group, ExpressZeitung and JanWalter/Legitim.ch, are small-scale operators entering our analysis with notably fewer posts than the remaining 23 alternative media (Table 1). This, coupled with their extreme content, likely contributed to their initial classification as outliers.

The measurement of the misinformation categories allowed us to capture no, one or multiple occurrences of the misinformation categories within a FB-post. Taking multiple coding into account, the distribution in Supplementary Appendix C, Table C1, shows that the alternative media grouped in cluster 1 spread misinformation in an average of 21% of their FB-posts and no misinformation in an average of 79% of their FB-posts. This ratio is 46:54% in cluster 2, 76:24% in cluster 3, and 98:2% in cluster 4 (see Supplementary Appendix C, Table C1).

Based on the discussed characteristics of the clusters regarding their use of the misinformation categories listed in Table 2, we name cluster 1 the light distorters, cluster 2 the heavy distorters, cluster 3 the ideological misinformers, and cluster 4 the extreme misinformers.

To better understand the four clusters, we delve into additional features, examining their reporting style, counter-positioning, and thematic content in the following section.

Specifying the Typology: Use of Misinformation Genre-Typical Features

Answering the second research question about differences in the use of genre-typical misinformation features across alternative media clusters, we employed analysis of variance and cross-tabulations to compare the variables’ frequencies (Supplementary Appendix C, Tables C2–C4). The dummy coding of most variables allows us to summarize our findings using percentages. The following patterns emerge:

Cluster 1 (light distorters) uses few genre-typical misinformation features. It uses less conspiracy rhetoric (3%), less overriding skepticism (0%), less populist in- and out-group rhetoric (17% resp. 22%), less bashing of politicians (2%) and government (8%) in its published content than the other clusters. While its reporting style tends towards sensationalism (45%) and emotions (24%), this aligns with the other clusters. Thematically, COVID-19, healthcare, security/safety, and cultural liberalism are prominent topics, similar to the other clusters, whereas human-interest (21%) and economy (7%) are more frequently covered. Overall, in accordance with the cluster-name, these alternative media promote a slightly ideologically skewed version of the traditional model of journalism.

Cluster 2 (heavy distorters) also uses few genre-typical misinformation features. Contrary to cluster 1, it uses slightly more populist in- and out-group references (28% resp. 42%). Criticism targets are mainly specific politicians (14%)—for example, Annalena Baerbock (Germany), Barack Obama (US), Laetitia Avia (France), or Priti Patel (UK). Similar to cluster 1, cluster 2’s reporting tends towards sensationalism (38%) and emotions (26%). As the cluster name suggests, there is a strong ideological bias in its reporting. Thematically, it differs from the other clusters by covering more security/safety (14%), cultural liberalism (12%), and migration (7%) topics. Cultural liberalism refers to controversial issues regarding identity politics (e.g., discrimination, racism, gender debates, civil-rights, religion).

Cluster 3 (ideological misinformers) uses significantly more conspiracy rhetoric (21%), pseudo-experts (5%), and sensationalism (58%) than the first two clusters, resembling the misinformation genre more closely. It shares similar levels of populist in- and out-group rhetoric (23% resp. 46%) with cluster 2, but criticizes politicians generically (12%), government (22%), and coronavirus measures (17%) more frequently. Thus, this cluster combines an antagonist reporting with COVID-19, healthcare, security/safety, and cultural liberalism issues, and unlike the previous clusters, spreads several fabrications and false connections.

Cluster 4 (extreme misinformers) uses significantly more conspiracy rhetoric (75%), overriding skepticism (20%), sensationalism (77%), and emotionalism (48%) than all other clusters, aligning most with the misinformation genre. Populist references to in- and out-groups are also more frequent. Criticism targets specific politicians (18%) and supranational institutions (14%); specifically, often the World Health Organization (WHO) or public figures like Bill Gates. Thematically, COVID-19 dominates (49%), with its danger being negated or downplayed in 29% of posts. Another focus was vaccines (18%). However, as with many other topics, most of the vaccine statements were completely false.

