Foundations of climate change denial: Anti-environmentalism and anti-science

PLOS One | https://doi.org/10.1371/journal.pone.0334544 November 12, 2025 5 / 23 makers) about AGW, and discovered they fell into three main categories reflecting trend, impact and policy denial--each of which had a number of sub-categories [7, p. 510]. The main claims and common examples of each are: (1) “The eviden- tiary basis of global warming is weak and even wrong” (e.g., “The scientific evidence for global warming is highly uncer- tain”), (2) “Global warming would be beneficial if it were to occur” (e.g., “Global warming would improve our quality of life”), and (3) “Global warming policies [designed to reduce GHG emissions] would do more harm than good” (e.g., “Proposed action would harm the national economy”). During this early stage of denial (1990–1997), there was so much emphasis on trend denial that attribution denial had not yet emerged as a distinct counterclaim. But as apparent from the Second and Third IPCC reports (in 1996 and 2001, respectively), scientific evidence for human contributions to global warming rapidly grew in the late 1990s, making attri- bution denial more common as Rahmstorf [29] noted based on arguments he often encountered as a prominent climate scientist. Boussalis and Coan [15] reported a major expansion and update of McCright’s and Dunlap’s analysis [7]). They col- lected over 16,000 documents from 19 North American CTTs from 1998–2013, and then used unsupervised computational modelling to machine code key themes and empirically identify topics in this data. Over this period, they found that CTTs

increased their production of climate denial documents “exponentially” (p. 97). They also found, as had prior analyses of denial books [27] and of conservative columnists’ op-eds [48], that the production of denial misinformation was not steady or random but responded to external events like pending AGW legislation. Boussalis and Coan identified 47 topics that they classified as dealing with either “science” or “politics and policy” and found they fell into five topic clusters consist- ing of “policy & regulation,” “domestic and international politics” and “energy and emissions” (politics and policy domain) and “climate science” and “scientific integrity” (science domain) [24, p. 93]. Furthermore, the pattern among the clusters mapped onto a two-dimensional space made sense, with the three policy-relevant clusters being close to one another but distant from the science cluster while the scientific integrity cluster is as close to the domestic and international politics cluster as it is to the science cluster–suggesting the political nature of questioning the integrity of climate scientists and institutions. Scientific integrity was a newer theme that became especially prominent with the “Climategate Controversy” in 2009– 2010, and Boussalis and Coan [24] describe it this way: Not only do topics dealing with scientific misconduct—both regarding scientists themselves, the scientific consensus on AGW, and the IPCC in general—form their own distinct cluster, the language used seems to have more in common with

politics than science; that is, scientists are presumed to wield ‘junk science’ to achieve political aims” (94-95). More generally, they report that despite claims by some commentators that AGW doubt was declining, the levels of doubt about both climate science and policy expressed by the CTTs remained strong throughout the 16 years they examined. Finally, while their sophisticated methodology allows for the analysis of denial claims in huge bodies of data, Bousallis and Coan conclude by noting that, “… more detailed qualitative research is necessary to determine the prevalence of both new and past claims” (p. 98), which is exactly the lacuna we work to fill. Farrell [49] conducted an even more comprehensive analysis focusing on 40,785 texts produced by the above- mentioned 164 AGW countermovement organizations including far right foundations, industry trade associations, lobby- ing firms as well as CTTs covering twenty years (1993–2013). Besides using large-scale computational text analysis to uncover major topics emphasized in this huge corpus, he also examined the role of corporate funding on the amount and content of the texts and the degree to which their content made its way into major newspapers, Presidential addresses, and Congressional proceedings (with most success with newspapers and Presidential communications). Of most rele- vance to the present study, Farrell’s [49] structural topic modeling methodology produced a 30-topic solution and further analysis showed formed four clusters he described as “questions about scientific evidence” for global warming, public