Introduction

Climate change poses one of the most significant threats to human civilization, with far-reaching consequences for ecosystems, economies, and public health worldwide1. However, due to its global scale and gradual, ambient influence on the environment, climate change and its ramifications are difficult to subjectively assess2,3. This presents substantial challenges for scientists and communicators tasked with disseminating climate information to the public and motivating climate change mitigation, as concern for climate change is moderated by a myriad of cognitive, socio-demographic, and political factors4. Of these factors, political orientation has consistently emerged as the defining predictor of climate concern (which we define as the worry and importance afforded to anthropogenic climate change) within the United States5, with Republicans and conservatives expressing considerably more skepticism toward the reality of climate change and its associated risks6,7,8. This trend is also observed internationally, as similar patterns have increasingly emerged following the global escalation of political polarization9. Such ideological resistance to climate initiatives creates significant barriers to mitigation efforts10, generating an urgent need for interventions to improve support for climate policy across the political spectrum.

In examining the relationship between threat and political orientation, the dominant theoretical paradigms propose that exposure to any type of threat is associated with either an ideological shift towards conservatism11,12 or a blanket intensification of existing political preferences13,14. Simply, as climate-based threats are made salient, these models predict that political conservatives will maintain or strengthen their preexisting beliefs about climate change. These hypotheses appear to challenge the efficacy of any behavioral intervention designed to mobilize more climate-friendly attitudes or behaviors among conservatives, as acknowledging climate change and its consequences as legitimate threats is a necessary prerequisite to broad climate mitigation15,16. Recent literature, however, suggest that these seminal theories persistently neglect to consider the political and affective context of threats15,17,18,19,20 leading to an overrepresentation of threats that increase the appeal of conservatism [e.g., terrorism21,22]. Accordingly, we are aware of few studies that explicitly validate the potential for threat to induce attitudinal shifts towards liberalism, and only one that does so in an environmental context15, evidencing a need for more thorough investigations of the relationship between threat and political orientation both broadly and within the context of climate change18,23. This study addresses the ongoing demand for research in this area by examining the influence of threat perception and fear on climate concern across the political spectrum, offering both practical and theoretical insights into this previously neglected relationship.

A robust body of research indicates that political conservatives in the U.S. perceive lower climate risk, express more climate skepticism, and exhibit greater distrust in climate science compared to liberals6,7,24,25,26,27 and moderates28. In recent years, this trend has been increasingly mirrored internationally, with similar polarized attitudes observed across Canada29, Western Europe5, Latin America30, and Australia31, with the most pronounced effects occurring in countries with a left-right political divide analogous to that of the United States5,32. One explanation for this divide is differences in threat perception between liberals and conservatives across two dimensions—intent and extent19. While conservatives are primarily concerned by local and intentional threats with malicious intent, liberals are concerned by global threats resulting from passivity or inaction19,23. These differences may be a reflection of the values underlying conservatism and liberalism33,34,35, particularly regarding preferences for stability versus change, respectively11,36. Thus, ideological differences in threat prioritization may underlie the increased climate concern observed among liberals—given that climate change is a global threat perpetuated through indirect anthropogenic processes19,37. In line with this model, and to establish that results from a new online survey yield similar results to prior research, we hypothesize that compared to liberals, moderates and especially conservatives have lower levels of climate concern (H1).

Fear related to climate change appears to influence risk perceptions and policy support (for a summary, see ref. 4). Findings from both neurological38 and survey-based research39 reveal strong associations between risk perception and incidental emotion. Within the context of climate change, worry is associated with greater perceptions of environmental risks40,41, the importance of global warming as a voting issue42, climate policy support24, adaptive behavior43,44, and information seeking45,46,47. For some outcomes, climate concern remains a strong predictor of behavior even after accounting for political orientation42,44, suggesting that conservatives who adopt more liberal attitudes towards climate change will reflect their beliefs with commensurate behaviors. However, due to the abstract nature of climate change, galvanizing public concern for climate threats is difficult3 without emphasizing the relationship between local, tangible climate experiences and the broader phenomenon of global warming2,48,49.

