Seth A. Rosenthal

As climate change increasingly harms public health, these impacts have potential to catalyze stronger public support for policies that protect the climate and public health. Prior research has examined Americans’ views about the health harms of climate change, but the current state of Americans’ views is unknown. This article presents a longitudinal comparison based on a 2014 study, mapping public perceptions among politically partisan audiences. We analyzed data from a 2024 nationally representative sample of U.S. adults ( N = 1013), comparing it to previous nationally representative surveys conducted in 2014 ( N = 1275) and 2018 ( N = 1114). Relative to 2014, we find that in 2024 more Americans are able to identify specific health harms caused by climate change (+5%). They are also more aware that climate change is harming other Americans’ health (+8%); various health harms will become more common in their communities over the next 10 years (+26%); the health impacts of climate change will disproportionately harm certain communities (+13%); and that natural gas harms health (+8%). They are also more trusting of various sources of information about climate and health, especially their primary care doctors (+11%), and more supportive of federal agencies increasing their efforts to protect people from the health harms of climate change (+7%). However, more Americans now incorrectly believe that wind (+9%) and solar (+7%) harm health. These increases in Americans’ awareness of climate and health are driven largely by Democrats, with Republicans’ awareness generally remaining constant or decreasing. We discuss the implications of these results for climate-health communicators and researchers.

As climate change intensifies, extreme heat events are becoming more frequent and severe, posing growing risks to human health, ecosystems, and infrastructure. In a large-scale survey experiment (N = 3,960 U.S. adults) we examine how the emotional valence of imagery—negative, neutral, or positive—in social media-based news about heat waves influences worry about heat waves, beliefs about climate change’s role in extreme heat, and willingness to share information about heat waves. Results show that identical messages accompanied by either negative or neutral imagery increased worry about heatwaves and strengthened beliefs that climate change is exacerbating heat waves, whereas the same messages accompanied by positive imagery did not. The influence of different images varied within each emotional valence category, highlighting the importance of stimulus sampling. These findings underscore the importance of strategic visual framing in climate communication, especially in multimodal, digital contexts.

Climate change impacts everyone, but those impacts are not distributed equally. Frontline communities (FLC) face disproportionate climate change harms, but that may not entail greater worry about them. Worries shape community priorities, and understanding these worries can improve engagement strategies. Here we integrate Climate Change in the American Mind project (n = 6,183) data with the U.S. Climate and Environmental Justice Screening Tool to explore climate change attitudes among people located in FLC compared with those not in FLC. People in FLC and non-FLC are equally worried about global warming. However, people in FLC are slightly more worried about several climate change impacts and are less likely to think there is scientific consensus that global warming is happening. Geographic analysis shows where FLC are most and least worried about global warming. This study shows a need for strategic communication in FLC regarding the connections between climate impacts and climate change.

Reducing climate change requires persuading people about climate change and climate solutions. The nearly universal way of evaluating which climate and environmental messages are most effective is to calculate and compare average treatment effects (ATEs). The problem with the ATE, however, is that it does not describe the underlying pattern of persuasion: whether effect size is better predicted by how many people are persuaded (i.e., breadth), or how much they are persuaded (i.e., depth). Here, we investigate this in a preregistered meta-analysis of our experimental data archive, avoiding publication bias (14 experiments; 94 treatments; total N = 41,265). We find that as treatment effects increase, the breadth and depth of persuasion both increase. However, the predictive power of breadth (𝛽 = 0.47, 95% CI [0.29, 0.65]) is far stronger than that of depth (𝛽 = 0.10, 95% CI [0.04, 0.15]). We also find that breadth and depth are only weakly correlated (𝛽 = 0.23, 95% CI [0.003, 0.46]), confirming they are distinct. Additionally, as average treatment effects increase, we also find that negative (i.e., backlash) effects decrease (𝛽 = -0.30, 95% CI [-0.46, -0.15]), suggesting that larger ATEs are a function of increased persuasion and decreased backlash. Through three case studies, we show how these results generalize and when they can be audience-dependent. These results have important implications for both scholars and practitioners of climate and environmental communication.

