Theory and Expectations

Responding to environmental degradation and climate change requires enacting key policies that have broad public support and the electoral backing of parties and candidates. Understanding the politics of environmental policy is therefore crucial to ensuring that successful environmental action is politically sustainable in the long term. The literature suggests that political support for environmental policies may be undermined by a host of reasons, such as the costs and distributional consequences that they entail (see, for example, Bechtel and Scheve Reference Bechtel and Scheve2013; Bergquist et al. Reference Bergquist, Mildenberger and Stokes2020; Bernauer and Gampfer Reference Bernauer and Gampfer2015; Egan and Mullin Reference Egan and Mullin2017; Stokes and Warshaw Reference Stokes and Warshaw2017; Colantone et al. Reference Colantone, Di Lonardo, Margalit and Percoco2024; Voeten Reference Voeten2024), the mobilization of political parties that are skeptical of the green transition (Spoon et al. Reference Spoon, Hobolt and De Vries2014; Dickson and Hobolt Reference Dickson and Hobolt2025), or societal skepticism towards political and scientific elites (Rhodes et al. Reference Rhodes, Axsen and Jaccard2017; Kitt et al. Reference Kitt, Axsen, Long and Rhodes2021; Huber et al. Reference Huber, Greussing and Eberl2022; Krange et al. Reference Krange, Kaltenborn and Hultman2021; Meijers et al. Reference Meijers, Van Drunen and Jacobs2023).

The existing literature has comparatively paid less attention to the role of protest events and outerparliamentary political activity when it comes to the environment. While only a small subset of the population is involved in environmental protests, their consequences are much more widespread (see, for example, Valentim Reference Valentim2023; Brehm and Gruhl Reference Brehm and Gruhl2024). So, the question becomes: how do these protest activities affect support for pro-environment and environment-critical movements in the wider population? A vast literature on protest activity has shown that protests have important consequences. They affect public opinion, the position-taking of political elites, and the agenda-setting of the media (Madestam et al. Reference Madestam, Shoag, Veuger and Yanagizawa-Drott2013; Wasow Reference Wasow2020; Gause Reference Gause2022a, Reference Gauseb), and these effects can be persistent (Mazumder Reference Mazumder2018), especially if exposure is local and repeated (Tertytchnaya and Lankina Reference Tertytchnaya and Lankina2020; Valentim Reference Valentim2023).

What is more, this literature suggests that the protest repertoire used—how violent, peaceful, or disruptive protests are—crucially affects public opinion (Wang and Piazza Reference Wang and Piazza2016; Enos et al. Reference Enos, Kaufman and Sands2019; Feinberg et al. Reference Feinberg, Willer and Kovacheff2020), legislators’ beliefs (Wouters and Walgrave Reference Wouters and Walgrave2017), and media reporting (Myers and Caniglia Reference Myers and Caniglia2004; Wouters Reference Wouters2013). Using evidence from 23,000 protest events in the United States between 1960 and 1995, Wang and Piazza (Reference Wang and Piazza2016) show that protesters face a trade-off between raising awareness by engaging in more disruptive protest activities but losing public support as a result of this disruption (see also Simpson et al. Reference Simpson, Willer and Feinberg2018; Shuman et al. Reference Shuman, Saguy, van Zomeren and Halperin2021, Reference Shuman, Hasan-Aslih, Van Zomeren, Saguy and Halperin2022). In the European context, a study by Muñoz and Anduiza (Reference Muñoz and Anduiza2019) exploits an unexpected riot event by an anti-austerity movement that broke out during a large-scale protest in Spain while a survey was in the field to show that street violence episodes reduced support for anti-austerity movements (see, for example, Hager et al. Reference Hager, Hensel, Hermle and Roth2022). Other studies demonstrate that protest size, disruptive protest action, and symbolic protest action are important drivers of media coverage of protests (see, for example, Smith et al. Reference Smith, McCarthy, McPhail and Augustyn2001; Amenta and Alan Elliott Reference Amenta and Alan Elliott2017; Hsiao and Radnitz Reference Hsiao and Radnitz2021; Wouters Reference Wouters2013) and public support (Feinberg et al. Reference Feinberg, Willer and Kovacheff2020; Wouters Reference Wouters2019) for protests in the United States and Europe. Building on these important findings, this study contributes to our understanding of how the degree of disruption and violence affects support for protest activities concerning the environment among the wider public, and how the kind of group that stages the protest matters.