Finalizing the Typology: Comparing Popularity across Countries

Addressing the final research question, we explore the popularity of the identified clusters across countries. First, we examine the reach of alternative media clusters via Facebook (FB) followership and audience engagement (see columns II and IV, Table 3): Ideological misinformers and extreme misinformers show smaller average reaches among FB-users than heavy and light distorter outlets (column II). However, the misinformer clusters generate more audience interactions through likes, shares, or comments (column IV). While the first two clusters attract broad but more passive audiences with their ideologically skewed content, the second two clusters attract smaller but more active and committed audiences with their strongly misinforming content.

Table 3. Clusters’ reach on facebook.

		FB-followers (April 6, 2020)		Received FB-interactions
		absolute number	relative number (on national FB-usage)^a	average interactions^b	weighted on followership
		Column I	Column II	Column III	Column IV
Cluster	Light distorters
	CH-ConvivaPlus	67,636	2.63%	301	.45
	CH-Infosperber	6,585	0.26%	65	.98
	DE-EpochTimesDE	796,077	4.41%	2,380	.30
	FR-Basta!	221,723	0.82%	1,899	.86
	FR-RussiaTodayFR	1,326,366	4.90%	5,008	.38
	FR-SputnikFR	578,799	2.14%	1,643	.28
	UK-EpochTimesUK	161,846	1.01%	59	.04
	UK-RussiaTodayUK	632,200	3.93%	2,991	.47
	US-EpochTimesUS	7,109,414	6.17%	5,504	.08
	Average		2.92%		.43
	Heavy distorters
	DE-JungeFreiheit	135,838	0.75%	1,840	1.35
	DE-RussiaTodayDE	459,931	2.55%	4,094	.89
	FR-FDSouche	274,357	1.01%	2,341	.85
	UK-DoubleDownNews	216,070	1.34%	4,395	2.03
	UK-TheCanary	161,239	1.00%	2,123	1.32
	US-Breitbart	4,125,483	3.58%	93,089	2.26
	US-CNSNews	2,138,806	1.86%	32,801	1.53
	US-DailyCaller	5,380,995	4.67%	35,187	.65
	US-OccupyDemocrats	8,920,460	7.75%	150,981	1.69
	Average		2.72%		1.40
	Ideological misinformers
	CH-Weltwoche	19,998	0.78%	119	.60
	DE-TichysEinblick	53,479	0.30%	1,988	3.72
	DE-NEOPresse	49,820	0.28%	515	1.03
	FR-ValeursActuelles	180,980	0.67%	3,458	1.91
	UK-AnotherAngryVoice	357,516	2.22%	10,390	2.91
	Average		0.85%		2.03
	Extreme misinformers
	CH-ExpressZeitung	12,342	0.48%	258	2.09
	CH-JanWalter/Legitim.ch	17,295	0.67%	300	1.73
	Average		0.58%		1.91

Country-codes: CH = Switzerland; DE = Germany; FR = France; UK = United Kingdom; US = United States.

^a The percentages (based on national FB-usage for news; Newman et al. 2020) are used for country comparison reasons; however, the followership may be global in scope.

^b Average of all FB-interactions received (i.e., likes,emoji-reactions,shares,comments).

Second, to evaluate popularity patterns in cross-national comparison, we use a cross-tabular breakdown (Table 4). The “Total” column in Table 4 shows that light distortion and heavy distortion are much more common journalistic programs for alternative media internationally than ideological misinformation and extreme misinformation – at least for those that are successful on social media and thus made it into our sample.

Table 4. Cross-tabulation of alternative media clusters and countries.

		Country
		CH	DE	FR	UK	US	Total
Cluster	1-Light distorters	2	1	3	2	1	9
	2-Heavy distorters	–	2	1	2	4	9
	3-Ideological misinformers	1	2	1	1	–	5
	4-Extreme misinformers	2	–	–	–	–	2
	All	5	5	5	5	5	25

Unit of analysis = alternative media (n = 25). Country-codes: CH = Switzerland; DE = Germany; FR = France; UK = United Kingdom; US = United States.