Extreme weather events are becoming increasingly frequent and severe worldwide. Emotional messages might be persuasive in communicating this information, but it remains unclear whether message efficacy depends more on emotional content or delivery. In a pre-registered experiment (N = 3,463), participants were randomly assigned to listen to one of seven audio messages: a control message or one of six messages on climate change’s impact on extreme weather, varying in emotional content (neutral, semi-emotional, emotional) and delivery (neutral vs. emotional). Compared to the control, all messages were equally effective at increasing belief in climate change’s impact on extreme weather, support for government action, and worry about extreme weather and climate change. However, participants were more willing to share the ‘neutral content + emotionally-delivered’ message than the same message delivered in a neutral tone. These findings provide valuable insights for scholars and practitioners on the role of emotions in climate change communication.

Air conditioning (AC) access is a critical adaptation to rising temperatures and extreme heat under climate change. Yet, fine-scale, national estimates of AC prevalence—and the social and geographic factors that shape access—remain scarce. Here, we use an original, nationally representative survey (n = 11,235) combined with high-resolution demographic, administrative, and remote sensing data to estimate AC access at state, core-based statistical area (CBSA), and county levels across the contiguous United States. Applying multilevel regression and post-stratification (MRP) with machine learning optimization, we map central AC, window units, and lack of AC, and validate estimates using four independent sources. We find sharp regional disparities: while 86% of counties report at least 90% access to some form of AC, prevalence remains low in higher-elevation western regions, particularly the Pacific Northwest and Rocky Mountains. Our analysis reveals that socioeconomic status, housing stock, and local climate are key predictors of AC access. These patterns expose an emerging adaptation gap, with implications for heat vulnerability, energy demand, and climate justice. By identifying where and why AC access is lacking, this work informs targeted interventions to reduce risk in a warming world.

In 2021, the Indian government announced the goal of net-zero carbon emissions in India by 2070. India’s vulnerability to climate impacts and central role in reducing global emissions make it essential to understand the policy features that affect Indians’ support for the 2070 goal. We conducted a conjoint experiment (N = 1,500) to test Indians’ support for the 2070 goal depending on differences among possible policy features including economic programs, allocation of funds collected via coal taxes, policies to mitigate pollution, who participates in the policy-making process, and adaptation policy. We find that Indians express stronger support for the 2070 goal when it includes jobs programs for Indians who lose their job in a coal-fired power plant, policies to mitigate pollution, investment or redistribution of funds collected via coal taxes, and adaptation policies to protect the most vulnerable Indians and critical infrastructure. Support for certain policy features varies depending on income, education, and caste.

Research shows that opinions rooted in moral justifications may be more stable over time. This suggests that communicators could create more durable persuasion by adding explicit moral claims (i.e., stating that the issue is a matter of right vs. wrong) to persuasive messages. Here, using animated videos about the transition from fossil fuels to renewable energy, we conducted a three-stage longitudinal experiment (N = 4,488) to test whether messages with explicit moral claims have more durable persuasive effects than similar messages without. We replicated this test across two distinct moral dimensions: harm to innocent people and contamination of nature’s purity. Our findings show that all message versions had strong immediate persuasive effects relative to a control condition, and much of those effects remained after a three-week period. However, there were no clear advantages—in immediate effects or in durability—from adding explicit moral claims to the messages.

Social media has become an important medium for climate change communication, where sponsored content can be delivered to specific audiences. However, studying message effects on social media platforms has limitations and therefore scholars often rely on surveys and controlled experiments, which can lack external validity. Here, we use survey methods to predict the real-world performance of advertisements on Facebook. We found that a 10 percentage point increase in self-reported likelihood of sharing a post predicted 55% more actual shares, and a 10 percentage point increase in perceptions that the post would be interesting to others predicted 86% more shares. We also found a U-shaped relationship between people’s emotional reactions to the posts and the number of shares, such that emotionally-neutral posts were shared less often than posts that elicited either a strongly positive or negative emotional response. We then discuss the strategic and practical applications of these findings.