Among climate hazards, environmental disasters have a unique affective potency owing to their capacity for substantial, proximal harms50,51. Drawing from relevant disaster literature (for a summary, see52,53), climate scientists typically categorize environmental disasters into two categories with respect to their causal structure54,55,56: natural disasters (e.g., floods, droughts, wildfires, hurricanes) and anthropogenic environmental disasters (e.g., pollution, oil spills, industrial accidents, dam failures). While natural disasters represent uncontrollable and unpredictable consequences of climate change, anthropogenic environmental disasters, by contrast, are controllable hazards that perpetuate the process of climate change57. Due to their inverse causal relationships to climate change and shared destructive capacities, both categories of disasters increase negative affect and concern for climate change—albeit through distinct mechanisms (for natural disasters, see refs. 58,59; for anthropogenic disasters, see refs. 49,55,60). While a comprehensive evaluation of the perceptual differences between natural and anthropogenic disasters is outside the scope of this paper, we nonetheless differentiate these constructs in accordance with the literature. Survey and experimental evidence overwhelmingly establish that the general population not only differentiates these disaster categories based on their perceived causes—ascribing ethical concerns to disasters directly involving human activity55—but additionally differentiates their ecological consequences, believing natural disasters to have less environmental impact than anthropogenic disasters61. Despite these differences, however, concern for both disaster categories is positively correlated with ecological risk perceptions61, suggesting a relationship to climate concern. In agreement with this line of research, we hypothesize that fear of natural and anthropogenic disasters is associated with increased levels of climate concern (H2).

A primary function of political systems is their ability to address societal threats, with political ideologies providing frameworks for how these threats are prioritized and resolved19. The study of individual ideological preferences, then, necessitates an understanding of how threats—and the solutions afforded by political parties—are perceived19,62,63. The dominant theoretical perspectives examining this relationship have described a universal association between threats and preferences for political conservatism64,65,66. Recent research, however, increasingly challenges these seminal models, suggesting they empirically fail to replicate20,67,68, draw overgeneralized conclusions69, and disproportionately use threat stimuli that privilege conservative values15,21,22,70. Instead, modern perspectives have begun to adopt models that understand the relationship between threat and ideology as contextual to both the type of threat and the capacity for an ideology to resolve the threat17,64,70,71, demonstrating an association between certain types of threat and preferences for political liberalism15. As left-wing parties are internationally perceived as responsible for resolving climate-based threats72, these models predict that individuals who feel threatened by environmental disasters will adopt more liberal climate attitudes—as a threat is inherently accompanied by unpleasant emotions that motivate individual actors to find a solution18,19,46,63.

Prior research into the influence of emotion on threat perception demonstrates that fear increases perceptions of risk and vulnerability73,74, especially within the context of climate-based threats49,75. As discussed above, conservatives who experience high levels of climate fear may be motivated to adopt more liberal attitudes towards climate change in response to the perceived threat, even if it requires the endorsement of ideologically inconsistent beliefs15,18,19,76,77. By extension, fear is unlikely to shift the attitudes of liberals and moderates, who ostensibly already hold liberal attitudes towards climate change7,28 and behave accordingly44. Given the growing body of evidence supporting the ability of threats to induce liberal attitudes, we hypothesize that the positive association between fear of environmental and natural disasters and climate concern is more pronounced for moderates and especially conservatives relative to liberals (H3).

To address our hypotheses, we utilize data from the SPEER23 Survey78—a comprehensive online survey administered by Qualtrics. Participants were recruited from existing Qualtrics panels using an “opt-in” sampling strategy. Panelists were recruited from a variety of sources and received compensation in accordance with the terms of their agreement with Qualtrics. Panelists were invited to participate via email or through the Qualtrics platform. The data collection process adhered to the relevant guidelines and regulations set forth by the University of Oklahoma Institutional Research Board, and followed the protocols outlined in Approval #15823. All participants provided informed consent prior to their participation. The survey respondents comprised 2188 adults (aged 18 and above) residing in the United States. Qualtrics employed quota-based recruitment strategies to ensure that the sample approximated a representative sample of U.S. adults, aligning with census benchmarks for age, gender, race/ethnicity, education level, and geographic region. Consequently, the data set offers a large and diverse sample that is generally representative of the broader U.S. adult population.