Looking at specific countries, the analysis shows that ideological misinformers and heavy distorters—both frequent disseminators of ideologically biased messages—are among the most popular outlets in Germany, the UK, and especially the US, where they represent four out of the five most popular outlets. In contrast, light distorters are popular in France (including two Russian-owned outlets, RussiaToday and Sputnik) and Switzerland. The Swiss case demonstrates a dual spectrum concerning the spread of misinformation, having two extremes (light distorters and extreme misinformers) among their most popular alternative media. Those findings confirm our cautious expectations that more polarized, fragmented, and low-trusted media systems (the US, the UK, and France) provide favorable conditions for mild and severe distortion. However, contrary to our expectations, in those media systems ideological and extreme misinformers barely appeared among the most popular alternative media. Switzerland stands out for the highest prevalence of extreme misinformation. Here, media laws are particularly liberal, and the culture of public discourse is particularly tolerant. Additionally, due to a generally small alternative media scene, niche outlets with more extreme editorial profiles have swept into our Swiss sample. Extreme misinformers in niche form probably also exist in the other media systems, but they did not rank among the five most popular on FB and did therefore not enter our sample.

It can be concluded that alternative media that aim to establish a larger national audience on social media choose a journalistic program that focuses more on (mild to severe) distortion than on crude misinformation. An intermediate category are ideologically motivated misinformers (cluster 3), who repeatedly cause a stir by deliberately crossing boundaries. They have a small, loyal, but vocal and presumably easily mobilized core audience on Facebook and may deserve particular attention in future scholarship on alternative media.

Conclusion

Social media have been vilified for facilitating the spread of misinformation. While alternative media have been attributed a role in this regard (Grinberg et al. 2019; Silverman et al. 2016), no one has yet examined their relationship with a broad range of misinformation formats conducting a systematic, international comparative content analysis.

Our main conclusion from the outset: The alternative media that are more successful among Facebook users in the respective country (see column II, Table 3) overwhelmingly choose not to adopt an editorial profile that focuses on extreme misinformation. Apparently, only niche outlets (such as ExpressZeitung and Legitim.ch from small Switzerland) can afford to do so. The alternative media (from clusters 1 and 2), which tend to be larger, serve their followers with reports that we have classified here as lightly distorted and heavily distorted. And this may be the basis of their success.

Central to this insight was the development of a typology that deliberately included a wide range of misleading information types. The advantages of our typology, as we will show, are simplification and organization, pattern recognition, and theory building.

Our four clusters first decompose a complex phenomenon by distinguishing two subtypes of distortion/bias on the one hand and two subtypes of misinformation/falsehood on the other. We consider this division of the misinformation coverage to be a heuristically valuable, order-creating achievement that could fertilize future theorizing about alternative media and misinformation conceptualizations. In terms of further theorizing, one might assume that our four types map a linear continuum of misinformation, because the proportion of fabricated falsehoods and false connections seems to increase successively from type I (light distorters) to type IV (extreme misinformers). But this impression of a linear continuum is not well supported by the data for two reasons. First, because the proportions of fabricated falsehoods, false connections, and ideological bias (as reported in Table 2) do not increase in a perfectly linear fashion from type I to IV, and second, because the other content characteristics tend to argue for four distinct types (see Table 5). The methodology used here also argues against the notion of a continuum.

Table 5. Characteristics of the four types of alternative media as misinformation spreaders.

