The power of protest in the media: examining portrayals of climate activism in UK news

The UK climate activism landscape has expanded rapidly in recent years with groups such as Extinction Rebellion (XR), Just Stop Oil, Insulate Britain, and Animal Rising employing non-violent civil disobedience to draw attention to climate inaction. In January 2023 XR announced "We Quit," temporarily shifting away from publicly disruptive tactics to focus pressure on elites in finance, media, and government. In contrast, peer groups continued disruptive actions aimed at public attention. This divergence raises key questions about how different tactics (e.g., legality, targets such as public vs. elites, and scale) shape media coverage. Given media’s role in agenda setting and shaping public perceptions, the study investigates the relationship between protest tactics and both the volume and sentiment of UK news coverage. The research addresses two pre-registered hypotheses: (H1) liberal newspapers report more positively on actions than conservative newspapers; and (H2) illegal actions receive more press coverage but more negative sentiment. The study also explores whether shifting to less confrontational tactics changed sentiment, and how targets, location, and partisanship affect coverage.

The literature establishes that news media coverage remains crucial for social movements’ influence despite the rise of social media. Coverage helps movements communicate frames, gain legitimacy, and pressure elites. However, few movements receive coverage, and the relationship between tactics and coverage is context-dependent. Research shows insider, non-confrontational tactics can draw more coverage, while experimental evidence identifies an activist’s dilemma where extreme tactics increase attention but reduce public support. Debates persist on whether public support matters relative to imposing costs on elites. XR has historically reflected on strategy, sometimes misapplying academic findings (e.g., the 3.5% rule) and recently repositioned toward a perceived moderate flank relative to more radical groups like Just Stop Oil, potentially benefiting from radical flank effects. Protests can increase overall climate coverage, and experimental work links protest coverage to shifts in public attitudes and elite responsiveness. Activists often choose tactics to maximize coverage. Visual portrayals of climate protest have evolved (the "Greta Effect"). Open questions include how legality, targets, geography, and media partisanship shape extent and tone of coverage, and whether moderate repositioning improves media sentiment.

Design and scope: The authors compiled 412 online news articles from the BBC and the 15 most popular UK newspapers (YouGov 2023) from November 1, 2022 to May 4, 2023, spanning two months before and four months after XR’s "We Quit" statement. While centered on XR, the dataset also includes articles that mentioned XR but focused on other groups, and targeted collection around two April 2023 actions by Just Stop Oil (snooker) and Animal Rising (Grand National) for a case study. Hypotheses (pre-registered): H1: Liberal newspapers will report more positively on actions compared to conservative newspapers. H2: Illegal actions will get more press coverage, but coverage will be more negative. Variables and coding: Key variables included partisanship (liberal/neutral/conservative; coded -1 to 1), word count, percentage coverage devoted to XR, comments/shares (where available), sentiment (coded -1 to 1 via hand dictionary coding), legality (0 illegal, 1 legal), target (Public, Elite, Government, Industry), scale (Local, London, National-coordinated), accuracy (count of major factual errors), action indicator, and temporal controls (day of week, month). "Press Coverage" was defined as an interaction of article word count and percentage devoted to XR. Primary coding was conducted by the first author; XR-specific event variables were secondarily coded by an XR member; intercoder agreement > 0.9 for those variables. Analytic approach: Ordinary Least Squares (OLS) regressions tested associations between predictors and (a) sentiment and (b) press coverage. Robustness checks included models with month and day-of-week controls. Analyses were conducted in R 4.0.0. H1 used Sentiment ~ Partisanship + Month + Day. H2 used (i) Press Coverage ~ Legality + Scale + Target + National Event + Partisanship + Month + Day and (ii) Sentiment ~ Legality + Scale + Target + National Event + Partisanship + Month + Day. Sub-analyses were run on (a) all articles, (b) XR-focused articles (>50% coverage), and (c) samples excluding articles about "The Big One" (TBO). Exploratory analyses included (1) Accuracy ~ Partisanship + Month + Day, (2) Welch t-tests comparing mean sentiment before vs. after "We Quit", and (3) a case study comparison of three April 2023 actions (TBO; Just Stop Oil’s snooker protest; Animal Rising’s Grand National protest) by total articles, total word count, and mean sentiment. Deviations from preregistration: Due to limited and non-representative comment/share data (~15% of articles), the planned "Cumulative Event Impact Score" was replaced with the Press Coverage measure. Month and day controls were added post-preregistration. Additional exploratory analyses (accuracy by partisanship; mean sentiment pre/post "We Quit") and the April 2023 case study were included.