Results

Tables 1 and 2 present the results of regression analyses for the relationship between fear and political orientation across the four domains of climate concern. Visual depictions of the distributions of the dependent variables and fear scales are displayed in Figs. 1 and 2. Across all baseline results (shown in Models 1a, 2a in Table 1 and 3a, and 4a in Table 2), compared to liberal orientation, moderate, and especially conservative orientation is negatively associated with all four climate concern dependent variables (supporting H1)—which is consistent with past literature. Baseline results also show that fear of environmental disasters is positively associated with all four climate concern domains—more so than any of the other fear variables. Fear of natural disasters is positively associated with climate risk, importance, and worry, but not belief in climate change alone. These results largely support Hypothesis 2. Interestingly, fear of strangers and fear of crime both negatively predict endorsement of climate risk and importance (albeit weakly), potentially due to their association with typically conservative threats15. Fear of human-made disasters and fear of personal tragedy are unrelated to climate concern.

Table 1 OLS regressions for belief and risk
Full size table
Table 2 OLS regressions for importance and worry
Full size table
Fig. 1: Each dependent variable has four categories.
Fig. 1: Each dependent variable has four categories.
Full size image

The distribution for each dependent variable is mostly normally distributed. The distribution for the climate belief dependent variable displayed in (a) is mostly normal with only a slight negative skew. Similarly, the distribution for the climate risk dependent variable in (b) is mostly normal with a slight negative skew. The distribution for the climate importance dependent variable displayed in (c) is only somewhat normally distributed. The distribution for the climate worry dependent variable displayed in (d) is mostly normal.

Fig. 2: Each fear scale has ten categories.
Fig. 2: Each fear scale has ten categories.
Full size image

The distribution for each dependent variable is mostly normally distributed. The distribution for the fear of natural disasters scale displayed in (a) is normal. The distribution for the fear of environmental disasters scale displayed in (b) is mostly normal with a slight negative skew.

Models 1b and 2b in Table 1 and 3b and 4b in Table 2 show interaction results for fear of natural disasters and political orientation on climate concern. Across all analyses (and in line with Hypothesis 3), significance tests for the slopes of the interaction terms show that fear of natural disasters increases climate concern among conservatives relative to liberals. To ease interpretation, we display the interaction results graphically. Results in Fig. 3 show that natural disaster fear has little effect in motivating liberals and moderates to adopt higher levels of climate concern across all four dependent variables. However, as hypothesized, fear of natural disasters greatly bolsters climate concern for conservatives. While regression results for interaction terms can indicate significance in the slope of a particular interaction effect, further testing is needed to test for moderation at specific levels of the independent variables. For instance, while the coefficient for the interaction effects between moderates and liberals when interacting with natural disaster fear is only significant for one dependent variable (climate importance), marginal tests performed at each level of the natural disaster fear scale reveal significance between liberals and moderates at lower levels of fear relative to higher levels of fear for the other dependent variables.

Fig. 3: Climate concern often varies across political orientation as disaster fear increases.
Fig. 3: Climate concern often varies across political orientation as disaster fear increases.
Full size image

Results displayed in a show that climate belief is roughly even across all levels of disaster fear for liberals and moderates.Conservatives, however, have higher levels of climate belief as disaster fear increases. The same basic pattern is shown in b with predicted levels of climate risk. Results in both c, d, demonstrate that disaster fear increases climate importance and climate worry for all political orientations—however, the positive effect of disaster fear on climate importance and worry is greater for conservatives relative to liberals and moderates. Overall, interaction plots for fear of natural disasters and political orientation show convergence among political orientations at higher levels of disaster fear when predicting climate concern.