	I: Light distorters	II: Heavy distorters	III: Ideological misinformers	IV: Extreme misinformers
Spread of misinformation	sporadic fabricated falsehoods (3%); false connections (3%); few ideological bias (16%)	sporadic fabricated falsehoods (3%); false connections (3%); heavy ideological bias (40%)	frequent fabricated falsehoods (15%); false connections (13%); very heavy ideological bias (53%)	persistent fabricated falsehoods (65%); false connections (27%); few ideological bias (16%)
Reporting style	Traditional	Antagonist	Antagonist and skeptic	Conspiratorist
News agenda	Broad agenda	Partisan agenda	Partisan agenda	Narrow agenda (i.e., focus on few topics)
Audience	Large reach, rather passive audience	Large reach, rather passive audience	Small reach, rather active audience	Small reach, rather active audience
Countries	All studied countries	US (most popular), UK, FR, DE	DE, CH, FR, UK	CH

Country codes: CH = Switzerland; DE = Germany; FR = France; UK = United Kingdom; US = United States.

Our typology was only made possible by a systematic comparative analysis. It now allows us to identify variations and relationships between variables – and to make assumptions about causal relationships for future hypothesis generation. Table 5 serves to identify such relationships between different types of distortion and misinformation, other editorial content characteristics, and related audience characteristics. We have referred to these relationships as “journalistic programs” behind the four types.

Table 4 further shows how the alternative media with their respective journalistic programs are distributed among the five countries studied. For the sake of contextualization, we characterize the four types summarized in Table 5.

The first type, light distorters, rarely uses falsehoods but occasionally spikes FB-posts with misinformation. Compared to the other types, these outlets come closest to established and professional media, as they not only use a few misinformation-typical style elements (e.g., conspiracy rhetoric, skepticism) but also cover a broad range of political topics and engage in little actor criticism. A clear political line is thus less evident. This confirms Müller and Freudenthaler (2022) study that found for German RussiaToday and EpochTimes an emphasis on general interest news and less pronounced populist partisanship. Nevertheless, these media outlets (especially the Russian-influenced outlets RussiaToday and Sputnik) often criticize the government – presumably as an intended foreign influence to create distrust. RussiaToday has a wide reach in Germany and the UK, while in France even two Russian-influenced outlets are among the most popular alternative media.

The second type, heavy distorters, frequently circulate ideologically biased content. Their reporting is antagonist through focusing on clear ingroups and outgroups (e.g., in the US, Trump supporters vs critics), pointing criticism primarily against politicians of the outgroup, and covering mainly migration, security, and political conflict issues – those issues refer to debates “against tradition” related to identity-politics. Typical in this category is the cherry-picking strategy, repeating a misleading message from another source or omitting context to provoke the own readership and to follow a political narrative: For example, “Rush [Limbaugh] believes Democratic Governors are keeping lockdown orders in place to hurt Trump politically” (from US-TheDailyCaller; May 13, 2020). Heavy distorters are most popular in our US sample, where low media trust and a fragmented media market facilitate a relatively large reach for such media, Also, in pronounced polarized environments with clear political camps, such partisan reporting—where opponents are criticized at the expense of facts (Rae 2021)—falls on fertile ground.

The third type, ideological misinformers, combines a consistent spread of ideological bias with frequently fabricated falsehoods and false connections, demonstrating a close relationship between misinformation and hyperpartisan news. Similar to type II, they show antagonistic and out-group reporting. However in contrast to typ II, they criticise more generically “the” politicians and “the” government, rather than specific political opponents, and they regularly spread conspiracy rhetoric. Suspected “intrigues” of societies’ elites appear, for example, in phrases of the German NEOPresse (e.g., “Merkel conspiracy,” “compulsory vaccination,” and “forced mail ballots”), or in the British outlet AnotherAngryVoice by regularly criticizing the Tories’ closeness to the BBC. Ideological misinformers are among the most used alternative media, except for the US. This summarizes that an antagonistic style (type II) dominates in the US, while elite skepticism (type III) is more pronounced in Western European countries.

The fourth type, extreme misinformers, is characterized by the most fabricated falsehoods, false connections, and ideological biases. These outlets use misinformation and a clickbait-style to fuel distrust of the country’s established politics and media environment. We further observed that extreme misinformers almost exclusively addressed the topical and polarizing COVID-19 issue and criticized Bill Gates and the WHO in line with conspiracy narratives.

