Polarisation of Climate and Environmental Attitudes in the United States, 1973–2022

Addressing global environmental problems requires collective action at multiple levels of governance. Citizen preferences can play an influential role for policymakers, and public acceptability strongly influences governmental actors and regulatory implementation frameworks within democratic states. As policymakers aim to align actions with constituent preferences, citizen demands can mitigate barriers to addressing environmental problems, particularly when pressure for policy solutions comes from a diversified set of party adherents. Over recent decades, environmental attitudes and policy preferences have become increasingly polarised in the United States. Republicans are currently less likely to be concerned about the environment or to support environmental policies. Open questions remain: Have increases in polarisation been symmetrical (both partisan groups moving away from the median) or asymmetrical (one group moving more)? How have these patterns changed over time? Do polarization patterns generalize across different environmental and climate change attitudes? This paper harmonizes four historical pooled cross-sectional datasets (1973–2022) with 83 survey years and n=110,237 respondents to explore seven dimensions of environmental and climate change attitudes. Using cross-classified random effect modelling and predicted probabilities, the study addresses: (i) whether polarization is symmetrical or asymmetrical; (ii) how patterns have changed in recent years; and (iii) whether polarization differs across types of attitudes and behaviors.

Theoretical expectations emphasize increasing divergence between Democratic and Republican elites over recent decades, with party differences likely at a fifty-year high. Public polarization has grown, with fewer weak partisans and more strong identifiers, and widening gaps across issues like employment, living standards, and health insurance. There is debate over whether mass polarization is symmetrical (both sides moving toward ideological poles) or asymmetrical (one side moving further), with some evidence that Republicans have shifted further from the median across ideology and policy attitudes. For environmental attitudes, political preferences are among the strongest and most consistent predictors in the U.S., and polarization appears to have increased. Group-level mechanisms include elite cues from legislators, lobbyists, NGOs, scientists, and media, with Republican elites increasingly challenging climate science and opposing regulatory action—a shift often tied to fossil fuel industry lobbying and conservative think tanks. A conservative countermovement, amplified by media and some religious elites, has promoted climate skepticism. Party sorting has made constituencies more ideologically homogeneous, potentially reinforcing polarization. At the individual level, social identity processes shape party affiliation and adherence to in-group norms; salient issues like climate can become identity markers, leading Democrats to adopt pro-environmental stances and Republicans to mobilize against positions identified with the out-group. Prior research documents rising partisan gaps on environmental spending (GSS), climate attitudes (Gallup), and Pew measures; however, no prior work compares longitudinal shifts within partisan groups across multiple datasets and a wide range of measures. This study addresses whether Republicans alone drive attitudinal differences via asymmetric polarization, or whether both parties have shifted. Empirical questions addressed: (i) Symmetry vs. asymmetry of polarization within Republicans and Democrats; (ii) temporal changes in polarization patterns; and (iii) differences across environmental attitudes vs. climate beliefs and attitudes (e.g., skepticism vs. concern).