To empirically test how political orientation moderates the effect of natural disaster fear on climate concern, we conduct a series of post-hoc tests for marginal estimates across varying levels of the fear scales. For all four measures of climate concern, all three orientations significantly differ at low and middle levels of concern, with liberals exhibiting the highest degree of concern for all dependent variables, followed by moderates and conservatives. Climate importance no longer significantly differs between liberals and moderates at a value of eight on the natural disaster fear scale and a value of nine for the other three dependent variables. Moderates and conservatives no longer differ in climate belief and importance at a value of eight on the disaster scale and no longer differ in climate risk and worry at a value of nine. Liberals and conservatives show no significant difference in climate belief and importance at a value of nine on the natural disaster scale and are no different in climate risk and worry at the highest level of fear.

To further demonstrate the relative convergence on climate concern for conservatives relative to liberals and moderates, we also report predicted scores on all dependent variables at the 25th and 75th percentiles for the natural disaster fear scale. Marginal estimates for the climate belief scale range from 2.59 (conservatives) to 3.16 (liberals) at the 25th percentile on the natural disaster fear scale. At the 75th percentile, the marginal estimates are much closer—ranging from 2.82 to 3.09. Marginal estimates for the climate risk scale at the 25th percentile on the disaster fear scale range between 2.45 (conservatives) and 3.06 (liberals). At the 75th percentile, marginal estimates range from 2.70 to 3.04. At the 25th percentile on the disaster fear scale, marginal estimates fall between 2.39 (conservatives) and 2.86 (liberals). The marginal estimates are much closer at the 75th percentile—ranging between 2.64 and 2.77. Finally, marginal estimates for climate worry at the 25th percentile of the disaster fear scale fall between 2.17 (conservatives) and 2.68 (liberals). At the 75th percentile, marginal estimates are between 2.44 and 2.76. To summarize, as natural disaster fear increases, political orientation matters less in predicting climate concern.

Interaction results for the slope of fear of environmental disasters and orientation on climate concern are less consistent and only somewhat supportive of Hypothesis 3. Results on the effects of the slopes of the interaction terms in Models 1c and 2c of Table 1 show that fear of environmental disasters significantly increases belief in climate change and climate risk among moderates and conservatives relative to liberals. Graphical interpretations of interaction effects displayed in Fig. 4 demonstrate that fear of environmental disasters increases scores on climate belief and risk more starkly for conservatives (and moderates to some degree) relative to liberals. Results for tests of significance between slopes of the interaction terms shown in Models 3c and 4c of Table 2 only partially support Hypothesis 3. Fear of environmental disasters does not affect climate importance and worry differently between liberals and conservatives. The slope of the interaction term for moderates, however, shows marginally higher scores on climate importance for moderates relative to liberals as fear of environmental disasters increases. The findings for the interaction effects of environmental fear are graphically depicted in Fig. 4.

Fig. 4: Climate concern somewhat converges between political orientations as environmental fear increases.
Fig. 4: Climate concern somewhat converges between political orientations as environmental fear increases.
Full size image

Results displayed in a demonstrate that environmental fear increases climate belief for all political orientationss, though more prominently for conservatives. b shows that climate risk increases for all political orientations as environmental fear increases, but more so for conservatives and especially moderates relative to liberals. Results displayed in c show a similar positive relationship between environmental fear and climate importance for liberals and conservatives, while the effect is more pronounced for moderates. Results displayed in d show that climate worry increases slightly more for moderates compared to liberals and conservatives as environmental fear increases.

Results of marginal effects again provide nuance when evaluating how political orientation interacts with environmental fear in predicting climate concern. Results on climate belief show significant differences between all three political orientations at low and middle levels of environmental fear, with liberals scoring highest in climate belief, followed by moderates and conservatives. Climate belief is no longer significantly different between liberals and moderates at a score of eight on the environmental fear scale. Moderates and conservates no longer significantly differ at a level of nine on the scale. Liberals and conservatives no longer differ on climate belief at a level of ten on the environmental fear scale. For both the climate risk and climate worry DVs, liberals score significantly higher than moderates at lower and middle levels of environmental fear and then converge at a score of eight on the scale. Liberals score significantly higher in climate risk and climate worry relative to conservatives across all levels of environmental fear, demonstrating no significant interaction. Results for climate risk show that liberals score significantly higher than moderates at low and middle scores of environmental fear and then no longer significantly different at a score of eight. Moderates and conservatives do not significantly vary at the lowest levels of environmental fear, however, moderates exhibit higher levels of worry at scores of three and above on the scale. Liberals demonstrate significantly higher levels of climate worry than conservatives across all levels of environmental fear, showing no interaction effect.