Environmental protest events have been on the rise in recent years (Carnegie Endowment for International Peace 2024). Pro-environment movements have staged protests to raise public awareness about the environment, biodiversity loss, and climate change to compel government action to avoid tipping points in the climate system and ecological collapse. The most famous example of a pro-environment movement is Extinction Rebellion (XR).Footnote ¹ Opposition to environmental policies and their redistributive effects also gave rise to environment-critical movements. Examples of such movements have, for example, been farmer groups.Footnote ² Across Europe, both pro-environment and environment-critical movements have resorted to disruptive actions to showcase their grievances as they have obstructed traffic on highways, vandalized private or public property, and, in some cases, resorted to the use of violent tactics. These protests have been widespread. Figure 1 shows protest events by pro-environment groups (XR, Just Stop Oil, and Fridays for Future combined) and farmers that have been critical of environmental policies in European countries from October 2019 to November 2024. To count these events, we rely on data from ACLED (Armed Conflict Location & Event Data), a non-profit organization that collects worldwide conflict and protest data. More information about the data collection and descriptive statistics can be found in Appendix section A.

Figure 1. Pro-environment (XR, Just Stop Oil, FF) and environment-critical (farmers) protests between 2019 and 2024.

Figure 1 gives us a general sense of the spread of pro-environment and environment-critical protests across Europe by homing in on protests by climate activists and farmers’ protests respectively. This is not to say that all these protests are the same. Pro-environment groups like Extinction Rebellion, Just Stop Oil, and Fridays for Future differ in their rhetoric, protest actions, and goals. Nonetheless, they are united in raising public awareness about environmental protection, biodiversity, and climate change more broadly. Likewise, while environment-critical farmer protests have different triggers across countries, they have mostly been fueled by concerns and resentment over perceived adverse effects on the economy and industry due to tighter environmental regulation (Finger et al. Reference Finger, Fabry, Kammer, Candel, Dalhaus and Meemken2024).

Larger questions about how to get environmental policy and the green transition right, and how to align them with equitable distributional outcomes, loom behind these disruptive protests (Iskander and Lowe Reference Iskander and Lowe2020; Valentim Reference Valentim2023). These protests are also an important warning sign to political elites: when governments get environmental policy wrong, or fail to do it fairly, they not only risk further environmental degradation and biodiversity loss, but also that political behavior in their countries might move in a more extreme direction. To understand the possible societal ramifications of pro-environment and environment-critical protests and the long-term support for environmental policy, it is important to study how pro-environment and environment-critical protests shape support among the wider public.

Important studies have explored the effects of environmental protests on public opinion. Valentim (Reference Valentim2023), for example, examined how pro-climate protests by the Fridays for Future movement in Germany affected voting for the Green Party using a difference-in-differences design. His findings show that repeated exposure to environmental protests increases the vote share of the Greens, but not other environmentally friendly behaviors (see also Brehm and Gruhl Reference Brehm and Gruhl2024). Based on survey experimental evidence from the United States, Bugden (Reference Bugden2020) finds that peaceful pro-climate protests increase support for pro-environment movements among Democratic and independent-leaning voters, but not among Republicans. Kountouris and Williams (Reference Kountouris and Williams2023) investigate Extinction Rebellion protests that happened during the fieldwork period of survey in the United Kingdom to examine how these protest events affect pro-climate attitudes and support for pro-environmental policy and behavior. The findings suggest that while exposure to protest increases support for pro-environmental behavior and policy, it lowers people’s willingness to pay more for environmentally friendly consumption (see also Ostarek et al. Reference Ostarek, Simpson, Rogers and Ozden2024). Finally, Zeitzoff and Gold (Reference Zeitzoff and Gold2024) examine public attitudes towards various environmental protest tactics, with a particular focus on cyber-based actions employed by the radical environmental movement in the United States. The results suggest that American respondents are more supportive of cyber tactics compared to strategies involving property damage or sabotage, and support is positively associated with heightened concern about climate change.

