Political affiliation shapes climate attention and choice beyond extreme weather exposure

When a hurricane floods a county or a wildfire chokes the air of a town, the intuitive assumption is that the experience itself will push people toward climate action. A new study published in Nature Climate Change dismantles that assumption with unusual precision. Using a large, representative United States sample, an incentivized decision task with real carbon emissions, and objective county-level weather records, researchers led by Annika M. Wyss and Emmanuel T. Guizar Rosales of the University of Bern found that political affiliation, not physical exposure to extreme weather, was the dominant force shaping how people attend to climate-relevant information and what they actually choose. The study, a preregistered Registered Report, suggests that the same storm can be encoded as climate evidence by one person and dismissed as unrelated by another, and that this interpretive filter operates before any choice is made.

The research team set out to answer a question that has divided the field for years: do extreme weather events act as catalysts for climate engagement? Prior evidence has been mixed. Some studies found that exposure to floods, heat waves and wildfires increased climate concern, media attention and even support for climate-related ballot measures. Others reported little to no effect. A recurring explanation is that the impact of such events depends on individual differences, above all political affiliation. In the polarized American context, liberals and conservatives differ systematically in their climate beliefs, and previous work has shown that wildfire exposure increased pro-environmental voting only in Democratic areas, while severe flooding predicted mitigation responses only among people who attributed the event to climate change. What remained unclear was the underlying psychological mechanism: where in the decision process does partisanship exert its influence?

To capture that mechanism directly, the team combined two methodological pillars. The first was a new variant of the Carbon Emission Task, a validated behavioral paradigm in which participants choose between two options that differ in both a financial bonus and the carbon emissions they generate. Crucially, the emissions are real: one randomly selected choice is implemented by retiring emission certificates in a cap-and-trade system in collaboration with a non-governmental organization, so participants’ decisions carry genuine environmental consequences. The second pillar was MouselabWEB, a process-tracing technology in which all information is initially hidden behind boxes. Participants reveal the bonus and emission details only by hovering the mouse over each box, and the software records every visit and its duration. This dwell time, known as ΔDuration when normalized across attributes, serves as an established measure of attention allocation, allowing researchers to observe the psychological trade-off between personal benefit and climate consequence as it unfolds.

Attention was then linked to the physical environment. The researchers matched each participant’s county of residence, identified through Federal Information Processing Standard codes, to records from the National Oceanic and Atmospheric Administration’s Storm Events Database. They focused on eight event types projected to intensify with climate change: excessive heat, drought, wildfire, flash floods, coastal floods, hail, strong wind and tornadoes. Exposure was operationalized as whether at least one such episode occurred in the participant’s county during the 30 days before study completion, with sensitivity analyses extending the window to 360 days and weighting events by severity and abnormality. Data were collected between August and September 2025 from a quota-matched sample of 2,000 respondents; after preregistered quality exclusions, the final analytic sample comprised 1,451 participants, roughly evenly split between Democrats and Republicans.

The headline result confirmed the first hypothesis with striking clarity. Political affiliation significantly predicted attention allocation, such that Democrats spent relatively more time inspecting carbon emission information than Republicans, with a standardized effect of b = 0.08 (95% confidence interval 0.04 to 0.12, P < 0.001). The effect persisted after adjusting for task-related and demographic controls and across the full sample. Notably, the two groups did not differ in their beliefs about the effectiveness of carbon certificates as an emissions reduction mechanism, meaning the partisan gap in attention was not a matter of distrusting the task itself. Political affiliation also predicted actual choices: Democrats were roughly twice as likely as Republicans to select the pro-climate option, with an odds ratio of 2.06 (P < 0.001). Partisan differences, in other words, emerge not only at the point of choice but already during information acquisition, consistent with theories of motivated reasoning and identity-protective cognition in which individuals preferentially attend to information that aligns with their group identity.

The second hypothesis, however, failed. Objectively measured exposure to extreme weather did not moderate the relationship between political affiliation and either attention allocation or pro-climate choices. The interaction between affiliation and exposure was statistically non-significant across every model specification, including duration-weighted exposure measures, alternative time windows from one to twelve months, and county-level controls such as population density, median income and the Partisan Voting Index. Equivalence tests confirmed that the interaction was practically indistinguishable from zero. Republicans were in fact slightly more likely than Democrats to have experienced at least one recorded event during the study period, 48.2 percent versus 43.4 percent, yet this objective exposure left the partisan gap in attention and choice untouched. Physical experience of a storm, flood or heat wave, by itself, appears insufficient to reshape how people process climate-relevant information.