As with results for natural disaster fear, we report marginal estimates across dependent variables at the 25th and 75th percentiles of environmental fear to provide more context. Marginal estimates for climate change at the 25th percentile of environmental fear range from 2.44 (conservatives) to 3.03 (liberals). At the 75th percentile, marginal estimates range from 3.02 to 3.35. For climate risk, marginal estimates range between 2.26 and 2.87 at the 25th percentile for conservatives and liberals, respectively. At the 75th percentile, estimates range between 2.95 and 3.37. For climate importance, marginal estimates at the 25th percentile on the environmental fear scale fall between 2.21 (conservatives) and 2.66 (liberals). Marginal estimates range between 2.83 and 3.20 at the 75th percentile. Finally, marginal estimates for climate worry at the 25th percentile of environmental fear range from 1.99 for conservatives to 2.47 for liberals. At the 75th percentile, marginal estimates fall between 2.66 and 3.07. In sum, while not as robust as the findings on natural disaster fear, we find substantial political convergence on climate concern as environmental fear increases.

Discussion

In this study, we find compelling support for both Hypothesis 1 and Hypothesis 2. In line with a robust body of literature (e.g., refs. 7,8,24,28,79), we find that conservatives, on average, express far lower concern for climate issues relative to liberals and moderates. We also find that fear of both natural and anthropogenic environmental disasters is generally associated with increased climate concern—which is consistent with prior findings (e.g., refs. 55,58,59,75). Results for Hypothesis 3 yield mixed support. While conservatives experience a more pronounced association between climate concern and fear of natural disasters, climate worry, and climate importance do not differ much across political orientations when considering fear of environmental disasters. Despite this, as levels of fear increase, conservatives do experience more pronounced growth in climate belief and climate risk compared to liberals, suggesting differences in the affective versus cognitive assessments made of threats induced by environmental disasters.

One explanation for the positive relationships observed between fear of anthropogenic environmental disasters and the affective components of climate concern across all political orientations can be found within relevant disaster literature. A large body of research suggests that anthropogenic environmental disasters are perceived as markedly more threatening56 and severe55 than natural disasters due to the attribution of blame afforded by the causal involvement of humans53,54,60. In addition to fear, such risk assessments may instead be mediated by anger, which is known to increase perceptions of threat and motivate behavioral responses independently of other emotions15,52,60,63, increasing climate risk perceptions relative to natural disasters73,74. As emotions tend to co-occur, anger may represent a confounding variable in the relationship between fear of disasters and climate concern15,47. Accordingly, anger may explain the mixed support observed for Hypothesis 3, as the influence of anger on perceived threats from anthropogenic disasters may increase perceptions of climate risks across all political orientations. Recent research indicates that threats associated with particularly strong negative emotions may be more effective at motivating attitudinal shifts, as taking action to resolve a salient threat (e.g., acknowledging potential solutions, adopting new perspectives, etc.) relieves the associated psychological discomfort18,63. While this study did not directly assess for this relationship or provide a measure for strong negative emotion, our findings are nonetheless consistent with the literature, and suggest a need for further research capable of elucidating the influence of discrete emotion on climate threat perception19.

Theoretically, this study has important implications for ongoing investigations into the relationship between threat perception and political orientation. Our findings contribute to a growing body of literature challenging the seminal theoretical foundations of the present field (e.g12.), providing much-needed data synthesizing relationships between threat, emotion, and political orientation while confirming the capacity for threat to induce preferences for aspects of liberalism18,19,23. Generally, existing literature synthesizing threat perception and political orientation has demonstrated extreme bias toward traditionally conservative threats, leaving a large gap in the literature concerning the effects of liberal threats on political attitudes15,18. We observe that perceptions of environmental threats are significantly predictive of climate concern among conservatives while having little influence on liberals, indicating a shift towards liberal climate attitudes among conservatives at the highest levels of fear. This finding provides support for the ideology-affordance model of threat and politics63, which proposes that experiencing threat motivates an attitudinal alignment with the political ideology perceived as most capable of providing a resolution. In the context of climate change, this model demonstrates that conservatives who feel threatened by environmental disasters are motivated to adopt politically liberal climate attitudes, as liberals are seen as responsible for solving climate-related threats. While our findings are primarily relevant to the political landscape of the United States, research on issue ownership has revealed that the threats associated with different political parties are relatively stable cross-culturally, suggesting the ability of this model to retain predictive power internationally72. However, as research on the capacity for threat to evoke preferences for liberalism remains scarce, future studies should examine alternative threat stimuli relating to climate change, as well as investigate other historically neglected threats outside of a climate context that may elicit liberal ideological shifts18.