These findings are instructive but they only focus on the effects of pro-environment protests. We build on this work to consider how different kinds of protest activities (varying based on their legality, level of disruption, and use of violence) of different types of groups (pro-environment protest movements v. environment-critical protest movements) matter. In keeping with the general protest literature discussed above, we expect that within democracies, when it comes to environmental protests, citizens show little support for violent and disruptive protests compared to legal and non-violent alternatives, regardless of the protest movement that stages them (Hypothesis 1).

Based on the work on rural resentment and the urban–rural divide in politics (see, for example, Berenguer et al. Reference Berenguer, Corraliza and Martin2005; Cramer Reference Cramer2016; Tosun et al. Reference Tosun, Schaub and Marek2024; De Lange et al. Reference de Lange, van der Brug and Harteveld2023), we expect that rural residents will show higher support for farmer-led protests than those of climate activists, compared to urban residents (Hypothesis 2). Rural resentment reflects long-term discontent among people residing in rural communities about the extent to which the central government favors urban areas at the expense of rural areas. Recent research suggests that rural residents often share perceptions of being victims of ‘distributive injustices’ by those in power (Ziblatt et al. Reference Ziblatt, Hilbig and Bischof2024, 4). Rural communities are usually less connected to national decision-making centers and often feel that their communities have been left behind by the state, and that they do not receive their fair share of public services (Ziblatt et al. Reference Ziblatt, Hilbig and Bischof2024; Cramer Reference Cramer2016; Cremaschi et al. Reference Cremaschi, Bariletto and De Vries2025; Patana Reference Patana2020). As a result, rural residents may show higher support for political parties and social groups that stress rural belonging and associated inequalities, and promise to restore the status of neglected places (Cramer Reference Cramer2016; Elgenius and Rydgren Reference Elgenius and Rydgren2017; Cremaschi et al. Reference Cremaschi, Bariletto and De Vries2025; Hochschild Reference Hochschild2018).

Yet, feeling closer to social groups and their grievances might not only originate from the urban–rural divide, but also from shared concerns about how environmental policy adversely affects lower educated and economically vulnerable people in society (Otjes and Krouwel Reference Otjes and Krouwel2022; Voeten Reference Voeten2024), or feelings that the green transition is being ‘imposed’ by societal elites, most notably scientists and politicians (Schwörer and Fernández-García Reference Schwörer and Fernández-García2023). What is more, these types of grievances are increasingly mobilized by populist radical right parties (Küppers Reference Küppers2022; Oswald et al. Reference Oswald, Fromm and Broda2021; Vihma et al. Reference Vihma, Reischl and Nonbo Andersen2021). As a result, we expect supporters of populist radical right parties, who generally have less trust in science and politics, to be more supportive of farmer-led protests compared to the same type of protest activities staged by climate activists (Hypothesis 3).

The Dutch Case

Before we present the findings from our two empirical studies, we first provide some background on the Dutch case. The Netherlands is one of the leaders in the green transition in Europe (Fritz et al. Reference Fritz, Willendorf, Zangemeister and Neuss2024). It is also a member state of the European Union, which has formulated a set of green policies, known as the Green Deal, to become the ‘world’s first climate-neutral continent’ by 2050 (Commission 2024). The dedication to climate and environmental action is in part a result of the fact that large parts of the Netherlands are below sea level, and therefore especially vulnerable to climate change and its consequences, such as rising sea levels. Both historically and more recently, the country has witnessed several flooding disasters with dire consequences for affected citizens. For these reasons, the Dutch public consistently ranks high in support for policies combating climate change and protecting the environment (Inglehart Reference Inglehart1995; Poushter et al. Reference Poushter, Fagan and Gubbala2022). A recent Eurobarometer report shows that 72 per cent of the Dutch population agrees that environmental issues have a direct effect on their daily life and health, and 83 per cent agrees that EU environmental legislation is necessary to protect the environment in the Netherlands (European Commission 2024).Footnote ³ That said, the Netherlands is one of the biggest exporters of agricultural goods worldwide (Baazil Reference Baazil2022). Protecting nature and the environment more generally might involve trade-offs. Environmental protection might especially require changes in the agriculture sector and thus have considerable consequences, especially for those employed in or reliant on the agricultural sector. This tension between the need to protect the environment and the interests of the agricultural sector to resist adjustments came to a boiling point in October 2019, with large-scale protests erupting across the country.