Another advantage of our typology is that it allows us to hypothesize about other applications. We strongly believe that our types are also applicable to non-alternative media. For example, we found a correlation between sensationalism and conspiracy rhetoric, suggesting that tabloid media should also be examined for misinformation, especially if they also tend to have a strong ideological bias. During our coding process, we also encountered misleading headlines in recognized quality media. For example, the headline of an online story from the German public broadcaster ZDF read “Corona in papayas?” and turned out to be insubstantial click-bait. Since previous research has already pointed out that the boundaries between alternative and mainstream media are in many ways fluid (Brems 2023; Kenix 2011), it makes sense to examine the first two clusters of light distortions and heavy distortions in particular for their prevalence in mainstream media.

The present study is not without limitations: First, it is important to note that we analyzed only the five most popular alternative media in each country. We did not include alternative media with smaller social media reaches, so our sample does not comprehensively cover the countries’ alternative media landscape. The identified types may therefore exist in all analyzed countries but are less widely used.

Second, we focused only on text and content that was directly consumable on Facebook and did not analyze the content of hyperlinked articles. As some random checks indicate, there was often more extreme and false content in a linked article than in the preview version and headline on Facebook. This may have strategic reasons to comply with the platforms’ regulations and avoid being banned (see e.g., Innes and Innes 2023). Additionally, our sample does not reflect a list of the most popular alternative media that are robust over time and online channels. First, in each country under study, there are further relevant outlets outside the social media sphere (e.g., “Politically Incorrect” in Germany). Second, the alternative media sphere seems to be a fast-changing field, where outlets change their name (e.g., from “WatergateTV” to “NEOPresse” in Germany) or disappear (e.g., “Westmonster” in the UK).

Third, we limited our sample to European democracies and the US, so our results cannot be generalized to countries with different journalism cultures (Hanitzsch et al. 2019; Waisbord 2022). Due to naturally different definitions and relationships between mainstream media and alternative media in other countries, we expect only limited applicability of our types. For example, in parts of Asia, Africa, or Latin America, alternative media have historically played pivotal roles in contesting state-controlled narratives, often serving as counter-hegemonic voices against dominant political regimes (Waisbord 2022). In some Asian contexts, alternative media can be tightly interwoven with grassroots movements, emphasizing local narratives and indigenous knowledge against globalizing forces. In certain African nations, misinformation often intertwines with oral traditions, community radio, and local vernaculars, challenging our conventional understanding of “fake news” (Wasserman 2020). These variances underscore the need for caution when transplanting concepts and frameworks across different cultural and geopolitical terrains (Frischlich and Humprecht 2021). Therefore, the adaption of our typology in other media systems is highly needed.

Fourth, we must keep in mind that this study was conducted during an atypical period where societies experienced extreme situations in their lives and information consumption habits (i.e., “infodemic”). This may explain why we found little media criticism (specific and generic) and why other societal actors were more often targets of criticism.

We would like to end by emphasizing the practical implications that our study has in addition to its theoretical contributions. Practically, a more granular understanding allows policymakers, educators, and platforms to tailor interventions more precisely. For instance, strategies to counter light distortion might differ from those addressing extreme misinformation. Such targeted approaches can enhance the efficacy of misinformation countermeasures, ensuring that resources are allocated in a manner that addresses the specific nature and impact of each category. With this study, we hope to pave the way for future comparative research on the role of distinct types of alternative media in the context of misinformation dissemination. In addition, our results can help researchers better identify misinformation and sensitize citizens to its identifying characteristics.

Disclosure statement

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

Additional information

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is part of a project funded by the Swiss National Science Foundation and the Research Foundation – Flanders (Grant No. 100017 L_182253).

Notes

1 Popularity = sum index of received likes, emoji reactions, shares, comments per FB-post.

2 The reliability scores (two coders, third pretest): Brennen and Prediger’s Kappa for dummy variables; Krippendorff’s alpha for nominal variables.