The third set of findings revealed where the interpretive filter operates. Subjective attribution, the degree to which participants viewed extreme weather events as caused by, indicative of, or representative of climate change, was strongly associated with both outcomes. Participants who attributed events more strongly to climate change attended more to carbon information (b = 0.05, P < 0.001) and made more pro-climate choices (odds ratio 1.41, P < 0.001), effects that held over and above political affiliation. Democrats reported significantly higher attribution than Republicans, 4.52 versus 3.94 on a five-point scale. More intriguingly, attribution moderated the effects of exposure itself. The interaction between objective exposure and attribution was negative and significant for attention (b = −0.05, P = 0.017) and for choices (odds ratio 0.66, P = 0.010). Johnson–Neyman analyses showed that exposure was positively associated with carbon attention at low attribution levels, non-significant at moderate levels, and significantly negative at high attribution levels. The authors propose two interpretations: among people who already strongly attribute events to climate change, attention may saturate, or threatening information may be avoided through defensive processing, whereas for those with weak attribution, exposure may act as a salient cue that briefly increases engagement with climate-relevant attributes.

Partisan differences also sharpened as attribution increased. The interaction between political affiliation and subjective attribution was positive and significant for attention (b = 0.07, P = 0.006) and for choices (odds ratio 1.64, P = 0.017), indicating that the gap between Democrats and Republicans widened among individuals who more strongly interpreted extreme weather in climate terms. The preregistered three-way interaction between affiliation, exposure and attribution, however, was not significant for either outcome, and this null result was robust across controls, exposure windows and event types. Exploratory analyses of self-reported subjective experiences added a further twist: Democrats reported more frequent extreme weather experiences than Republicans despite lower objective exposure, and subjective experience predicted both greater carbon attention (b = 0.02, P = 0.020) and pro-climate choices (odds ratio 1.46, P < 0.001). This pattern supports the idea that political affiliation functions as a perceptual filter, shaping not only how environmental events are interpreted but also which events are noticed, recalled and encoded as climate-relevant in the first place.

The study’s design carries considerable methodological weight. As a Registered Report, the hypotheses, analysis plan and smallest effect sizes of interest were specified and accepted before data collection, limiting analytical flexibility. By combining objective exposure records with subjective attribution measures, the team avoided the common method bias that has plagued previous studies relying on self-reported exposure. Mixed-effects models with random intercepts for participants and trials accounted for repeated observations, and multicollinearity was negligible. One limitation is that preregistered quality criteria removed roughly one quarter of the initial sample, potentially constraining generalizability, though sensitivity analyses on the full sample yielded substantively identical conclusions. The authors also acknowledge that effects of extreme weather are heterogeneous, depending on event type and how exposure is operationalized, with some indication of three-way interactions only for specific cases such as strong wind and coastal flood events lasting at least 120 days.

The applied implications are sobering. Simply increasing exposure to extreme weather, which climate change guarantees, is unlikely by itself to foster greater climate attention or pro-climate choices. The decisive variable is how those experiences are contextualized and interpreted in light of existing belief systems. From a multi-level perspective, attention allocation reflects the interplay between top-down influences such as political identity and bottom-up inputs such as environmental events, and experiences shape decision-making only insofar as they are cognitively encoded within existing frameworks. The findings therefore point toward communication strategies that help people connect extreme weather events to climate change, rather than relying on the events to speak for themselves. In a warming world where storms will keep arriving, the study suggests that the storm’s meaning, not the storm itself, is what moves minds.

Subject of Research: Psychology

Article Title: Political affiliation impacts climate attention and choice beyond extreme weather exposure

Article References: Political affiliation impacts climate attention and choice beyond extreme weather exposure. (n.d.). https://doi.org/10.1038/s41558-026-02765-z

Image Credits: AI Generated

DOI: 10.1038/s41558-026-02765-z

Keywords: climate change, political affiliation, extreme weather, decision-making, attention allocation, motivated reasoning, Nature Climate Change, behavioral science, Political, affiliation, impacts, climate