The origins of these protests can be traced back to the summer of 2019, when the Dutch Council of State, the highest administrative court in the Netherlands, released a report stating that ‘drastic measures’ were needed for the country to meet its nature preservation goals and reduce nitrogen emissions. It ruled that the Netherlands’ nitrogen permits system was failing to prevent nitrogen emissions from harming protected nature reserves and biodiversity in general. Nitrogen emissions mainly consist of ammonia. By far, the most ammonia is released into the air through agriculture. In the Netherlands, approximately 91 per cent of total ammonia emissions stems from farming, especially dairy farming (Centraal Bureau voor de Statistiek (CBS) 2020). Although it was far from clear how this recommendation would be implemented by the government, farmer protests erupted in October 2019 against the more general environmental policy goals of the Dutch government to protect biodiversity and nature in the country.

While the farming sector is relatively small in the Netherlands (51,000 farms, not only including livestock farms, but also horticultural companies), and dairy farming accounts for just 1 per cent of Dutch gross domestic product (Schouten Reference Schouten2023), large-scale protests by Dutch farmers erupted on 1 October 2019. On that day, 2,000 tractors blocked major highways in the Netherlands. This was in part a reaction to one of the government parties (Democraten66, D66) proposing a reduction of Dutch cattle by 50 per cent in order to meet the nature preservation goals entailed in European regulation transposed by Dutch parliament in national regulation. Subsequent waves of protests intensified when the Minister of Agriculture, Nature and Food Quality set a maximum for the protein content in concentrates for dairy cattle in 2020, and later on between 2021 and 2023, when the government presented new reports on how to cut back nitrogen emissions. The main mobilizing group was the Farmers Defence Force, which mobilized farmers, mostly online, to vandalize public property, block highways, and visit government ministers at their private homes. Recently, the Dutch government has stalled its plans to combat the nitrogen crisis, in part because of a snap election in the fall of 2023 after a government collapse due to the issue of migration and refugee settlement.

The farmer protests that began in October 2019 were very quickly followed by pro-environment protests. The Dutch branch of Extinction Rebellion, a pro-environment group founded in the United Kingdom in December 2018, started staging protest actions around the Netherlands in the Spring of 2019. This group, but also other environmental groups like Greenpeace, staged protests to raise awareness about the environment, climate change, biodiversity loss, and governments’ ongoing subsidies for fossil fuels. Similar to protesting farmers, their repertoire of protest actions also includes blocking roads and highways and vandalizing public or private property. Most notably, since the summer of 2022, XR has organized regular protests on the A12 highway, an important and busy highway in The Hague. In 2022, XR and Greenpeace activists also stormed an area holding private jets at Schiphol International Airport, where they cycled around and chained themselves to private jets (Moloney Reference Moloney2022). These protests lasted for years and became known as the nitrogen crisis. The environment question deeply polarized Dutch society: not only did environmental stances come to define voters’ left–right placements (Otjes and Krouwel Reference Otjes and Krouwel2022), and a new farmers’ party secured an electoral breakthrough (Otjes and Krouwel Reference Otjes and Krouwel2022), but also the Dutch Commissioner for Climate in the European Commission returned to Dutch politics to lead an electoral coalition between two left parties to advocate more pro-environment stances (Tullis Reference Tullis2023). For a timeline of the pro-environment and environment-critical protests in the Netherlands, please see Appendix section B.

Empirical Approach

To better understand how disruptive, or even violent, protest features by pro-environment and environment-critical groups affect public support for protests, we present evidence from two empirical studies. Study 1 reports evidence from a pre-registered vignette experiment embedded in a representative survey of the Dutch population. Study 2 documents findings from a quantitative and qualitative content analysis of media reporting of the most popular tabloid and broadsheet newspapers in the Netherlands.