- Hypothesis 1 (H1) supported: Conservative outlets were significantly more negative toward climate activism than liberal or neutral ones. In the full sample, conservative newspapers’ sentiment was -0.247 points vs neutral and -0.521 points vs liberal on a -1 to 1 scale; liberal outlets were more positive than neutral. In XR-focused coverage, conservative outlets were -0.513 points more negative than liberal outlets; the conservative vs neutral difference lost significance. - Hypothesis 2 (H2) mixed: Contrary to the expectation that illegal actions would garner more press coverage, legal actions attracted more press coverage on average (significant in the full sample) and were covered more positively (significant in the full sample; legal actions’ sentiment +0.259 vs illegal). Thus, legal actions received both more and more favorable coverage. - Targets and geography: - Industry-targeted actions drew more coverage than other targets (significant in two models). - Public-targeted actions received more favorable sentiment than other targets. - London-based actions attracted significantly more coverage than actions elsewhere; nationally coordinated actions sometimes had more positive sentiment, but this was largely driven by The Big One and lost significance when TBO was excluded. - Partisanship and volume: Both conservative and liberal outlets tended to publish more on climate activism than neutral outlets (some evidence in press coverage models). - Accuracy by partisanship (exploratory): Conservative newspapers made more factual errors on average than liberal or neutral outlets, by about 0.2 to 0.315 additional errors depending on model; no significant difference between liberal and neutral outlets. - Sentiment before vs. after "We Quit" (exploratory): Mean sentiment did not improve after XR shifted tactics. Depending on subset, it was statistically indistinguishable (before -0.125 vs after -0.2064; p=0.25) or slightly lower post-statement (before -0.125 vs after -0.271; p=0.033 in XR-focused subset). - Case study of April 2023 actions: - Coverage volume (articles): Grand National 57; Snooker 44; TBO without marathon 40; TBO with marathon 53. - Total word counts: Grand National 43,109; Snooker 29,755; TBO without marathon 24,852; TBO with marathon 32,068. - Mean sentiment: Grand National -0.214; Snooker -0.364; TBO without marathon -0.175; TBO with marathon -0.038. - Interpretation: More disruptive public-facing actions (Grand National, Snooker) garnered as much or more coverage than TBO but with more negative sentiment; all events had negative mean sentiment overall, with the TBO+marathon subset closest to neutral.

Findings indicate that media partisanship strongly shapes the tenor and accuracy of coverage of climate activism, with conservative outlets more negative and less accurate on average. Contrary to the activist’s dilemma, legal actions both received more coverage and were framed more positively, suggesting that law-abiding tactics can simultaneously increase attention and favorable tone in legacy media. Geography and target matter: London-based actions received more coverage, indicating a London-centric media bias; industry-targeted actions drew more coverage, possibly due to inherent newsworthiness or clear antagonists; public-targeted actions garnered more favorable tone, while government-targeted actions tended to be more negative. The Big One’s overlap with the London Marathon appears to have boosted positivity for that subset, illustrating how coincident events can influence coverage beyond a single news cycle. Despite XR’s move away from disruptive tactics, average sentiment did not improve, suggesting inelasticity in media framing amid a landscape dominated by conservative outlets. These results provide strategic insights for activists balancing the goals of maximizing attention versus securing favorable narratives, while highlighting the complex interplay between tactics, targets, location, and outlet partisanship in shaping media responses.

If the primary goal is maximizing media attention, disruptive demonstrations in London appear most effective. If the goal is more positive coverage, legal, nationally oriented actions targeting the public provide a plausible strategy. Overall, conservative outlets’ dominance contributes to a negative tilt in coverage of climate activism. The study expands evidence on how tactics, targets, geography, and outlet ideology influence both the extent and affective tone of coverage. Future work should examine social media dynamics, extend cross-nationally (e.g., to the U.S.), and explore more nuanced measures of disruptiveness and outcomes beyond media sentiment, such as elite responsiveness and policy change.

- Media scope: Focus on legacy UK outlets (BBC and top newspapers) may not capture social media dynamics or local news ecosystems; however, legacy media still reaches large audiences and influences political elites. - Disruptiveness measure: The study did not include an objective disruptiveness rating separate from legality, potentially omitting an important dimension. - Reaction data: Limited availability of comments/shares (~15% of articles) precluded use of the preregistered Cumulative Event Impact Score and constrained analyses of public reactions. - Timeframe and events: The 7-month window and coincident events (e.g., TBO overlapping the London Marathon) may influence generalizability; sub-analyses mitigated but did not remove this concern. - Geographic concentration: A London-centric media bias may limit generalizability to actions outside London. - Country context: Results are specific to the UK media environment and may differ elsewhere.