Practically, our findings provide direction for climate communicators and policymakers looking to increase climate concern among conservatives and increase their behavioral responses to climate change. Principally, understanding the circumstances under which shifts in political preferences can occur is paramount to developing effective interventions18. Our study presents evidence that conservatives can be motivated to increase endorsements of climate concern under circumstances of environmental threat. If interventions can be designed to assist conservatives in connecting their understandings of local, salient climate threats (e.g., disasters) to the broader phenomenon of global warming, our study suggests that this may be associated with increased climate concern2,48,49. Increased climate concern across the domains of belief, worry, risk perception, and importance are well-correlated with a variety of behavioral outcomes. Worry about climate change consistently predicts its importance as a voting issue, even after accounting for political orientation62, and is correlated with support for climate policies—as are climate risk perceptions and belief in climate change24,40,43. Perception of climate risks is strongly associated with adaptive behavioral intentions43. Belief in climate change is further associated with climate action for both conservatives and liberals44, as political polarization does not translate into differences in behavioral outcomes. Given the clear potential for behavioral outcomes, future research should seek to utilize the present findings to both design more effective interventions (e.g., messaging and awareness campaigns) and to confirm the relationship between climate concern and behavior across the political spectrum.

Climate communicators and policymakers should additionally be mindful of the potential risks associated with increasing concern for climate change in the absence of accessible behavioral responses. If individuals do not feel that climate-friendly behaviors are available, convenient, or effective—both personally and politically—they may not elect to change their behaviors even when experiencing high levels of concern about climate change (for a summary, see ref. 79). High levels of climate concern in the absence of a conspicuous solution may subsequently increase the belief that climate change is unstoppable, which is associated with reduced behavioral outcomes, including decreased desire for mitigatory policies80. Thus, groups or individuals seeking to increase behavioral responses to climate change should remain aware of the importance of providing accessible opportunities for pro-climate behavior in addition to any efforts to increase climate concern. Through outlining both proximal ecological threats to personal safety and encouraging the consideration of relationships between local experiences and climate change, our findings suggest that future climate initiatives may successfully increase climate concern among historically resistant groups, such as political conservatives. Pairing this concern with explicit behavioral recommendations or instruction may be a fruitful avenue for future research.

Given the nature of the data, this study has some additional methodological and statistical limitations worth considering. To begin, there are noted discrepancies in samples between traditional secondary surveys (e.g., General Social Survey and American National Election Survey) and online, opt-in surveys81. While Qualtrics met most demographic targets within range to provide an approximated representative sample, we note that our sample still skews toward liberals and undercounts conservatives and Republicans—an issue that is common with many online surveys and political polls82. Additionally, we are limited by the measures available. For example, 37 percent of respondents identify as politically independent. Other research83 notes that independents usually lean toward one party or the other—meaning very few Americans are truly independent. We have no measure to discern those who lean in an ideological direction from true independents, thus there may be some political misclassification for our political affiliation control. Furthermore, the available survey items only allow us to evaluate attitudes toward climate concerns. Survey questions that assess actual behaviors would provide more meaningful measurement. Finally, while we find statistical significance for our interaction terms, the interaction models do not add much-explained variance to the models as a whole. Much of the variance in climate concern appears to be attributed to the direct effects of the independent variables in the baseline models. The interaction models explain nuances within the focal relationships but do not substantively add to the model fit.