Experimental Design

We showed respondents two fictional protest scenarios. These vignettes resemble real protests that have occurred in the Netherlands between 2019 and 2023. We debriefed respondents about the experiment and randomization of the vignettes at the end of our survey. In the two vignettes, we randomized the levels of three protest features: the ’protest movement’, the ’protest actions’, and whether ’violence was used or not’. These protest features are shown in Table 1. More details about the rationale behind the vignette design can be found in Appendix section C.

Table 1. Protest features and levels

Note: The first column displays protest features, the second column shows levels of these protest features, and the third column shows how levels are operationalized as text in the vignettes.

Protest movement

We first randomize the protest movement. We contrast a pro-environment movement against an environment-critical movement. For the pro-environment group we use the generic term ‘climate activists’ (Dutch: ‘Klimaatactivisten’) in our protest scenarios. We chose the term ‘climate activists’ over specific group names, such as ‘Extinction Rebellion’, to account for the multiplicity of pro-environment groups (for example Greenpeace, Just Stop Oil), and because the use of the term ’climate activists’ is more recognizable than alternative wordings such as ‘environmental activists’. For the environment-critical group we choose ‘farmers’. We focus on farmers because they have been the most visible environment-critical protest movement in the Netherlands (Schelfaut Reference Schelfaut2019). Dutch farmer protests were coveretd worldwide in international media and acknowledged as a leading example of green backlash movements (Tullis Reference Tullis2023).

Protest actions

We then randomize protest actions. We loosely modeled these actions after the pyramid of environmental tactics (Bugden Reference Bugden2020; Brown Reference Brown, Sowers, VanDeveer and Weinthal2021). The pyramid conceptually describes escalatory steps in the repertoire of environmental protest tactics. At the bottom of the pyramid sits legal protest, followed by civil disobedience, sabotage, and, eventually, violence (Brown Reference Brown, Sowers, VanDeveer and Weinthal2021). In our vignette, we operationalized legal protest as having ‘protested at the Malieveld’. The Malieveld is a big open field close to the central station in The Hague allocated for legal protests by the public authorities. We operationalized civil obedience as having ‘obstructed traffic on a highway’. Lastly, we operationalized sabotage as having ‘vandalized state property’.Footnote ⁵

Use of violence

Finally, we randomize whether violence, at the top of the environmental protest tactics pyramid, was used or not. We operationalized violence in the vignettes as ‘a number of demonstrators assaulted police officers’. We also include a slightly modified level of violence: ‘a number of demonstrators were arrested for assaulting police officers’. We add this second attribute level to our attribute randomization to see if the cue about protesters being arrested changes perceptions about the acceptability of violence. The final level is non-violence, which we operationalized as ‘communication with the police went smoothly’.

Treatment scenarios and outcomes

The protest features are randomized within and across two vignettes. We first show respondents the following introductory text:

The Netherlands has experienced several protests. We are interested in your opinion on two protests. You will now see the description of the first protest. Read the description carefully.

After this, respondents see the first protest scenario followed by some outcome questions and then proceed to the second protest vignette. In the second vignette, we adjust the feature order to account for a possible ordering effect. The protest vignettes take on the following form:

Protest 1: In this protest, farmers [climate activists] protested at the Malieveld [vandalized state property/obstructed traffic on a highway] to express their dissatisfaction with the current Cabinet’s climate plans. During the protest, a number of protesters assaulted police officers [a number of protesters were arrested for assaulting police officers/communication with the police went smoothly].

Protest 2: In this protest, there was a demonstration at the Malieveld [state property was vandalized/traffic was obstructed on a highway] by farmers [climate activists] to express their dissatisfaction with the current Cabinet’s environmental plans. During the protest, a number of protesters assaulted police officers [a number of protesters were arrested for assaulting police officers/communication with the police went smoothly].