Data sources and scope: The study pools cross-sectional surveys from four sources covering 1973–2021/2022 and n=110,237 respondents across seven measures. - General Social Survey (GSS) cumulative 1973–2021: measures of support for federal environmental spending and confidence in the scientific community. - Pew Research Center: (a) Climate change as a top priority (16 time points, 2001–2020); (b) Climate change as a major threat (8 time points, 2009–2019). Collected in separate, independent survey series. - Gallup Poll Social Series (GPSS) 1989–2021: climate change worry, whether climate change effects have already begun, and whether the seriousness of climate change is generally exaggerated (27 time points). Seven dependent variables (all coded as binary for analysis): 1) Environmental spending (GSS): “Too little” vs. other responses to federal spending on improving/protecting the environment. 2) Confidence in scientific community (GSS): “A great deal” vs. “only some/hardly any.” 3) Climate change worry (GPSS): “A fair amount/a great deal” vs. “only a little/not at all.” 4) Climate change already happening (GPSS): “Already begun” vs. not yet begun. 5) Climate change exaggerated (GPSS): “Generally exaggerated” vs. not exaggerated. 6) Climate change is a top priority (Pew): “Top priority” vs. not. 7) Climate change is a major threat (Pew): “Major threat” vs. not. Key predictor and controls: Party identification harmonized to three categories (Republican, Independent, Democrat) across datasets. Controls include age and age-squared, gender (female indicator), education (categorical degree attainment), and race/ethnicity (white vs. non-white). Analytical approach: Cross-classified random effects models (CCREM) within a hierarchical age-period-cohort framework. Level 1 models individual responses as a function of party ID and controls. Level 2 includes random intercepts for survey year period and 5-year birth cohorts, and a random coefficient for party affiliation by survey year to capture time-varying partisan effects. This addresses the complex nesting of individuals within both period and cohort and facilitates interpretation of historical trends independent of demographic composition changes. Postestimation: Compute average predicted probabilities by party (Republican, Independent, Democrat) for each year to facilitate substantive interpretation. Non-overlapping 95% confidence intervals indicate statistically significant differences. Predicted probabilities are also estimated by birth cohort as a robustness check. Robustness checks: Compare CCREM results with three alternative estimation strategies using the climate priority outcome as an example: (1) fixed-effect logistic models with period-by-party interactions; (2) random intercepts logistic models for period; (3) random effects logistic models with party identification nested in period. Patterns are substantively similar across approaches; CCREM tends to yield comparatively smaller Democratic effects, possibly due to explicit incorporation of period and cohort effects. Data harmonization and availability: Data harmonized across sources; descriptive statistics and raw means provided in supplementary materials. Analyses performed in Stata 16.1. Replication code available via Harvard Dataverse (https://doi.org/10.7910/DVN/DWXVIE).

Overall pattern: Contemporary environmental and climate change attitudes are symmetrically polarized—Democrats exhibit higher pro-environmental attitudes and Republicans lower, roughly equidistant from the median. Historically, polarization dynamics vary by measure, revealing two asymmetric patterns: (1) Republicans became less pro-environmental beginning in the early 1990s; (2) Democrats have become more pro-environmental since the mid-2010s. Environmental attitudes (GSS, 1973–2021): - Environmental spending (“too little”): Asymmetric polarization. Republican support decreased substantially beginning in the early 1990s, stabilized mid-2000s to late-2010s, then dropped in 2021 (predicted probability fell from ~0.45 to 0.34; single data point cautioned). Democrats increased recently: predicted probability rose from ~0.70 (2015) to 0.85 (2021). - Confidence in the scientific community (“a great deal”): Minimal partisan differences historically, but a large 2021 divergence: Democrats about +0.20 higher, Republicans ~−0.05 lower than prior years (needs confirmation with future data). Climate change beliefs (GPSS, 1989–2021): Symmetric polarization across beliefs. - Worry “a great deal”: Minimal partisan differences in 1989. Republicans declined ~30 percentage points to ~0.30 by mid-2000s and remained stable. Democrats increased from 0.70 (2011) to 0.91 (2021). - Effects already begun: Near parity in 1997 (both ~0.50), diverging by 2021: Democrats 0.83 vs. Republicans 0.30. - Seriousness exaggerated: Already polarized by 1997. Republicans increased from 0.46 (1997) to 0.79 (2021). Democrats approached 0 by 2021. Independents remained ~0.30–0.40. Climate change attitudes (Pew, 2001–2020/2019): Asymmetric polarization driven by rising Democratic concern post mid-2010s. - “Top priority”: Republicans stable around ~0.15 over two decades; Democrats increased from ~0.45 (mid-2000s) to 0.69 (2020). - “Major threat”: Large partisan gap already by 2007 (~0.45 difference). Republicans largely stable; Democrats increased from ~0.70 to 0.83 (2007–2020). Period and cohort effects: Significant period variance components indicate general greening during certain periods (spending 0.04; priority 0.09; threat 0.06), alongside increased skepticism for exaggeration (0.02). Birth cohort variance is statistically significant but substantively small; no strong evidence of a “rising green generation.” Robustness: Predicted probability patterns by party are consistent across CCREM and alternative models; CCREM provides conservative estimates for Democratic effects. Overall, party affiliation dynamics over time dominate period or cohort explanations.