Ultimately, understanding the psychological dynamics that influence political preferences is crucial for consolidating social support for climate action. By embracing nuanced approaches that acknowledge and address individuals’ fears and concerns, researchers can lay the groundwork for meaningful collective efforts to mitigate climate impacts. This study offers valuable and novel insights into the intricate interplay between fear, political orientation, and climate concern, illuminating promising avenues for future research initiatives.

Methods

We use four linear dependent variables to measure varying dimensions of climate concern. Adapted from prior climate opinion surveys84,85, these variables measure belief in anthropogenic contributions to climate change (climate belief), perceptions of risk to people and the environment (climate risk), personal importance of climate change (climate importance), and worry about climate change (climate worry). These constructs are frequently used within climate research due to their ability to capture varying cognitive and experiential factors that independently influence climate threat perception and behavioral outcomes4,24,42, allowing for more thorough associations to be drawn with the extant literature. All four dependent variables are linear and were recoded to a scale of 1-4 to achieve a Gaussian or normal distribution. A correlation matrix for all DVs is provided in Supplementary Table 3. A full description of survey item wording and general coding for all variables can be found in Supplementary Table 1 and descriptive statistics for all variables are provided in Supplementary Table 2. Distributions for dependent variables and fear scales can be found in Table 1 and Table 2, respectively.

Independent variables include measures for fear and political orientation. We measure self-reported fear across six established domains based on items from Chapman University’s Survey of American Fears86. To comprehensively account for various types of fears, including those beyond environmental and climate concerns, our study assesses fear across six well-established domains. These domains were determined based on items from Chapman University’s Survey of American Fears86. Each of the six fears is measured with linear self-placement scales. Interaction terms are only created using the scales for disaster fear and environmental fear scales. We refer to environmental disasters as those that are caused by or related to human activity that primarily result in damage to the environment. In doing so, we draw from disaster classifications developed by risk and climate scientists to distinguish human-caused disasters from natural disasters, as well as use the Survey of American Fears87 in our separation of disasters harming the environment (“Environmental Disasters,” e.g., pollution, industrial accidents) from those that primarily impact human populations (“Man-Made Disasters,” e.g., war, terrorism, pandemics). Separating disasters by consequence as well as causal structure is necessary to focus our analysis on only disasters related directly to the environment, thus avoiding confounding influences on their relationship to climate concern. Environmental fear is modeled with fear of disasters for all analyses. A bivariate correlation test (r = 0.616) indicates a strong correlation between the two types of fear. Given this correlation, a post-hoc test of multicollinearity (performed on the baseline model without interaction terms) shows an overall mean VIF value of 1.61. VIF values for the environmental and disaster scales are 2.21 and 2.13, respectively. This finding demonstrates that the variance of the coefficients is inflated by 121% and 113%, respectively, due to multicollinearity with other variables in the model. These results demonstrate no serious issues with multicollinearity in the model.

The original political orientation variable assesses self-placement on a 7-point scale from “very liberal” to “very conservative”. Responses were coded into three dummy variables for liberal, moderate, and conservative by collapsing the liberal and conservative endpoints together—leaving the moderate category for respondents who indicated no ideological “lean.” Extant climate literature generally categorizes political orientation along a liberal-conservative dichotomy, with little explicit consideration given to political moderates. We include moderates for more robust analysis and to address a gap in the literature on politics and climate attitudes.

Interaction terms are created by multiplying scales of fears of both natural disasters and environmental disasters by moderate and conservative dummy variables (with liberal as the reference category). In each model, we include a full array of control variables for political party, age, sex, race, socio-economic status, family status, religious variables, and subjective weather perception (see Supplementary Table 1 for more detail on coding).

Analyses are conducted using OLS regressions modeled in Stata. Baseline models report regression coefficients for all focal and control independent variables. Subsequent models include regression effects for interaction terms between political orientation and both fear of natural disasters and fear of environmental disasters. Graphical figures are displayed to aid in the interpretation of significant interaction effects based on model results. We replicate OLS regressions with ordinal logistic regressions. Given that the results are remarkably consistent, we employ OLS regression for ease of interpretation and to more accurately report the amount of variance explained in each model.