After each protest vignette, we ask respondents to what extent they support or oppose the protesters’ actions. We use a five-point Likert scale with a neutral midpoint that ranges from ’strongly oppose’ (=0) to ’strongly support’ (=1).Footnote ⁶

Study 2: Content Analysis of Newspaper Reporting

We turn to media coverage of environmental protests to better understand the asymmetric bias in support for environmental protest groups. Previous work suggests that media coverage and framing of protest events can affect public support for protest movements (McLeod and Detenber Reference McLeod and Detenber1999; Shultziner and Stukalin Reference Shultziner and Stukalin2021) and that different protest features, including protest tactics, protest size, and the protest group, can all have an influence on how these protests are portrayed in news coverage (Wasow Reference Wasow2020; Wouters Reference Wouters2013; Boyle et al. Reference Boyle, McLeod and Armstrong2012). We analyze how the pro-environment (climate activists) and environment-critical (farmers) protest groups were featured in two national-level newspapers in the Netherlands: broadsheet newspaper the NRC and tabloid newspaper De Telegraaf. Both newspapers are the most-read newspapers in their genre—broadsheet v. tabloid. We chose these newspapers because of the different demographics of their readership. On average, NRC’s readers are highly educated, high earners, and ideologically center- to left-leaning. On the other hand, De Telegraaf readers are, on average, more right-leaning, less educated, and represent a larger cross-section of the Dutch population (Bakker and Scholten Reference Bakker and Scholten2019; Leruth et al. Reference Leruth, Kutiyski, Krouwel, Startin, Caiani and Guerra2017).

We collected newspaper articles through the Nexis Uni database. We selected articles from 1 October 2019, when the first big farmer protest started in the Netherlands, until 20 June 2023, the end of our survey data collection. To find relevant articles, we used a separate search string for each protest group. These can be found in Appendix section K. Our query yielded 3,459 articles: 1,203 articles for our protesting climate activists query and 2,256 articles for our protesting farmers query. We excluded articles that were not about the Netherlands (n = 778), articles that did not mention the respective protest groups or only mentioned them in passing (n = 1,667), and articles that, while referring to the respective protest groups, did not mention any protests (n = 127). This leaves us with a final sample of 880 news articles to analyze.Footnote ¹⁵

In addition to news articles, we include columns, op-ed pieces, and letters to the editor in our analysis. Mass media provide a platform where protesters, targets, and third parties can interact (Koopmans and Olzak Reference Koopmans and Olzak2004). Columns, op-ed pieces, and letters to the editor are therefore important ’target’ and ’third-party’ responses (Wouters Reference Wouters2016) that are crucial to our analysis. While these pieces might not necessarily represent the views of editors, they have an important agenda-setting function (Sommer and Maycroft Reference Sommer and Maycroft2008) and strong persuasive effects on the general public (Coppock et al. Reference Coppock, Ekins and David2018). We therefore believe it is crucial to include them in our overall assessment of the newspapers’ coverage of protests.

Results From the Content Analysis

Figure 10 shows the count of articles per newspaper and protest group over time. In terms of overall reporting on environmental protests, the broadsheet newspaper NRC and tabloid newspaper Telegraaf show very similar over-time patterns, closely mirroring the ebb and flow of protest events. But in terms of volume, there is much more coverage of protesting farmers than of protesting climate activists in both the main tabloid and broadsheet newspapers. This mirrors the asymmetric bias we find in our heterogeneity analyses: farmers’ protests receive far more coverage than climate protests.

Figure 10. Over-time count of articles about protest groups across newspapers.

Next, we conduct a sentiment analysis of newspaper articles to examine if there are differences in coverage of protesting climate activists and protesting farmers. We code sentiment at the article level. We manually coded whether references to protesting climate activists or protesting farmers are predominantly negative (−1), predominantly positive (1), or neutral/balanced (0). More details about our exclusion and coding strategy can be found in Appendix section K.

We measured the reliability of the manual codings by conducting an interrater reliability test. We used Cohen’s K as a test statistic (two coders), and report a Cohen’s K of 0.65 (81.1 per cent agreement). This value reflects substantial agreement between two coders (Landis and Koch Reference Landis and Koch1977) and is in line with what we would expect based on other studies that examine sentiment in media content (Dunaway and Lawrence Reference Dunaway and Lawrence2015; Boukes et al. Reference Boukes, Van de Velde, Araujo and Vliegenthart2020).