Findings indicate that while the current U.S. public exhibits symmetric polarization on environmental and climate attitudes—Democrats higher and Republicans lower—historical trajectories differ by construct. Environmental attitudes and climate change attitudes show two waves of asymmetric polarization: an initial rightward shift among Republicans in the early 1990s, followed by a more recent increase in pro-environmental attitudes among Democrats starting mid-2010s. In contrast, climate change beliefs (worry, belief that effects have begun, exaggeration) display symmetric divergence since the mid-1990s. These patterns align with literature documenting a conservative countermovement against environmental regulation and climate action, amplified by think tanks, lobbying (with a >10:1 spending imbalance favoring fossil fuel–aligned sectors over renewable/environmental groups), and sympathetic media. The more recent Democratic “greening” may reflect greater acceptance of scientific consensus, responsiveness to extreme weather experiences, evolving coalitions (e.g., increased share of highly educated women), norm shifts within the Democratic constituency, and elite cueing (e.g., increased congressional attention to climate impacts). Exploratory analyses suggest rising pro-environmental attitudes across Democrats irrespective of age. Policy implications: Asymmetric polarization shapes coalition-building for climate action. While unified Democratic control enabled passage of the Inflation Reduction Act (2022), U.S. governance is often divided, implying that durable federal climate policy may require at least some Republican support despite climate policy’s large partisan gap. The recent Democratic shift could offset decades of conservative movement away from environmental consensus, culminating in symmetric polarization where both parties are now far from the median but on opposite sides. Yet, climate remains a particularly “sticky” policy domain relative to others, posing challenges for bipartisan progress.

This study harmonizes multiple historical U.S. survey series spanning five decades and seven measures to map the evolution of partisan polarization on environmental and climate attitudes. It shows symmetric polarization in the contemporary period, but historically distinct dynamics: an earlier Republican shift away from pro-environmental positions in the 1990s, and a more recent Democratic increase in pro-environmental attitudes since the mid-2010s. These shifts have elevated average U.S. pro-environmental sentiment by roughly 10–20%. Contributions include a cross-dataset, within-party, longitudinal comparison across multiple attitude dimensions using CCREM to account for period and cohort effects, and intuitive postestimation via predicted probabilities. The results highlight the centrality of party affiliation over demographic cohort dynamics in explaining trends. Future research directions: (1) Identify causal drivers of the recent increase in Democratic pro-environmental attitudes (e.g., elite cues, norm shifts, coalition composition, extreme weather exposure, science communication). (2) Examine less polarized environmental policy domains (e.g., energy efficiency, renewable energy) to assess whether similar polarization dynamics emerge. (3) Collect additional waves to validate 2021–2022 shifts (especially environmental spending support and confidence in science) and further disentangle age, period, and cohort mechanisms within parties.

- Measurement scope: The dependent variables are broad indicators and may be sensitive to partisan response bias. - Data constraints: Some recent findings rely on single recent survey waves (e.g., 2021 drops among Republicans for spending; 2021 divergence in confidence in science) and require confirmation with additional data. - Generalizability across policy domains: Other environmental areas (e.g., energy efficiency, renewable energy) are often less polarized; patterns here may not fully generalize. - Age–period–cohort identification: While CCREM incorporates period and cohort random effects and robustness checks, inherent identification challenges in pooled cross-sectional APC analyses remain. - Secondary data use: Reliance on existing surveys limits control over measurement and timing across sources.