Figure 11 shows the mean values of sentiment scores for protesting climate activists and protesting farmers in the main tabloid and broadsheet newspapers in the Netherlands. Both the within-newspaper difference in mean scores and the between-newspaper differences in mean scores are statistically significant.Footnote ¹⁶ Overall, the sentiment scores mirror the levels of protest support across subgroups of Dutch society. In the broadsheet newspaper, with a more center-left leaning readership, references to climate activists are balanced (μ = 0.09) while references to protesting farmers are somewhat negative (μ = −0.23). In the tabloid newspaper, with a more right-leaning readership, references to protesting climate activists are on average quite negative (μ = −0.45)) while references to protesting farmers (μ = −0.06) are mostly balanced. A key takeaway from Figure 11 is also that the differences in sentiment scores between the pro-environment (climate activists) and environment-critical (farmers) protest groups are small in the Dutch main broadsheet newspaper, but markedly different in the coverage of the main tabloid. Coverage of pro-environment protesters is considerably more negative in De Telegraaf. The media analysis thus shows that both the volume (more articles about environment-critical protests in both newspapers) and sentiment (the biggest tabloid newspaper being mostly negative about the pro-environment group) may contribute to the asymmetric bias in public support for environmental protests.

Figure 11. Sentiment scores for newspaper articles about protest groups. Sentiment scores range from −1 (fully negative) to 1 (fully positive). 0 represents a neutral/balanced mid-point. Thin bars are 95 per cent CIs, thick bars are 90 per cent CIs.

In addition to our quantitative sentiment estimates, we also provide a more qualitative account of how protest groups are covered in Dutch newspapers. In the broadsheet newspaper, some op-ed pieces emphasize the similarity between farmer-led and climate activists’ protest actions. For example, in one article, an interviewee points out that ‘the climate and farmers’ protests have major similarities. Both are well-organized and disrupt public order. An important difference is that climate activists protest nonviolently, while farmers not always do’ (NRC, 2 February 2023). Criticism of farmer protests in the NRC homes in on this: farmers seem to use more extreme actions than their pro-environment counterparts. In November 2019, a well-known NRC columnist, Tom-Jan Meeus, writes that ‘protests are sometimes accompanied by hardening and radicalization, and ominous signs of this are now being seen among farmers’. Meeus is worried that ‘high expectations have been raised, but it has long been clear that they cannot be met. And we know the price of excessive optimism: despair and extremism’ (NRC, 7 November 2019).

In De Telegraaf, articles are mostly critical of climate activists. While displaying more understanding of farmers’ concerns, they are not outright positive. Criticism of climate activists in some articles is often directed towards the perceived ‘globalist identity and lifestyle’ of protesters. For example, after an XR protest at Eindhoven Airport, a reader suggests that a ‘fitting punishment’ would be to ban these protesters from ever flying again: ‘only then will they contribute to the environment’ (De Telegraaf, 28 March 2023). Another op-ed piece criticizes the ‘predominantly white leaders of “Extinction Rebellion”’ who ‘grew up behind screens in peace and prosperity. The world was their village, Globalism their new mother tongue and a weekend in New York the crowning glory of their adolescence’ (De Telegraaf, 11 October 2019). The predicament of protesting farmers is sometimes also blamed on left-wing ‘elites’. Discussing the proposed nitrogen emissions reduction, a reader argues that ‘[t]hese are all decisions of the left-wing environmental mafia, Amsterdam elites, who think that chocolate milk comes from a cow with brown spots’ (De Telegraaf, 1 October 2019). Criticism of farmers, similar to NRC, often focuses on the extremity of farmers’ actions, while also being somewhat understanding of their position. In a sent-in letter, a reader writes that she ‘personally experienced the farmers’ anger yesterday and the behavior of these frustrated farmers was very dangerous. Their protest actions were very understandable to me, but to endanger traffic in such a way was downright criminal’ (De Telegraaf, 24 June 2022).

Additional Analyses

We further explore the negative bias towards pro-environment protesters in the tabloid newspaper De Telegraaf that we documented in the previous section. Our qualitative assessment suggested that both negative sentiment towards pro-environment protesters (climate activists) and more favorable attitudes towards environment-critical protesters (farmers) were often couched in anti-elite terms. This suggests that anti-establishment sentiment could be important for the asymmetric bias we have documented. Anti-elite rhetoric is a core rhetorical strategy of populist parties that pit the ‘normal people’ against vilified elites (De Vries and Hobolt Reference De Vries and Hobolt2020; Mudde Reference Mudde2004). Dutch right-wing populist parties, such as the Farmer-Citizen Movement (BoerBurgerBeweging, BBB) or the Freedom Party (Partij voor de Vrijheid, PVV), played a central role in public debate about environmental policy in recent years (Otjes and Krouwel Reference Otjes and Krouwel2022). In the run-up to the 2023 Dutch General Elections, they heavily politicized environment-critical stances and rallied behind the protesting farmers. We therefore hypothesize that supporters of right-wing populist parties will be more supportive of environment-critical protesters compared to the same type of protest activities staged by pro-environment protesters (Hypothesis 3).

To explore this link with right-wing populist parties, we run an additional model with marginal means where we split our sample based on whether respondents intended to vote for a right-wing populist party during the 2023 Dutch General Elections or not. These marginal means are shown in Figure 12.Footnote ¹⁷

Figure 12. Marginal means of protest features – split by right-wing populist vote or not. 0.5 implies a neutral stance. The ‘marginal mean differences’ are estimates of the differences between subgroup levels.

The marginal means in Figure 12 mirror the asymmetric bias we found for other subgroups (for example left–right, education, political trust, trust in science), but the magnitude is even larger. Among voters who did not vote for a right-wing populist party, mean levels of support across protest features show very small biases between protest groups. However, for respondents who intended to vote for a right-wing populist party during the 2023 General Elections, the biases between the pro-environment and environment-critical protesters are extremely stark. Overall, this group of voters opposes ( < 0.5) all pro-environment protests, including legal protests, while showing support ( > 0.5) for legal, disruptive, and even violent environment-critical protests. The marginal mean differences between subgroup levels are all statistically significant.

In a final step, we explore if we also find an asymmetric bias in respondents’ own thoughts on pro-environment and environment-critical protests. In our survey, we asked participants if they wanted to share their opinions about the protest vignettes they saw with the Minister of the Interior using the prompt below.Footnote ¹⁸

As a Dutch person, you can take action by telling the Minister of the Interior what action should be taken against the protests you have read about. On the next page you have the opportunity to share your opinion on the subject completely anonymously. We process comments from you and others, anonymously, in a letter that we send to the Minister of the Interior. Your comments, if you choose to share them, will be completely anonymous. Would you like to share your opinion about the protests with the Minister of the Interior?

In total, 52.1 per cent of respondents (1,316) chose to share their comments. We filtered out empty comments and those with fewer than fifteen characters. We then conducted the same sentiment analysis that we did for the media data. To do this, we subsetted on comments that contained references to climate activists (for pro-environment sentiment scores), farmers (for environment-critical sentiment scores), or both (n = 261). For each answer, we then coded whether references to these protest groups were predominantly positive (1), neutral (0), or predominantly negative (−1). We then grouped sentiment scores based on whether respondents intended to vote for a right-wing populist party or not (see footnote Figure 12). The sentiment scores are shown in Figure 13. These scores again mirror the strong asymmetric bias we find between populist voters and non-populist voters in Figure 12, and between tabloid and broadsheet newspapers in Figure 11: non-right-wing-populist voters’ comments about farmers and climate activists are very similar in tone, while right-wing populist voters’ comments have a strong negative bias towards climate activists, and even a strong positive bias towards farmers.Footnote ¹⁹

Figure 13. Sentiment scores open-ended comments about protest groups. Sentiment scores range from −1 (fully negative) to 1 (fully positive). 0 represents a neutral/balanced mid-point. Thin bars are 95 per cent CIs, thick bars are 90 per cent CIs.

These findings help place the pro-environment and environment-critical protests in the context of recent political developments in the Netherlands. These protests have deeply polarized Dutch society since 2019, and the asymmetric bias reported in public support for protest and in tabloid newspaper coverage might help to understand the electoral breakthrough of the right-wing populist party, BBB, amplifying skepticism towards pro-environmental policy and protests in the 2023 elections (Haan and Vermeer Reference Haan and Vermeer2023).